Upload folder using huggingface_hub

3b-mb_qwen/README.md

@@ -0,0 +1,164 @@
+---
+library_name: transformers
+license: other
+base_model: Qwen/Qwen2.5-3B-Instruct
+tags:
+- generated_from_trainer
+datasets:
+- mb_qwen.jsonl
+model-index:
+- name: outputs/out
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
+<details><summary>See axolotl config</summary>
+
+axolotl version: `0.6.0`
+```yaml
+base_model: Qwen/Qwen2.5-3B-Instruct
+model_type: AutoModelForCausalLM
+tokenizer_type: AutoTokenizer
+trust_remote_code: false
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
+output_dir: ./outputs/out2
+chat_template: qwen_25
+datasets:
+  - path: mb_qwen.jsonl
+    type: chat_template
+    field_messages: messages
+    message_field_role: role
+    message_field_content: content
+    roles:
+      system:
+        - system
+      user:
+        - user
+      assistant:
+        - assistant
+
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.005
+output_dir: ./outputs/out
+eval_sample_packing: False
+
+sequence_len: 8192
+sample_packing: False
+pad_to_sequence_len: False
+
+wandb_project: mergedbench
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+plugins:
+  - axolotl.integrations.liger.LigerPlugin
+liger_rope: true
+liger_rms_norm: true
+liger_swiglu: true
+liger_fused_linear_cross_entropy: true
+
+gradient_accumulation_steps: 4
+micro_batch_size: 8
+eval_batch_size: 4
+num_epochs: 3
+optimizer: paged_adamw_8bit
+lr_scheduler: cosine
+learning_rate: 2e-5
+
+train_on_inputs: false
+group_by_length: false
+bf16: auto
+fp16:
+tf32: false
+
+gradient_checkpointing: true
+gradient_checkpointing_kwargs:
+  use_reentrant: false
+early_stopping_patience:
+resume_from_checkpoint:
+logging_steps: 1
+xformers_attention:
+flash_attention: true
+
+warmup_steps: 30
+evals_per_epoch: 3
+eval_max_new_tokens: 128
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed: deepspeed_configs/zero1.json
+weight_decay: 0.01
+fsdp:
+fsdp_config:
+special_tokens:
+```
+
+</details><br>
+
+# outputs/out
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct) on the mb_qwen.jsonl dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.2918
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 2e-05
+- train_batch_size: 8
+- eval_batch_size: 4
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 2
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 64
+- total_eval_batch_size: 8
+- optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_steps: 30
+- num_epochs: 3.0
+
+### Training results
+
+| Training Loss | Epoch  | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 1.1276        | 0.0041 | 1    | 1.1255          |
+| 0.3639        | 0.3350 | 82   | 0.3423          |
+| 0.303         | 0.6701 | 164  | 0.3124          |
+| 0.2298        | 1.0082 | 246  | 0.3009          |
+| 0.2219        | 1.3432 | 328  | 0.3102          |
+| 0.196         | 1.6782 | 410  | 0.3017          |
+| 0.1716        | 2.0163 | 492  | 0.2929          |
+| 0.1586        | 2.3514 | 574  | 0.2984          |
+| 0.1578        | 2.6864 | 656  | 0.2918          |
+
+
+### Framework versions
+
+- Transformers 4.48.1
+- Pytorch 2.5.1+cu121
+- Datasets 3.2.0
+- Tokenizers 0.21.0

3b-mb_qwen/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

3b-mb_qwen/checkpoint-338/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

3b-mb_qwen/checkpoint-338/config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "Qwen/Qwen2.5-3B-Instruct",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 70,
+  "model_type": "qwen2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.48.1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151665
+}

3b-mb_qwen/checkpoint-338/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.48.1"
+}

3b-mb_qwen/checkpoint-338/latest

@@ -0,0 +1 @@
+global_step337

3b-mb_qwen/checkpoint-338/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec1dcaf6f5430f06caba435ab343745ffde124cf70490a36f66d78187bf075e4
+size 4956450288

3b-mb_qwen/checkpoint-338/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3981a3a9d6b4e220344b35a25a971d7825c700fbd96ff52859b234be31da7df
+size 1835586736

3b-mb_qwen/checkpoint-338/model.safetensors.index.json

@@ -0,0 +1,442 @@
+{
+  "metadata": {
+    "total_size": 6791987200
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00002-of-00002.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors"
+  }
+}

3b-mb_qwen/checkpoint-338/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dcb161b22b2558dbf7e3f8c871050cec383d11a40423fab11f18d5e630639bf
+size 14512

3b-mb_qwen/checkpoint-338/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d50af6aef769414a6f28fa1b1bc51ce707dc8ecd15474e03f99a2f10fde086be
+size 14512

3b-mb_qwen/checkpoint-338/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd0ff87d03adec7a7483b66c31fc3a08e9184f59f52667e0a62a335c052ee5c8
+size 1064

3b-mb_qwen/checkpoint-338/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

3b-mb_qwen/checkpoint-338/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
+size 11421896

3b-mb_qwen/checkpoint-338/tokenizer_config.json

@@ -0,0 +1,208 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

3b-mb_qwen/checkpoint-338/trainer_state.json

@@ -0,0 +1,2447 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9926144756277697,
+  "eval_steps": 57,
+  "global_step": 338,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005908419497784343,
+      "grad_norm": 4.501461029052734,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 1.062,
+      "step": 1
+    },
+    {
+      "epoch": 0.005908419497784343,
+      "eval_loss": 1.0835397243499756,
+      "eval_runtime": 4.3539,
+      "eval_samples_per_second": 12.632,
+      "eval_steps_per_second": 1.608,
+      "step": 1
+    },
+    {
+      "epoch": 0.011816838995568686,
+      "grad_norm": 4.469114303588867,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.0268,
+      "step": 2
+    },
+    {
+      "epoch": 0.01772525849335303,
+      "grad_norm": 4.554893970489502,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0401,
+      "step": 3
+    },
+    {
+      "epoch": 0.023633677991137372,
+      "grad_norm": 4.374792575836182,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.0423,
+      "step": 4
+    },
+    {
+      "epoch": 0.029542097488921712,
+      "grad_norm": 3.4377498626708984,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9965,
+      "step": 5
+    },
+    {
+      "epoch": 0.03545051698670606,
+      "grad_norm": 3.1242499351501465,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.9479,
+      "step": 6
+    },
+    {
+      "epoch": 0.0413589364844904,
+      "grad_norm": 1.8368685245513916,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.8296,
+      "step": 7
+    },
+    {
+      "epoch": 0.047267355982274745,
+      "grad_norm": 1.7457680702209473,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.8159,
+      "step": 8
+    },
+    {
+      "epoch": 0.053175775480059084,
+      "grad_norm": 1.2953853607177734,
+      "learning_rate": 6e-06,
+      "loss": 0.664,
+      "step": 9
+    },
+    {
+      "epoch": 0.059084194977843424,
+      "grad_norm": 1.1054794788360596,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.6486,
+      "step": 10
+    },
+    {
+      "epoch": 0.06499261447562776,
+      "grad_norm": 0.8712942004203796,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.6415,
+      "step": 11
+    },
+    {
+      "epoch": 0.07090103397341212,
+      "grad_norm": 1.4441039562225342,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6255,
+      "step": 12
+    },
+    {
+      "epoch": 0.07680945347119646,
+      "grad_norm": 1.4984484910964966,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.5561,
+      "step": 13
+    },
+    {
+      "epoch": 0.0827178729689808,
+      "grad_norm": 0.8376960754394531,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.5534,
+      "step": 14
+    },
+    {
+      "epoch": 0.08862629246676514,
+      "grad_norm": 0.7184750437736511,
+      "learning_rate": 1e-05,
+      "loss": 0.5062,
+      "step": 15
+    },
+    {
+      "epoch": 0.09453471196454949,
+      "grad_norm": 0.8381787538528442,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.5531,
+      "step": 16
+    },
+    {
+      "epoch": 0.10044313146233383,
+      "grad_norm": 0.7621350288391113,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 0.4876,
+      "step": 17
+    },
+    {
+      "epoch": 0.10635155096011817,
+      "grad_norm": 0.6955872178077698,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5019,
+      "step": 18
+    },
+    {
+      "epoch": 0.11225997045790251,
+      "grad_norm": 0.5844917297363281,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 0.4368,
+      "step": 19
+    },
+    {
+      "epoch": 0.11816838995568685,
+      "grad_norm": 0.5807573795318604,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4965,
+      "step": 20
+    },
+    {
+      "epoch": 0.1240768094534712,
+      "grad_norm": 0.5376399755477905,
+      "learning_rate": 1.4e-05,
+      "loss": 0.4841,
+      "step": 21
+    },
+    {
+      "epoch": 0.12998522895125553,
+      "grad_norm": 0.5053263902664185,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.4573,
+      "step": 22
+    },
+    {
+      "epoch": 0.1358936484490399,
+      "grad_norm": 0.5155225396156311,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.451,
+      "step": 23
+    },
+    {
+      "epoch": 0.14180206794682423,
+      "grad_norm": 0.52030348777771,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4199,
+      "step": 24
+    },
+    {
+      "epoch": 0.14771048744460857,
+      "grad_norm": 0.5321907997131348,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4532,
+      "step": 25
+    },
+    {
+      "epoch": 0.1536189069423929,
+      "grad_norm": 0.5318155288696289,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.4813,
+      "step": 26
+    },
+    {
+      "epoch": 0.15952732644017725,
+      "grad_norm": 0.5176340937614441,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4288,
+      "step": 27
+    },
+    {
+      "epoch": 0.1654357459379616,
+      "grad_norm": 0.43893975019454956,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.3766,
+      "step": 28
+    },
+    {
+      "epoch": 0.17134416543574593,
+      "grad_norm": 0.43830162286758423,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 0.4159,
+      "step": 29
+    },
+    {
+      "epoch": 0.17725258493353027,
+      "grad_norm": 0.45950719714164734,
+      "learning_rate": 2e-05,
+      "loss": 0.4505,
+      "step": 30
+    },
+    {
+      "epoch": 0.1831610044313146,
+      "grad_norm": 0.40500667691230774,
+      "learning_rate": 1.9999783114048658e-05,
+      "loss": 0.3726,
+      "step": 31
+    },
+    {
+      "epoch": 0.18906942392909898,
+      "grad_norm": 0.43435147404670715,
+      "learning_rate": 1.9999132465602526e-05,
+      "loss": 0.442,
+      "step": 32
+    },
+    {
+      "epoch": 0.19497784342688332,
+      "grad_norm": 0.44813328981399536,
+      "learning_rate": 1.999804808288491e-05,
+      "loss": 0.437,
+      "step": 33
+    },
+    {
+      "epoch": 0.20088626292466766,
+      "grad_norm": 0.48166996240615845,
+      "learning_rate": 1.9996530012933285e-05,
+      "loss": 0.4107,
+      "step": 34
+    },
+    {
+      "epoch": 0.206794682422452,
+      "grad_norm": 0.398764044046402,
+      "learning_rate": 1.9994578321597258e-05,
+      "loss": 0.3882,
+      "step": 35
+    },
+    {
+      "epoch": 0.21270310192023634,
+      "grad_norm": 0.44229164719581604,
+      "learning_rate": 1.999219309353572e-05,
+      "loss": 0.4154,
+      "step": 36
+    },
+    {
+      "epoch": 0.21861152141802068,
+      "grad_norm": 0.44369620084762573,
+      "learning_rate": 1.998937443221316e-05,
+      "loss": 0.3863,
+      "step": 37
+    },
+    {
+      "epoch": 0.22451994091580502,
+      "grad_norm": 0.44270017743110657,
+      "learning_rate": 1.9986122459895182e-05,
+      "loss": 0.3945,
+      "step": 38
+    },
+    {
+      "epoch": 0.23042836041358936,
+      "grad_norm": 0.42152372002601624,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 0.4094,
+      "step": 39
+    },
+    {
+      "epoch": 0.2363367799113737,
+      "grad_norm": 0.4120837450027466,
+      "learning_rate": 1.9978319165308373e-05,
+      "loss": 0.4411,
+      "step": 40
+    },
+    {
+      "epoch": 0.24224519940915806,
+      "grad_norm": 0.4064903259277344,
+      "learning_rate": 1.997376818152453e-05,
+      "loss": 0.3818,
+      "step": 41
+    },
+    {
+      "epoch": 0.2481536189069424,
+      "grad_norm": 0.3692624270915985,
+      "learning_rate": 1.9968784563700586e-05,
+      "loss": 0.3874,
+      "step": 42
+    },
+    {
+      "epoch": 0.25406203840472674,
+      "grad_norm": 0.4399218261241913,
+      "learning_rate": 1.9963368528011867e-05,
+      "loss": 0.3749,
+      "step": 43
+    },
+    {
+      "epoch": 0.25997045790251105,
+      "grad_norm": 0.3779003620147705,
+      "learning_rate": 1.9957520309390786e-05,
+      "loss": 0.3656,
+      "step": 44
+    },
+    {
+      "epoch": 0.2658788774002954,
+      "grad_norm": 0.3946981132030487,
+      "learning_rate": 1.9951240161516643e-05,
+      "loss": 0.3612,
+      "step": 45
+    },
+    {
+      "epoch": 0.2717872968980798,
+      "grad_norm": 0.3969726264476776,
+      "learning_rate": 1.99445283568046e-05,
+      "loss": 0.3932,
+      "step": 46
+    },
+    {
+      "epoch": 0.2776957163958641,
+      "grad_norm": 0.4239075183868408,
+      "learning_rate": 1.9937385186393888e-05,
+      "loss": 0.387,
+      "step": 47
+    },
+    {
+      "epoch": 0.28360413589364847,
+      "grad_norm": 0.3688453733921051,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 0.3524,
+      "step": 48
+    },
+    {
+      "epoch": 0.2895125553914328,
+      "grad_norm": 0.4294806718826294,
+      "learning_rate": 1.9921806006577102e-05,
+      "loss": 0.3787,
+      "step": 49
+    },
+    {
+      "epoch": 0.29542097488921715,
+      "grad_norm": 0.3867166042327881,
+      "learning_rate": 1.9913370672952074e-05,
+      "loss": 0.3756,
+      "step": 50
+    },
+    {
+      "epoch": 0.30132939438700146,
+      "grad_norm": 0.43365901708602905,
+      "learning_rate": 1.990450532516116e-05,
+      "loss": 0.3896,
+      "step": 51
+    },
+    {
+      "epoch": 0.3072378138847858,
+      "grad_norm": 0.38658151030540466,
+      "learning_rate": 1.9895210347758233e-05,
+      "loss": 0.3703,
+      "step": 52
+    },
+    {
+      "epoch": 0.31314623338257014,
+      "grad_norm": 0.37093815207481384,
+      "learning_rate": 1.98854861439333e-05,
+      "loss": 0.3763,
+      "step": 53
+    },
+    {
+      "epoch": 0.3190546528803545,
+      "grad_norm": 0.40044137835502625,
+      "learning_rate": 1.9875333135495e-05,
+      "loss": 0.3752,
+      "step": 54
+    },
+    {
+      "epoch": 0.3249630723781389,
+      "grad_norm": 0.39133360981941223,
+      "learning_rate": 1.986475176285232e-05,
+      "loss": 0.3589,
+      "step": 55
+    },
+    {
+      "epoch": 0.3308714918759232,
+      "grad_norm": 0.38397374749183655,
+      "learning_rate": 1.985374248499546e-05,
+      "loss": 0.3701,
+      "step": 56
+    },
+    {
+      "epoch": 0.33677991137370755,
+      "grad_norm": 0.3795414865016937,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 0.3584,
+      "step": 57
+    },
+    {
+      "epoch": 0.33677991137370755,
+      "eval_loss": 0.3953791558742523,
+      "eval_runtime": 4.6385,
+      "eval_samples_per_second": 11.857,
+      "eval_steps_per_second": 1.509,
+      "step": 57
+    },
+    {
+      "epoch": 0.34268833087149186,
+      "grad_norm": 0.3709493577480316,
+      "learning_rate": 1.9830442142386e-05,
+      "loss": 0.3647,
+      "step": 58
+    },
+    {
+      "epoch": 0.34859675036927623,
+      "grad_norm": 0.35005033016204834,
+      "learning_rate": 1.9818152088336786e-05,
+      "loss": 0.3317,
+      "step": 59
+    },
+    {
+      "epoch": 0.35450516986706054,
+      "grad_norm": 0.3652004599571228,
+      "learning_rate": 1.9805436150436352e-05,
+      "loss": 0.3394,
+      "step": 60
+    },
+    {
+      "epoch": 0.3604135893648449,
+      "grad_norm": 0.3940984904766083,
+      "learning_rate": 1.9792294880266346e-05,
+      "loss": 0.3711,
+      "step": 61
+    },
+    {
+      "epoch": 0.3663220088626292,
+      "grad_norm": 0.35634928941726685,
+      "learning_rate": 1.977872884785815e-05,
+      "loss": 0.3455,
+      "step": 62
+    },
+    {
+      "epoch": 0.3722304283604136,
+      "grad_norm": 0.3972924053668976,
+      "learning_rate": 1.9764738641668137e-05,
+      "loss": 0.3652,
+      "step": 63
+    },
+    {
+      "epoch": 0.37813884785819796,
+      "grad_norm": 0.40372708439826965,
+      "learning_rate": 1.9750324868552133e-05,
+      "loss": 0.3662,
+      "step": 64
+    },
+    {
+      "epoch": 0.38404726735598227,
+      "grad_norm": 0.396133691072464,
+      "learning_rate": 1.9735488153739128e-05,
+      "loss": 0.3726,
+      "step": 65
+    },
+    {
+      "epoch": 0.38995568685376664,
+      "grad_norm": 0.398989737033844,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 0.3595,
+      "step": 66
+    },
+    {
+      "epoch": 0.39586410635155095,
+      "grad_norm": 0.4102807939052582,
+      "learning_rate": 1.9704548491640195e-05,
+      "loss": 0.3308,
+      "step": 67
+    },
+    {
+      "epoch": 0.4017725258493353,
+      "grad_norm": 0.344397634267807,
+      "learning_rate": 1.9688446886429885e-05,
+      "loss": 0.3653,
+      "step": 68
+    },
+    {
+      "epoch": 0.4076809453471196,
+      "grad_norm": 0.3550814390182495,
+      "learning_rate": 1.9671925023615572e-05,
+      "loss": 0.3412,
+      "step": 69
+    },
+    {
+      "epoch": 0.413589364844904,
+      "grad_norm": 0.4047009348869324,
+      "learning_rate": 1.9654983619869242e-05,
+      "loss": 0.3578,
+      "step": 70
+    },
+    {
+      "epoch": 0.4194977843426883,
+      "grad_norm": 0.41112563014030457,
+      "learning_rate": 1.9637623410061392e-05,
+      "loss": 0.3694,
+      "step": 71
+    },
+    {
+      "epoch": 0.4254062038404727,
+      "grad_norm": 0.3775319755077362,
+      "learning_rate": 1.961984514722914e-05,
+      "loss": 0.3571,
+      "step": 72
+    },
+    {
+      "epoch": 0.43131462333825704,
+      "grad_norm": 0.3610381782054901,
+      "learning_rate": 1.960164960254358e-05,
+      "loss": 0.3713,
+      "step": 73
+    },
+    {
+      "epoch": 0.43722304283604135,
+      "grad_norm": 0.38662371039390564,
+      "learning_rate": 1.9583037565276314e-05,
+      "loss": 0.311,
+      "step": 74
+    },
+    {
+      "epoch": 0.4431314623338257,
+      "grad_norm": 0.3574771285057068,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 0.3353,
+      "step": 75
+    },
+    {
+      "epoch": 0.44903988183161003,
+      "grad_norm": 0.3932562470436096,
+      "learning_rate": 1.9544567260379455e-05,
+      "loss": 0.3536,
+      "step": 76
+    },
+    {
+      "epoch": 0.4549483013293944,
+      "grad_norm": 0.3974682092666626,
+      "learning_rate": 1.9524710661483594e-05,
+      "loss": 0.3556,
+      "step": 77
+    },
+    {
+      "epoch": 0.4608567208271787,
+      "grad_norm": 0.37172290682792664,
+      "learning_rate": 1.9504440907401113e-05,
+      "loss": 0.3568,
+      "step": 78
+    },
+    {
+      "epoch": 0.4667651403249631,
+      "grad_norm": 0.37170422077178955,
+      "learning_rate": 1.948375887737699e-05,
+      "loss": 0.3556,
+      "step": 79
+    },
+    {
+      "epoch": 0.4726735598227474,
+      "grad_norm": 0.3596966862678528,
+      "learning_rate": 1.9462665468539582e-05,
+      "loss": 0.332,
+      "step": 80
+    },
+    {
+      "epoch": 0.47858197932053176,
+      "grad_norm": 0.35934680700302124,
+      "learning_rate": 1.944116159586169e-05,
+      "loss": 0.3276,
+      "step": 81
+    },
+    {
+      "epoch": 0.4844903988183161,
+      "grad_norm": 0.40984946489334106,
+      "learning_rate": 1.94192481921209e-05,
+      "loss": 0.3685,
+      "step": 82
+    },
+    {
+      "epoch": 0.49039881831610044,
+      "grad_norm": 0.3622114658355713,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.3336,
+      "step": 83
+    },
+    {
+      "epoch": 0.4963072378138848,
+      "grad_norm": 0.34888842701911926,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 0.3625,
+      "step": 84
+    },
+    {
+      "epoch": 0.5022156573116692,
+      "grad_norm": 0.37125518918037415,
+      "learning_rate": 1.9351060388513304e-05,
+      "loss": 0.3304,
+      "step": 85
+    },
+    {
+      "epoch": 0.5081240768094535,
+      "grad_norm": 0.4107120931148529,
+      "learning_rate": 1.9327518542959717e-05,
+      "loss": 0.3755,
+      "step": 86
+    },
+    {
+      "epoch": 0.5140324963072378,
+      "grad_norm": 0.3420109748840332,
+      "learning_rate": 1.9303572095859545e-05,
+      "loss": 0.3457,
+      "step": 87
+    },
+    {
+      "epoch": 0.5199409158050221,
+      "grad_norm": 0.35079535841941833,
+      "learning_rate": 1.9279222085942396e-05,
+      "loss": 0.3454,
+      "step": 88
+    },
+    {
+      "epoch": 0.5258493353028065,
+      "grad_norm": 0.3775666058063507,
+      "learning_rate": 1.9254469569443274e-05,
+      "loss": 0.3501,
+      "step": 89
+    },
+    {
+      "epoch": 0.5317577548005908,
+      "grad_norm": 0.3327409625053406,
+      "learning_rate": 1.9229315620056805e-05,
+      "loss": 0.3507,
+      "step": 90
+    },
+    {
+      "epoch": 0.5376661742983752,
+      "grad_norm": 0.37142789363861084,
+      "learning_rate": 1.9203761328890626e-05,
+      "loss": 0.3453,
+      "step": 91
+    },
+    {
+      "epoch": 0.5435745937961596,
+      "grad_norm": 0.36256077885627747,
+      "learning_rate": 1.91778078044181e-05,
+      "loss": 0.3588,
+      "step": 92
+    },
+    {
+      "epoch": 0.5494830132939439,
+      "grad_norm": 0.3861102759838104,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 0.3479,
+      "step": 93
+    },
+    {
+      "epoch": 0.5553914327917282,
+      "grad_norm": 0.3359353542327881,
+      "learning_rate": 1.9124707575986642e-05,
+      "loss": 0.318,
+      "step": 94
+    },
+    {
+      "epoch": 0.5612998522895125,
+      "grad_norm": 0.33662593364715576,
+      "learning_rate": 1.909756317536643e-05,
+      "loss": 0.3421,
+      "step": 95
+    },
+    {
+      "epoch": 0.5672082717872969,
+      "grad_norm": 0.35831600427627563,
+      "learning_rate": 1.9070024148017375e-05,
+      "loss": 0.3409,
+      "step": 96
+    },
+    {
+      "epoch": 0.5731166912850812,
+      "grad_norm": 0.39858701825141907,
+      "learning_rate": 1.9042091688505104e-05,
+      "loss": 0.3319,
+      "step": 97
+    },
+    {
+      "epoch": 0.5790251107828656,
+      "grad_norm": 0.3343643546104431,
+      "learning_rate": 1.9013767008461236e-05,
+      "loss": 0.3352,
+      "step": 98
+    },
+    {
+      "epoch": 0.5849335302806499,
+      "grad_norm": 0.3519919216632843,
+      "learning_rate": 1.89850513365308e-05,
+      "loss": 0.3634,
+      "step": 99
+    },
+    {
+      "epoch": 0.5908419497784343,
+      "grad_norm": 0.32900717854499817,
+      "learning_rate": 1.895594591831896e-05,
+      "loss": 0.3415,
+      "step": 100
+    },
+    {
+      "epoch": 0.5967503692762186,
+      "grad_norm": 0.34432175755500793,
+      "learning_rate": 1.8926452016336987e-05,
+      "loss": 0.3169,
+      "step": 101
+    },
+    {
+      "epoch": 0.6026587887740029,
+      "grad_norm": 0.33144107460975647,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 0.3431,
+      "step": 102
+    },
+    {
+      "epoch": 0.6085672082717873,
+      "grad_norm": 0.3299802839756012,
+      "learning_rate": 1.8866303895308856e-05,
+      "loss": 0.3411,
+      "step": 103
+    },
+    {
+      "epoch": 0.6144756277695717,
+      "grad_norm": 0.30740225315093994,
+      "learning_rate": 1.883565228531919e-05,
+      "loss": 0.3355,
+      "step": 104
+    },
+    {
+      "epoch": 0.620384047267356,
+      "grad_norm": 0.34325993061065674,
+      "learning_rate": 1.88046174095592e-05,
+      "loss": 0.3188,
+      "step": 105
+    },
+    {
+      "epoch": 0.6262924667651403,
+      "grad_norm": 0.3394065797328949,
+      "learning_rate": 1.8773200614234587e-05,
+      "loss": 0.3153,
+      "step": 106
+    },
+    {
+      "epoch": 0.6322008862629247,
+      "grad_norm": 0.35468512773513794,
+      "learning_rate": 1.874140326211766e-05,
+      "loss": 0.3387,
+      "step": 107
+    },
+    {
+      "epoch": 0.638109305760709,
+      "grad_norm": 0.36726799607276917,
+      "learning_rate": 1.8709226732488216e-05,
+      "loss": 0.3457,
+      "step": 108
+    },
+    {
+      "epoch": 0.6440177252584933,
+      "grad_norm": 0.3223711848258972,
+      "learning_rate": 1.86766724210737e-05,
+      "loss": 0.3588,
+      "step": 109
+    },
+    {
+      "epoch": 0.6499261447562777,
+      "grad_norm": 0.3537541925907135,
+      "learning_rate": 1.8643741739988672e-05,
+      "loss": 0.3506,
+      "step": 110
+    },
+    {
+      "epoch": 0.6558345642540621,
+      "grad_norm": 0.3755073845386505,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 0.3221,
+      "step": 111
+    },
+    {
+      "epoch": 0.6617429837518464,
+      "grad_norm": 0.31778833270072937,
+      "learning_rate": 1.8576756998832667e-05,
+      "loss": 0.3161,
+      "step": 112
+    },
+    {
+      "epoch": 0.6676514032496307,
+      "grad_norm": 0.3517738878726959,
+      "learning_rate": 1.8542705844371544e-05,
+      "loss": 0.3442,
+      "step": 113
+    },
+    {
+      "epoch": 0.6735598227474151,
+      "grad_norm": 0.3254755139350891,
+      "learning_rate": 1.8508284131333604e-05,
+      "loss": 0.3372,
+      "step": 114
+    },
+    {
+      "epoch": 0.6735598227474151,
+      "eval_loss": 0.363791823387146,
+      "eval_runtime": 4.0908,
+      "eval_samples_per_second": 13.445,
+      "eval_steps_per_second": 1.711,
+      "step": 114
+    },
+    {
+      "epoch": 0.6794682422451994,
+      "grad_norm": 0.3458060622215271,
+      "learning_rate": 1.8473493352836032e-05,
+      "loss": 0.3329,
+      "step": 115
+    },
+    {
+      "epoch": 0.6853766617429837,
+      "grad_norm": 0.33962881565093994,
+      "learning_rate": 1.8438335018005052e-05,
+      "loss": 0.3478,
+      "step": 116
+    },
+    {
+      "epoch": 0.691285081240768,
+      "grad_norm": 0.33980926871299744,
+      "learning_rate": 1.8402810651910444e-05,
+      "loss": 0.3484,
+      "step": 117
+    },
+    {
+      "epoch": 0.6971935007385525,
+      "grad_norm": 0.355694979429245,
+      "learning_rate": 1.8366921795499394e-05,
+      "loss": 0.3686,
+      "step": 118
+    },
+    {
+      "epoch": 0.7031019202363368,
+      "grad_norm": 0.3415476083755493,
+      "learning_rate": 1.8330670005529657e-05,
+      "loss": 0.3204,
+      "step": 119
+    },
+    {
+      "epoch": 0.7090103397341211,
+      "grad_norm": 0.3336890935897827,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 0.3323,
+      "step": 120
+    },
+    {
+      "epoch": 0.7149187592319055,
+      "grad_norm": 0.34337785840034485,
+      "learning_rate": 1.8257083930592102e-05,
+      "loss": 0.3283,
+      "step": 121
+    },
+    {
+      "epoch": 0.7208271787296898,
+      "grad_norm": 0.3578524887561798,
+      "learning_rate": 1.8219752837581466e-05,
+      "loss": 0.3326,
+      "step": 122
+    },
+    {
+      "epoch": 0.7267355982274741,
+      "grad_norm": 0.32392922043800354,
+      "learning_rate": 1.8182065194788024e-05,
+      "loss": 0.3141,
+      "step": 123
+    },
+    {
+      "epoch": 0.7326440177252584,
+      "grad_norm": 0.36127492785453796,
+      "learning_rate": 1.814402263699584e-05,
+      "loss": 0.3461,
+      "step": 124
+    },
+    {
+      "epoch": 0.7385524372230429,
+      "grad_norm": 0.33812931180000305,
+      "learning_rate": 1.8105626814384173e-05,
+      "loss": 0.3404,
+      "step": 125
+    },
+    {
+      "epoch": 0.7444608567208272,
+      "grad_norm": 0.3138431906700134,
+      "learning_rate": 1.8066879392455932e-05,
+      "loss": 0.3237,
+      "step": 126
+    },
+    {
+      "epoch": 0.7503692762186115,
+      "grad_norm": 0.33033978939056396,
+      "learning_rate": 1.8027782051965408e-05,
+      "loss": 0.3416,
+      "step": 127
+    },
+    {
+      "epoch": 0.7562776957163959,
+      "grad_norm": 0.3907163143157959,
+      "learning_rate": 1.7988336488845374e-05,
+      "loss": 0.3352,
+      "step": 128
+    },
+    {
+      "epoch": 0.7621861152141802,
+      "grad_norm": 0.315248042345047,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 0.3225,
+      "step": 129
+    },
+    {
+      "epoch": 0.7680945347119645,
+      "grad_norm": 0.3284492790699005,
+      "learning_rate": 1.7908407553898282e-05,
+      "loss": 0.3217,
+      "step": 130
+    },
+    {
+      "epoch": 0.7740029542097489,
+      "grad_norm": 0.3439176082611084,
+      "learning_rate": 1.7867927649163838e-05,
+      "loss": 0.3367,
+      "step": 131
+    },
+    {
+      "epoch": 0.7799113737075333,
+      "grad_norm": 0.31954073905944824,
+      "learning_rate": 1.782710645583473e-05,
+      "loss": 0.3133,
+      "step": 132
+    },
+    {
+      "epoch": 0.7858197932053176,
+      "grad_norm": 0.38416293263435364,
+      "learning_rate": 1.7785945744619642e-05,
+      "loss": 0.3484,
+      "step": 133
+    },
+    {
+      "epoch": 0.7917282127031019,
+      "grad_norm": 0.34139737486839294,
+      "learning_rate": 1.774444730095456e-05,
+      "loss": 0.3042,
+      "step": 134
+    },
+    {
+      "epoch": 0.7976366322008862,
+      "grad_norm": 0.3623535931110382,
+      "learning_rate": 1.7702612924925377e-05,
+      "loss": 0.3318,
+      "step": 135
+    },
+    {
+      "epoch": 0.8035450516986706,
+      "grad_norm": 0.32973209023475647,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.3092,
+      "step": 136
+    },
+    {
+      "epoch": 0.8094534711964549,
+      "grad_norm": 0.30704402923583984,
+      "learning_rate": 1.761794364889855e-05,
+      "loss": 0.321,
+      "step": 137
+    },
+    {
+      "epoch": 0.8153618906942393,
+      "grad_norm": 0.34877485036849976,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 0.3266,
+      "step": 138
+    },
+    {
+      "epoch": 0.8212703101920237,
+      "grad_norm": 0.3538282811641693,
+      "learning_rate": 1.7531952607241033e-05,
+      "loss": 0.3703,
+      "step": 139
+    },
+    {
+      "epoch": 0.827178729689808,
+      "grad_norm": 0.35590365529060364,
+      "learning_rate": 1.7488466077924525e-05,
+      "loss": 0.3506,
+      "step": 140
+    },
+    {
+      "epoch": 0.8330871491875923,
+      "grad_norm": 0.33215418457984924,
+      "learning_rate": 1.7444654719990128e-05,
+      "loss": 0.3207,
+      "step": 141
+    },
+    {
+      "epoch": 0.8389955686853766,
+      "grad_norm": 0.3381923735141754,
+      "learning_rate": 1.7400520433851457e-05,
+      "loss": 0.3237,
+      "step": 142
+    },
+    {
+      "epoch": 0.844903988183161,
+      "grad_norm": 0.3371356129646301,
+      "learning_rate": 1.735606513392984e-05,
+      "loss": 0.3394,
+      "step": 143
+    },
+    {
+      "epoch": 0.8508124076809453,
+      "grad_norm": 0.344291627407074,
+      "learning_rate": 1.7311290748571273e-05,
+      "loss": 0.3604,
+      "step": 144
+    },
+    {
+      "epoch": 0.8567208271787297,
+      "grad_norm": 0.3567575216293335,
+      "learning_rate": 1.72661992199628e-05,
+      "loss": 0.3518,
+      "step": 145
+    },
+    {
+      "epoch": 0.8626292466765141,
+      "grad_norm": 0.33762165904045105,
+      "learning_rate": 1.7220792504048227e-05,
+      "loss": 0.3146,
+      "step": 146
+    },
+    {
+      "epoch": 0.8685376661742984,
+      "grad_norm": 0.3404117822647095,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 0.3192,
+      "step": 147
+    },
+    {
+      "epoch": 0.8744460856720827,
+      "grad_norm": 0.3535095751285553,
+      "learning_rate": 1.7129041402350317e-05,
+      "loss": 0.3364,
+      "step": 148
+    },
+    {
+      "epoch": 0.880354505169867,
+      "grad_norm": 0.3418992757797241,
+      "learning_rate": 1.708270099647198e-05,
+      "loss": 0.3327,
+      "step": 149
+    },
+    {
+      "epoch": 0.8862629246676514,
+      "grad_norm": 0.3172495663166046,
+      "learning_rate": 1.7036053362924896e-05,
+      "loss": 0.3404,
+      "step": 150
+    },
+    {
+      "epoch": 0.8921713441654358,
+      "grad_norm": 0.3307952284812927,
+      "learning_rate": 1.6989100525152346e-05,
+      "loss": 0.3279,
+      "step": 151
+    },
+    {
+      "epoch": 0.8980797636632201,
+      "grad_norm": 0.29014381766319275,
+      "learning_rate": 1.694184451983651e-05,
+      "loss": 0.3027,
+      "step": 152
+    },
+    {
+      "epoch": 0.9039881831610044,
+      "grad_norm": 0.3290538191795349,
+      "learning_rate": 1.689428739681012e-05,
+      "loss": 0.3297,
+      "step": 153
+    },
+    {
+      "epoch": 0.9098966026587888,
+      "grad_norm": 0.3165034353733063,
+      "learning_rate": 1.684643121896755e-05,
+      "loss": 0.3225,
+      "step": 154
+    },
+    {
+      "epoch": 0.9158050221565731,
+      "grad_norm": 0.3677435517311096,
+      "learning_rate": 1.679827806217533e-05,
+      "loss": 0.328,
+      "step": 155
+    },
+    {
+      "epoch": 0.9217134416543574,
+      "grad_norm": 0.3617594242095947,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 0.3299,
+      "step": 156
+    },
+    {
+      "epoch": 0.9276218611521418,
+      "grad_norm": 0.31069889664649963,
+      "learning_rate": 1.6701089179528032e-05,
+      "loss": 0.3146,
+      "step": 157
+    },
+    {
+      "epoch": 0.9335302806499262,
+      "grad_norm": 0.3610530197620392,
+      "learning_rate": 1.6652057669453606e-05,
+      "loss": 0.3223,
+      "step": 158
+    },
+    {
+      "epoch": 0.9394387001477105,
+      "grad_norm": 0.3169001638889313,
+      "learning_rate": 1.6602737611807975e-05,
+      "loss": 0.3194,
+      "step": 159
+    },
+    {
+      "epoch": 0.9453471196454948,
+      "grad_norm": 0.33033737540245056,
+      "learning_rate": 1.655313114595666e-05,
+      "loss": 0.3317,
+      "step": 160
+    },
+    {
+      "epoch": 0.9512555391432792,
+      "grad_norm": 0.35510334372520447,
+      "learning_rate": 1.6503240423688768e-05,
+      "loss": 0.3249,
+      "step": 161
+    },
+    {
+      "epoch": 0.9571639586410635,
+      "grad_norm": 0.356079638004303,
+      "learning_rate": 1.6453067609123656e-05,
+      "loss": 0.3274,
+      "step": 162
+    },
+    {
+      "epoch": 0.9630723781388478,
+      "grad_norm": 0.36350899934768677,
+      "learning_rate": 1.6402614878617037e-05,
+      "loss": 0.3553,
+      "step": 163
+    },
+    {
+      "epoch": 0.9689807976366323,
+      "grad_norm": 0.3371831476688385,
+      "learning_rate": 1.6351884420666616e-05,
+      "loss": 0.3245,
+      "step": 164
+    },
+    {
+      "epoch": 0.9748892171344166,
+      "grad_norm": 0.3398657739162445,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 0.3043,
+      "step": 165
+    },
+    {
+      "epoch": 0.9807976366322009,
+      "grad_norm": 0.34537115693092346,
+      "learning_rate": 1.6249599136564837e-05,
+      "loss": 0.349,
+      "step": 166
+    },
+    {
+      "epoch": 0.9867060561299852,
+      "grad_norm": 0.31506776809692383,
+      "learning_rate": 1.619804874726171e-05,
+      "loss": 0.315,
+      "step": 167
+    },
+    {
+      "epoch": 0.9926144756277696,
+      "grad_norm": 0.32844215631484985,
+      "learning_rate": 1.6146229504018777e-05,
+      "loss": 0.3247,
+      "step": 168
+    },
+    {
+      "epoch": 0.9985228951255539,
+      "grad_norm": 0.3447742760181427,
+      "learning_rate": 1.609414365460921e-05,
+      "loss": 0.3193,
+      "step": 169
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3447742760181427,
+      "learning_rate": 1.6041793458370812e-05,
+      "loss": 0.3359,
+      "step": 170
+    },
+    {
+      "epoch": 1.0059084194977843,
+      "grad_norm": 0.27635836601257324,
+      "learning_rate": 1.5989181186108003e-05,
+      "loss": 0.2579,
+      "step": 171
+    },
+    {
+      "epoch": 1.0059084194977843,
+      "eval_loss": 0.3496532440185547,
+      "eval_runtime": 4.0258,
+      "eval_samples_per_second": 13.662,
+      "eval_steps_per_second": 1.739,
+      "step": 171
+    },
+    {
+      "epoch": 1.0118168389955686,
+      "grad_norm": 0.27547529339790344,
+      "learning_rate": 1.5936309119993333e-05,
+      "loss": 0.2532,
+      "step": 172
+    },
+    {
+      "epoch": 1.017725258493353,
+      "grad_norm": 0.2674752473831177,
+      "learning_rate": 1.5883179553468465e-05,
+      "loss": 0.2413,
+      "step": 173
+    },
+    {
+      "epoch": 1.0236336779911375,
+      "grad_norm": 0.3056715428829193,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 0.2418,
+      "step": 174
+    },
+    {
+      "epoch": 1.0295420974889218,
+      "grad_norm": 0.27895164489746094,
+      "learning_rate": 1.5776157148703094e-05,
+      "loss": 0.2516,
+      "step": 175
+    },
+    {
+      "epoch": 1.035450516986706,
+      "grad_norm": 0.2935872972011566,
+      "learning_rate": 1.5722268952793806e-05,
+      "loss": 0.254,
+      "step": 176
+    },
+    {
+      "epoch": 1.0413589364844904,
+      "grad_norm": 0.28329288959503174,
+      "learning_rate": 1.566813254093538e-05,
+      "loss": 0.2356,
+      "step": 177
+    },
+    {
+      "epoch": 1.0472673559822747,
+      "grad_norm": 0.29026728868484497,
+      "learning_rate": 1.5613750261413256e-05,
+      "loss": 0.2404,
+      "step": 178
+    },
+    {
+      "epoch": 1.053175775480059,
+      "grad_norm": 0.3126751780509949,
+      "learning_rate": 1.555912447317792e-05,
+      "loss": 0.2303,
+      "step": 179
+    },
+    {
+      "epoch": 1.0590841949778433,
+      "grad_norm": 0.26517724990844727,
+      "learning_rate": 1.5504257545742585e-05,
+      "loss": 0.2175,
+      "step": 180
+    },
+    {
+      "epoch": 1.0649926144756279,
+      "grad_norm": 0.26433265209198,
+      "learning_rate": 1.5449151859080395e-05,
+      "loss": 0.2169,
+      "step": 181
+    },
+    {
+      "epoch": 1.0709010339734122,
+      "grad_norm": 0.2908313274383545,
+      "learning_rate": 1.5393809803521213e-05,
+      "loss": 0.2236,
+      "step": 182
+    },
+    {
+      "epoch": 1.0768094534711965,
+      "grad_norm": 0.2951337397098541,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 0.2305,
+      "step": 183
+    },
+    {
+      "epoch": 1.0827178729689808,
+      "grad_norm": 0.29755067825317383,
+      "learning_rate": 1.528242619819224e-05,
+      "loss": 0.2385,
+      "step": 184
+    },
+    {
+      "epoch": 1.0886262924667651,
+      "grad_norm": 0.2879098355770111,
+      "learning_rate": 1.5226389479930296e-05,
+      "loss": 0.2377,
+      "step": 185
+    },
+    {
+      "epoch": 1.0945347119645494,
+      "grad_norm": 0.2590835392475128,
+      "learning_rate": 1.517012605557746e-05,
+      "loss": 0.2312,
+      "step": 186
+    },
+    {
+      "epoch": 1.1004431314623337,
+      "grad_norm": 0.2694130837917328,
+      "learning_rate": 1.5113638365682996e-05,
+      "loss": 0.2347,
+      "step": 187
+    },
+    {
+      "epoch": 1.106351550960118,
+      "grad_norm": 0.29442402720451355,
+      "learning_rate": 1.5056928860524181e-05,
+      "loss": 0.2428,
+      "step": 188
+    },
+    {
+      "epoch": 1.1122599704579026,
+      "grad_norm": 0.29042768478393555,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2501,
+      "step": 189
+    },
+    {
+      "epoch": 1.118168389955687,
+      "grad_norm": 0.2620311975479126,
+      "learning_rate": 1.4942854253524479e-05,
+      "loss": 0.2395,
+      "step": 190
+    },
+    {
+      "epoch": 1.1240768094534712,
+      "grad_norm": 0.26113441586494446,
+      "learning_rate": 1.488549409991953e-05,
+      "loss": 0.2532,
+      "step": 191
+    },
+    {
+      "epoch": 1.1299852289512555,
+      "grad_norm": 0.2995262145996094,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 0.2319,
+      "step": 192
+    },
+    {
+      "epoch": 1.1358936484490398,
+      "grad_norm": 0.27327674627304077,
+      "learning_rate": 1.477014053300299e-05,
+      "loss": 0.2348,
+      "step": 193
+    },
+    {
+      "epoch": 1.1418020679468242,
+      "grad_norm": 0.26245003938674927,
+      "learning_rate": 1.4712152123405018e-05,
+      "loss": 0.228,
+      "step": 194
+    },
+    {
+      "epoch": 1.1477104874446087,
+      "grad_norm": 0.28888335824012756,
+      "learning_rate": 1.4653959313887813e-05,
+      "loss": 0.2436,
+      "step": 195
+    },
+    {
+      "epoch": 1.153618906942393,
+      "grad_norm": 0.2724781632423401,
+      "learning_rate": 1.4595564628691944e-05,
+      "loss": 0.2442,
+      "step": 196
+    },
+    {
+      "epoch": 1.1595273264401773,
+      "grad_norm": 0.2921780049800873,
+      "learning_rate": 1.4536970600814789e-05,
+      "loss": 0.2412,
+      "step": 197
+    },
+    {
+      "epoch": 1.1654357459379616,
+      "grad_norm": 0.27938568592071533,
+      "learning_rate": 1.4478179771900634e-05,
+      "loss": 0.2465,
+      "step": 198
+    },
+    {
+      "epoch": 1.171344165435746,
+      "grad_norm": 0.29516273736953735,
+      "learning_rate": 1.4419194692130453e-05,
+      "loss": 0.2415,
+      "step": 199
+    },
+    {
+      "epoch": 1.1772525849335302,
+      "grad_norm": 0.27947136759757996,
+      "learning_rate": 1.436001792011128e-05,
+      "loss": 0.2295,
+      "step": 200
+    },
+    {
+      "epoch": 1.1831610044313146,
+      "grad_norm": 0.26482367515563965,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 0.2273,
+      "step": 201
+    },
+    {
+      "epoch": 1.1890694239290989,
+      "grad_norm": 0.2728091776371002,
+      "learning_rate": 1.424109957521806e-05,
+      "loss": 0.2227,
+      "step": 202
+    },
+    {
+      "epoch": 1.1949778434268834,
+      "grad_norm": 0.28748828172683716,
+      "learning_rate": 1.4181363160687693e-05,
+      "loss": 0.2402,
+      "step": 203
+    },
+    {
+      "epoch": 1.2008862629246677,
+      "grad_norm": 0.2891993820667267,
+      "learning_rate": 1.4121445370371922e-05,
+      "loss": 0.224,
+      "step": 204
+    },
+    {
+      "epoch": 1.206794682422452,
+      "grad_norm": 0.24767152965068817,
+      "learning_rate": 1.4061348803336135e-05,
+      "loss": 0.221,
+      "step": 205
+    },
+    {
+      "epoch": 1.2127031019202363,
+      "grad_norm": 0.2819165885448456,
+      "learning_rate": 1.400107606640056e-05,
+      "loss": 0.2231,
+      "step": 206
+    },
+    {
+      "epoch": 1.2186115214180206,
+      "grad_norm": 0.27328819036483765,
+      "learning_rate": 1.394062977402717e-05,
+      "loss": 0.229,
+      "step": 207
+    },
+    {
+      "epoch": 1.224519940915805,
+      "grad_norm": 0.2674582302570343,
+      "learning_rate": 1.3880012548206292e-05,
+      "loss": 0.2155,
+      "step": 208
+    },
+    {
+      "epoch": 1.2304283604135893,
+      "grad_norm": 0.2989075481891632,
+      "learning_rate": 1.3819227018342865e-05,
+      "loss": 0.2184,
+      "step": 209
+    },
+    {
+      "epoch": 1.2363367799113738,
+      "grad_norm": 0.30796098709106445,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 0.2288,
+      "step": 210
+    },
+    {
+      "epoch": 1.2422451994091581,
+      "grad_norm": 0.29833805561065674,
+      "learning_rate": 1.3697161600496525e-05,
+      "loss": 0.2368,
+      "step": 211
+    },
+    {
+      "epoch": 1.2481536189069424,
+      "grad_norm": 0.26458829641342163,
+      "learning_rate": 1.3635887007368467e-05,
+      "loss": 0.2376,
+      "step": 212
+    },
+    {
+      "epoch": 1.2540620384047267,
+      "grad_norm": 0.2781698703765869,
+      "learning_rate": 1.3574454699677893e-05,
+      "loss": 0.2167,
+      "step": 213
+    },
+    {
+      "epoch": 1.259970457902511,
+      "grad_norm": 0.268433153629303,
+      "learning_rate": 1.3512867342185705e-05,
+      "loss": 0.2229,
+      "step": 214
+    },
+    {
+      "epoch": 1.2658788774002954,
+      "grad_norm": 0.2726047933101654,
+      "learning_rate": 1.3451127606378425e-05,
+      "loss": 0.223,
+      "step": 215
+    },
+    {
+      "epoch": 1.2717872968980797,
+      "grad_norm": 0.29567429423332214,
+      "learning_rate": 1.3389238170352318e-05,
+      "loss": 0.2105,
+      "step": 216
+    },
+    {
+      "epoch": 1.277695716395864,
+      "grad_norm": 0.30303359031677246,
+      "learning_rate": 1.3327201718697232e-05,
+      "loss": 0.2602,
+      "step": 217
+    },
+    {
+      "epoch": 1.2836041358936485,
+      "grad_norm": 0.27332380414009094,
+      "learning_rate": 1.326502094238013e-05,
+      "loss": 0.2288,
+      "step": 218
+    },
+    {
+      "epoch": 1.2895125553914328,
+      "grad_norm": 0.2703614830970764,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 0.2308,
+      "step": 219
+    },
+    {
+      "epoch": 1.2954209748892171,
+      "grad_norm": 0.2788908779621124,
+      "learning_rate": 1.3140237210812741e-05,
+      "loss": 0.2254,
+      "step": 220
+    },
+    {
+      "epoch": 1.3013293943870015,
+      "grad_norm": 0.27442580461502075,
+      "learning_rate": 1.3077639668330124e-05,
+      "loss": 0.2158,
+      "step": 221
+    },
+    {
+      "epoch": 1.3072378138847858,
+      "grad_norm": 0.28895896673202515,
+      "learning_rate": 1.3014908626486032e-05,
+      "loss": 0.2404,
+      "step": 222
+    },
+    {
+      "epoch": 1.31314623338257,
+      "grad_norm": 0.24982582032680511,
+      "learning_rate": 1.2952046806376806e-05,
+      "loss": 0.2201,
+      "step": 223
+    },
+    {
+      "epoch": 1.3190546528803546,
+      "grad_norm": 0.28909650444984436,
+      "learning_rate": 1.2889056934771577e-05,
+      "loss": 0.2384,
+      "step": 224
+    },
+    {
+      "epoch": 1.324963072378139,
+      "grad_norm": 0.28018954396247864,
+      "learning_rate": 1.282594174399399e-05,
+      "loss": 0.2324,
+      "step": 225
+    },
+    {
+      "epoch": 1.3308714918759232,
+      "grad_norm": 0.29922735691070557,
+      "learning_rate": 1.2762703971803684e-05,
+      "loss": 0.2457,
+      "step": 226
+    },
+    {
+      "epoch": 1.3367799113737076,
+      "grad_norm": 0.289288729429245,
+      "learning_rate": 1.2699346361277538e-05,
+      "loss": 0.2366,
+      "step": 227
+    },
+    {
+      "epoch": 1.3426883308714919,
+      "grad_norm": 0.2790012061595917,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 0.2359,
+      "step": 228
+    },
+    {
+      "epoch": 1.3426883308714919,
+      "eval_loss": 0.35204342007637024,
+      "eval_runtime": 4.4578,
+      "eval_samples_per_second": 12.338,
+      "eval_steps_per_second": 1.57,
+      "step": 228
+    },
+    {
+      "epoch": 1.3485967503692762,
+      "grad_norm": 0.36030444502830505,
+      "learning_rate": 1.2572282623397268e-05,
+      "loss": 0.2405,
+      "step": 229
+    },
+    {
+      "epoch": 1.3545051698670605,
+      "grad_norm": 0.24079382419586182,
+      "learning_rate": 1.2508582007711074e-05,
+      "loss": 0.2148,
+      "step": 230
+    },
+    {
+      "epoch": 1.3604135893648448,
+      "grad_norm": 0.26674559712409973,
+      "learning_rate": 1.2444772576785828e-05,
+      "loss": 0.2457,
+      "step": 231
+    },
+    {
+      "epoch": 1.3663220088626291,
+      "grad_norm": 0.25345727801322937,
+      "learning_rate": 1.2380857098495355e-05,
+      "loss": 0.2229,
+      "step": 232
+    },
+    {
+      "epoch": 1.3722304283604136,
+      "grad_norm": 0.2623337507247925,
+      "learning_rate": 1.2316838345313517e-05,
+      "loss": 0.231,
+      "step": 233
+    },
+    {
+      "epoch": 1.378138847858198,
+      "grad_norm": 0.27783095836639404,
+      "learning_rate": 1.225271909419395e-05,
+      "loss": 0.2251,
+      "step": 234
+    },
+    {
+      "epoch": 1.3840472673559823,
+      "grad_norm": 0.25021976232528687,
+      "learning_rate": 1.2188502126449616e-05,
+      "loss": 0.226,
+      "step": 235
+    },
+    {
+      "epoch": 1.3899556868537666,
+      "grad_norm": 0.2695038318634033,
+      "learning_rate": 1.2124190227632138e-05,
+      "loss": 0.2438,
+      "step": 236
+    },
+    {
+      "epoch": 1.395864106351551,
+      "grad_norm": 0.24312005937099457,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 0.2138,
+      "step": 237
+    },
+    {
+      "epoch": 1.4017725258493354,
+      "grad_norm": 0.2761548161506653,
+      "learning_rate": 1.1995292799452472e-05,
+      "loss": 0.244,
+      "step": 238
+    },
+    {
+      "epoch": 1.4076809453471197,
+      "grad_norm": 0.2740529477596283,
+      "learning_rate": 1.1930712861298553e-05,
+      "loss": 0.2416,
+      "step": 239
+    },
+    {
+      "epoch": 1.413589364844904,
+      "grad_norm": 0.2605426013469696,
+      "learning_rate": 1.186604917424549e-05,
+      "loss": 0.2515,
+      "step": 240
+    },
+    {
+      "epoch": 1.4194977843426884,
+      "grad_norm": 0.27557292580604553,
+      "learning_rate": 1.1801304543222349e-05,
+      "loss": 0.232,
+      "step": 241
+    },
+    {
+      "epoch": 1.4254062038404727,
+      "grad_norm": 0.2512328624725342,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.2311,
+      "step": 242
+    },
+    {
+      "epoch": 1.431314623338257,
+      "grad_norm": 0.2634104788303375,
+      "learning_rate": 1.1671583686415833e-05,
+      "loss": 0.2207,
+      "step": 243
+    },
+    {
+      "epoch": 1.4372230428360413,
+      "grad_norm": 0.2541881203651428,
+      "learning_rate": 1.1606613087558748e-05,
+      "loss": 0.2207,
+      "step": 244
+    },
+    {
+      "epoch": 1.4431314623338256,
+      "grad_norm": 0.24408863484859467,
+      "learning_rate": 1.1541572798340076e-05,
+      "loss": 0.2155,
+      "step": 245
+    },
+    {
+      "epoch": 1.44903988183161,
+      "grad_norm": 0.25305289030075073,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 0.2245,
+      "step": 246
+    },
+    {
+      "epoch": 1.4549483013293945,
+      "grad_norm": 0.26579606533050537,
+      "learning_rate": 1.1411294436778562e-05,
+      "loss": 0.2295,
+      "step": 247
+    },
+    {
+      "epoch": 1.4608567208271788,
+      "grad_norm": 0.26332345604896545,
+      "learning_rate": 1.1346062015544997e-05,
+      "loss": 0.2363,
+      "step": 248
+    },
+    {
+      "epoch": 1.466765140324963,
+      "grad_norm": 0.2519514262676239,
+      "learning_rate": 1.1280771205923269e-05,
+      "loss": 0.2215,
+      "step": 249
+    },
+    {
+      "epoch": 1.4726735598227474,
+      "grad_norm": 0.2569345533847809,
+      "learning_rate": 1.1215424840045254e-05,
+      "loss": 0.223,
+      "step": 250
+    },
+    {
+      "epoch": 1.4785819793205317,
+      "grad_norm": 0.25557035207748413,
+      "learning_rate": 1.1150025752452693e-05,
+      "loss": 0.2511,
+      "step": 251
+    },
+    {
+      "epoch": 1.4844903988183162,
+      "grad_norm": 0.26646342873573303,
+      "learning_rate": 1.1084576779974257e-05,
+      "loss": 0.2476,
+      "step": 252
+    },
+    {
+      "epoch": 1.4903988183161005,
+      "grad_norm": 0.27917614579200745,
+      "learning_rate": 1.1019080761602473e-05,
+      "loss": 0.2284,
+      "step": 253
+    },
+    {
+      "epoch": 1.4963072378138849,
+      "grad_norm": 0.2594425082206726,
+      "learning_rate": 1.0953540538370591e-05,
+      "loss": 0.2319,
+      "step": 254
+    },
+    {
+      "epoch": 1.5022156573116692,
+      "grad_norm": 0.23648317158222198,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 0.225,
+      "step": 255
+    },
+    {
+      "epoch": 1.5081240768094535,
+      "grad_norm": 0.24810343980789185,
+      "learning_rate": 1.0822338850923644e-05,
+      "loss": 0.2222,
+      "step": 256
+    },
+    {
+      "epoch": 1.5140324963072378,
+      "grad_norm": 0.25305667519569397,
+      "learning_rate": 1.0756683077869133e-05,
+      "loss": 0.2178,
+      "step": 257
+    },
+    {
+      "epoch": 1.519940915805022,
+      "grad_norm": 0.23994190990924835,
+      "learning_rate": 1.069099448202878e-05,
+      "loss": 0.2274,
+      "step": 258
+    },
+    {
+      "epoch": 1.5258493353028064,
+      "grad_norm": 0.28112536668777466,
+      "learning_rate": 1.0625275912789307e-05,
+      "loss": 0.2157,
+      "step": 259
+    },
+    {
+      "epoch": 1.5317577548005907,
+      "grad_norm": 0.2910768687725067,
+      "learning_rate": 1.0559530220837593e-05,
+      "loss": 0.2337,
+      "step": 260
+    },
+    {
+      "epoch": 1.537666174298375,
+      "grad_norm": 0.26320862770080566,
+      "learning_rate": 1.049376025803703e-05,
+      "loss": 0.2156,
+      "step": 261
+    },
+    {
+      "epoch": 1.5435745937961596,
+      "grad_norm": 0.2653874456882477,
+      "learning_rate": 1.0427968877303809e-05,
+      "loss": 0.2269,
+      "step": 262
+    },
+    {
+      "epoch": 1.549483013293944,
+      "grad_norm": 0.24998469650745392,
+      "learning_rate": 1.0362158932483165e-05,
+      "loss": 0.2252,
+      "step": 263
+    },
+    {
+      "epoch": 1.5553914327917282,
+      "grad_norm": 0.25920990109443665,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 0.2274,
+      "step": 264
+    },
+    {
+      "epoch": 1.5612998522895125,
+      "grad_norm": 0.2827723026275635,
+      "learning_rate": 1.023049476986304e-05,
+      "loss": 0.248,
+      "step": 265
+    },
+    {
+      "epoch": 1.567208271787297,
+      "grad_norm": 0.27848076820373535,
+      "learning_rate": 1.0164646263284993e-05,
+      "loss": 0.2372,
+      "step": 266
+    },
+    {
+      "epoch": 1.5731166912850814,
+      "grad_norm": 0.2601296305656433,
+      "learning_rate": 1.0098790614814658e-05,
+      "loss": 0.212,
+      "step": 267
+    },
+    {
+      "epoch": 1.5790251107828657,
+      "grad_norm": 0.24360589683055878,
+      "learning_rate": 1.0032930681085028e-05,
+      "loss": 0.2152,
+      "step": 268
+    },
+    {
+      "epoch": 1.58493353028065,
+      "grad_norm": 0.3080978989601135,
+      "learning_rate": 9.967069318914977e-06,
+      "loss": 0.2218,
+      "step": 269
+    },
+    {
+      "epoch": 1.5908419497784343,
+      "grad_norm": 0.26208099722862244,
+      "learning_rate": 9.901209385185345e-06,
+      "loss": 0.2184,
+      "step": 270
+    },
+    {
+      "epoch": 1.5967503692762186,
+      "grad_norm": 0.2984671890735626,
+      "learning_rate": 9.835353736715007e-06,
+      "loss": 0.2432,
+      "step": 271
+    },
+    {
+      "epoch": 1.602658788774003,
+      "grad_norm": 0.26782581210136414,
+      "learning_rate": 9.769505230136962e-06,
+      "loss": 0.2126,
+      "step": 272
+    },
+    {
+      "epoch": 1.6085672082717872,
+      "grad_norm": 0.28440967202186584,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 0.2214,
+      "step": 273
+    },
+    {
+      "epoch": 1.6144756277695715,
+      "grad_norm": 0.2926226854324341,
+      "learning_rate": 9.637841067516837e-06,
+      "loss": 0.2256,
+      "step": 274
+    },
+    {
+      "epoch": 1.6203840472673559,
+      "grad_norm": 0.25548121333122253,
+      "learning_rate": 9.572031122696196e-06,
+      "loss": 0.2304,
+      "step": 275
+    },
+    {
+      "epoch": 1.6262924667651402,
+      "grad_norm": 0.28455373644828796,
+      "learning_rate": 9.506239741962971e-06,
+      "loss": 0.2299,
+      "step": 276
+    },
+    {
+      "epoch": 1.6322008862629247,
+      "grad_norm": 0.262614369392395,
+      "learning_rate": 9.440469779162407e-06,
+      "loss": 0.2251,
+      "step": 277
+    },
+    {
+      "epoch": 1.638109305760709,
+      "grad_norm": 0.27394819259643555,
+      "learning_rate": 9.374724087210698e-06,
+      "loss": 0.2117,
+      "step": 278
+    },
+    {
+      "epoch": 1.6440177252584933,
+      "grad_norm": 0.2843812108039856,
+      "learning_rate": 9.309005517971222e-06,
+      "loss": 0.2268,
+      "step": 279
+    },
+    {
+      "epoch": 1.6499261447562779,
+      "grad_norm": 0.25647154450416565,
+      "learning_rate": 9.24331692213087e-06,
+      "loss": 0.2187,
+      "step": 280
+    },
+    {
+      "epoch": 1.6558345642540622,
+      "grad_norm": 0.27861371636390686,
+      "learning_rate": 9.17766114907636e-06,
+      "loss": 0.2311,
+      "step": 281
+    },
+    {
+      "epoch": 1.6617429837518465,
+      "grad_norm": 0.270049512386322,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 0.2265,
+      "step": 282
+    },
+    {
+      "epoch": 1.6676514032496308,
+      "grad_norm": 0.2750328779220581,
+      "learning_rate": 9.04645946162941e-06,
+      "loss": 0.2253,
+      "step": 283
+    },
+    {
+      "epoch": 1.673559822747415,
+      "grad_norm": 0.2412230521440506,
+      "learning_rate": 8.980919238397532e-06,
+      "loss": 0.2394,
+      "step": 284
+    },
+    {
+      "epoch": 1.6794682422451994,
+      "grad_norm": 0.2524693012237549,
+      "learning_rate": 8.915423220025747e-06,
+      "loss": 0.2258,
+      "step": 285
+    },
+    {
+      "epoch": 1.6794682422451994,
+      "eval_loss": 0.3460842967033386,
+      "eval_runtime": 4.0784,
+      "eval_samples_per_second": 13.486,
+      "eval_steps_per_second": 1.716,
+      "step": 285
+    },
+    {
+      "epoch": 1.6853766617429837,
+      "grad_norm": 0.25439098477363586,
+      "learning_rate": 8.849974247547307e-06,
+      "loss": 0.2266,
+      "step": 286
+    },
+    {
+      "epoch": 1.691285081240768,
+      "grad_norm": 0.257929265499115,
+      "learning_rate": 8.784575159954748e-06,
+      "loss": 0.2133,
+      "step": 287
+    },
+    {
+      "epoch": 1.6971935007385524,
+      "grad_norm": 0.24912972748279572,
+      "learning_rate": 8.719228794076733e-06,
+      "loss": 0.2129,
+      "step": 288
+    },
+    {
+      "epoch": 1.7031019202363367,
+      "grad_norm": 0.27103564143180847,
+      "learning_rate": 8.653937984455007e-06,
+      "loss": 0.2276,
+      "step": 289
+    },
+    {
+      "epoch": 1.709010339734121,
+      "grad_norm": 0.2718878984451294,
+      "learning_rate": 8.588705563221444e-06,
+      "loss": 0.2276,
+      "step": 290
+    },
+    {
+      "epoch": 1.7149187592319055,
+      "grad_norm": 0.26431816816329956,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 0.2328,
+      "step": 291
+    },
+    {
+      "epoch": 1.7208271787296898,
+      "grad_norm": 0.2725984752178192,
+      "learning_rate": 8.458427201659926e-06,
+      "loss": 0.2292,
+      "step": 292
+    },
+    {
+      "epoch": 1.7267355982274741,
+      "grad_norm": 0.2515108585357666,
+      "learning_rate": 8.393386912441257e-06,
+      "loss": 0.226,
+      "step": 293
+    },
+    {
+      "epoch": 1.7326440177252584,
+      "grad_norm": 0.2476361244916916,
+      "learning_rate": 8.328416313584169e-06,
+      "loss": 0.2277,
+      "step": 294
+    },
+    {
+      "epoch": 1.738552437223043,
+      "grad_norm": 0.25414201617240906,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.2268,
+      "step": 295
+    },
+    {
+      "epoch": 1.7444608567208273,
+      "grad_norm": 0.26264503598213196,
+      "learning_rate": 8.198695456777653e-06,
+      "loss": 0.2193,
+      "step": 296
+    },
+    {
+      "epoch": 1.7503692762186116,
+      "grad_norm": 0.26917147636413574,
+      "learning_rate": 8.133950825754511e-06,
+      "loss": 0.2251,
+      "step": 297
+    },
+    {
+      "epoch": 1.756277695716396,
+      "grad_norm": 0.2692192792892456,
+      "learning_rate": 8.069287138701452e-06,
+      "loss": 0.232,
+      "step": 298
+    },
+    {
+      "epoch": 1.7621861152141802,
+      "grad_norm": 0.27494263648986816,
+      "learning_rate": 8.004707200547534e-06,
+      "loss": 0.2461,
+      "step": 299
+    },
+    {
+      "epoch": 1.7680945347119645,
+      "grad_norm": 0.28247448801994324,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 0.2226,
+      "step": 300
+    },
+    {
+      "epoch": 1.7740029542097489,
+      "grad_norm": 0.2632560133934021,
+      "learning_rate": 7.875809772367867e-06,
+      "loss": 0.216,
+      "step": 301
+    },
+    {
+      "epoch": 1.7799113737075332,
+      "grad_norm": 0.26561063528060913,
+      "learning_rate": 7.81149787355039e-06,
+      "loss": 0.2286,
+      "step": 302
+    },
+    {
+      "epoch": 1.7858197932053175,
+      "grad_norm": 0.24065916240215302,
+      "learning_rate": 7.747280905806051e-06,
+      "loss": 0.2201,
+      "step": 303
+    },
+    {
+      "epoch": 1.7917282127031018,
+      "grad_norm": 0.288473904132843,
+      "learning_rate": 7.683161654686486e-06,
+      "loss": 0.2179,
+      "step": 304
+    },
+    {
+      "epoch": 1.797636632200886,
+      "grad_norm": 0.27798035740852356,
+      "learning_rate": 7.619142901504649e-06,
+      "loss": 0.2341,
+      "step": 305
+    },
+    {
+      "epoch": 1.8035450516986706,
+      "grad_norm": 0.28387168049812317,
+      "learning_rate": 7.555227423214174e-06,
+      "loss": 0.226,
+      "step": 306
+    },
+    {
+      "epoch": 1.809453471196455,
+      "grad_norm": 0.28974682092666626,
+      "learning_rate": 7.491417992288927e-06,
+      "loss": 0.2296,
+      "step": 307
+    },
+    {
+      "epoch": 1.8153618906942393,
+      "grad_norm": 0.26052042841911316,
+      "learning_rate": 7.427717376602739e-06,
+      "loss": 0.2002,
+      "step": 308
+    },
+    {
+      "epoch": 1.8212703101920238,
+      "grad_norm": 0.29558730125427246,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 0.263,
+      "step": 309
+    },
+    {
+      "epoch": 1.827178729689808,
+      "grad_norm": 0.24457122385501862,
+      "learning_rate": 7.300653638722463e-06,
+      "loss": 0.224,
+      "step": 310
+    },
+    {
+      "epoch": 1.8330871491875924,
+      "grad_norm": 0.2517196834087372,
+      "learning_rate": 7.2372960281963165e-06,
+      "loss": 0.2134,
+      "step": 311
+    },
+    {
+      "epoch": 1.8389955686853767,
+      "grad_norm": 0.27632561326026917,
+      "learning_rate": 7.174058256006012e-06,
+      "loss": 0.2229,
+      "step": 312
+    },
+    {
+      "epoch": 1.844903988183161,
+      "grad_norm": 0.2603515684604645,
+      "learning_rate": 7.110943065228425e-06,
+      "loss": 0.2299,
+      "step": 313
+    },
+    {
+      "epoch": 1.8508124076809453,
+      "grad_norm": 0.24517123401165009,
+      "learning_rate": 7.047953193623195e-06,
+      "loss": 0.2096,
+      "step": 314
+    },
+    {
+      "epoch": 1.8567208271787297,
+      "grad_norm": 0.24135427176952362,
+      "learning_rate": 6.985091373513972e-06,
+      "loss": 0.2072,
+      "step": 315
+    },
+    {
+      "epoch": 1.862629246676514,
+      "grad_norm": 0.2676647901535034,
+      "learning_rate": 6.92236033166988e-06,
+      "loss": 0.2173,
+      "step": 316
+    },
+    {
+      "epoch": 1.8685376661742983,
+      "grad_norm": 0.2504200041294098,
+      "learning_rate": 6.859762789187259e-06,
+      "loss": 0.2192,
+      "step": 317
+    },
+    {
+      "epoch": 1.8744460856720826,
+      "grad_norm": 0.26364269852638245,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 0.2193,
+      "step": 318
+    },
+    {
+      "epoch": 1.880354505169867,
+      "grad_norm": 0.24448218941688538,
+      "learning_rate": 6.734979057619873e-06,
+      "loss": 0.2208,
+      "step": 319
+    },
+    {
+      "epoch": 1.8862629246676514,
+      "grad_norm": 0.24706940352916718,
+      "learning_rate": 6.67279828130277e-06,
+      "loss": 0.2211,
+      "step": 320
+    },
+    {
+      "epoch": 1.8921713441654358,
+      "grad_norm": 0.24761930108070374,
+      "learning_rate": 6.610761829647685e-06,
+      "loss": 0.2222,
+      "step": 321
+    },
+    {
+      "epoch": 1.89807976366322,
+      "grad_norm": 0.2566414475440979,
+      "learning_rate": 6.548872393621578e-06,
+      "loss": 0.2136,
+      "step": 322
+    },
+    {
+      "epoch": 1.9039881831610044,
+      "grad_norm": 0.2611066401004791,
+      "learning_rate": 6.487132657814297e-06,
+      "loss": 0.2146,
+      "step": 323
+    },
+    {
+      "epoch": 1.909896602658789,
+      "grad_norm": 0.27130842208862305,
+      "learning_rate": 6.4255453003221115e-06,
+      "loss": 0.2184,
+      "step": 324
+    },
+    {
+      "epoch": 1.9158050221565732,
+      "grad_norm": 0.2548243999481201,
+      "learning_rate": 6.364112992631537e-06,
+      "loss": 0.2299,
+      "step": 325
+    },
+    {
+      "epoch": 1.9217134416543575,
+      "grad_norm": 0.2533697187900543,
+      "learning_rate": 6.302838399503477e-06,
+      "loss": 0.2043,
+      "step": 326
+    },
+    {
+      "epoch": 1.9276218611521418,
+      "grad_norm": 0.2540424168109894,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 0.2039,
+      "step": 327
+    },
+    {
+      "epoch": 1.9335302806499262,
+      "grad_norm": 0.2535569965839386,
+      "learning_rate": 6.180772981657139e-06,
+      "loss": 0.2019,
+      "step": 328
+    },
+    {
+      "epoch": 1.9394387001477105,
+      "grad_norm": 0.29982754588127136,
+      "learning_rate": 6.119987451793711e-06,
+      "loss": 0.2228,
+      "step": 329
+    },
+    {
+      "epoch": 1.9453471196454948,
+      "grad_norm": 0.23110415041446686,
+      "learning_rate": 6.059370225972834e-06,
+      "loss": 0.2188,
+      "step": 330
+    },
+    {
+      "epoch": 1.951255539143279,
+      "grad_norm": 0.2608148753643036,
+      "learning_rate": 5.998923933599443e-06,
+      "loss": 0.2236,
+      "step": 331
+    },
+    {
+      "epoch": 1.9571639586410634,
+      "grad_norm": 0.26010897755622864,
+      "learning_rate": 5.938651196663865e-06,
+      "loss": 0.2032,
+      "step": 332
+    },
+    {
+      "epoch": 1.9630723781388477,
+      "grad_norm": 0.26297712326049805,
+      "learning_rate": 5.878554629628081e-06,
+      "loss": 0.2224,
+      "step": 333
+    },
+    {
+      "epoch": 1.9689807976366323,
+      "grad_norm": 0.2658803164958954,
+      "learning_rate": 5.818636839312309e-06,
+      "loss": 0.2153,
+      "step": 334
+    },
+    {
+      "epoch": 1.9748892171344166,
+      "grad_norm": 0.23885361850261688,
+      "learning_rate": 5.758900424781939e-06,
+      "loss": 0.2029,
+      "step": 335
+    },
+    {
+      "epoch": 1.9807976366322009,
+      "grad_norm": 0.2604767978191376,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 0.2059,
+      "step": 336
+    },
+    {
+      "epoch": 1.9867060561299852,
+      "grad_norm": 0.2535778284072876,
+      "learning_rate": 5.6399820798887266e-06,
+      "loss": 0.2204,
+      "step": 337
+    },
+    {
+      "epoch": 1.9926144756277697,
+      "grad_norm": 0.2699243128299713,
+      "learning_rate": 5.580805307869549e-06,
+      "loss": 0.2158,
+      "step": 338
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 507,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 169,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.797158580880671e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

3b-mb_qwen/checkpoint-338/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d657c9786dc6c8c08c64e914a96a01397e0a80c1d965337767408bc8f80e5cf
+size 10744

3b-mb_qwen/checkpoint-338/zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)

3b-mb_qwen/checkpoint-488/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

3b-mb_qwen/checkpoint-488/config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "Qwen/Qwen2.5-3B-Instruct",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 70,
+  "model_type": "qwen2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.48.1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151665
+}

3b-mb_qwen/checkpoint-488/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.48.1"
+}

3b-mb_qwen/checkpoint-488/latest

@@ -0,0 +1 @@
+global_step488

3b-mb_qwen/checkpoint-488/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b83e248771da77240f443146ced8bcec08670e1b062eafb60d717aad9c60d9d8
+size 4956450288

3b-mb_qwen/checkpoint-488/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cff6cf812ec94cd8ce4d0f6391ebeb2b6ba5f3610b37009cb666fa27b6651784
+size 1835586736

3b-mb_qwen/checkpoint-488/model.safetensors.index.json

@@ -0,0 +1,442 @@
+{
+  "metadata": {
+    "total_size": 6791987200
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00002-of-00002.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors"
+  }
+}

3b-mb_qwen/checkpoint-488/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dcb161b22b2558dbf7e3f8c871050cec383d11a40423fab11f18d5e630639bf
+size 14512

3b-mb_qwen/checkpoint-488/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d50af6aef769414a6f28fa1b1bc51ce707dc8ecd15474e03f99a2f10fde086be
+size 14512

3b-mb_qwen/checkpoint-488/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d8b2a59c30f5e09b1d7ce944fea889fdfc7000e147a68a8ad08ea9263213eb2
+size 1064

3b-mb_qwen/checkpoint-488/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

3b-mb_qwen/checkpoint-488/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
+size 11421896

3b-mb_qwen/checkpoint-488/tokenizer_config.json

@@ -0,0 +1,208 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

3b-mb_qwen/checkpoint-488/trainer_state.json

@@ -0,0 +1,3497 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.996935648621042,
+  "eval_steps": 82,
+  "global_step": 488,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0040858018386108275,
+      "grad_norm": 4.078765869140625,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 1.1276,
+      "step": 1
+    },
+    {
+      "epoch": 0.0040858018386108275,
+      "eval_loss": 1.1255302429199219,
+      "eval_runtime": 5.3213,
+      "eval_samples_per_second": 14.846,
+      "eval_steps_per_second": 1.879,
+      "step": 1
+    },
+    {
+      "epoch": 0.008171603677221655,
+      "grad_norm": 4.422896385192871,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.1678,
+      "step": 2
+    },
+    {
+      "epoch": 0.012257405515832482,
+      "grad_norm": 4.083443641662598,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0831,
+      "step": 3
+    },
+    {
+      "epoch": 0.01634320735444331,
+      "grad_norm": 4.063514709472656,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.0868,
+      "step": 4
+    },
+    {
+      "epoch": 0.020429009193054137,
+      "grad_norm": 3.104428291320801,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0167,
+      "step": 5
+    },
+    {
+      "epoch": 0.024514811031664963,
+      "grad_norm": 3.144237756729126,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1182,
+      "step": 6
+    },
+    {
+      "epoch": 0.028600612870275793,
+      "grad_norm": 1.837905764579773,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.8753,
+      "step": 7
+    },
+    {
+      "epoch": 0.03268641470888662,
+      "grad_norm": 1.8454015254974365,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.9588,
+      "step": 8
+    },
+    {
+      "epoch": 0.03677221654749745,
+      "grad_norm": 1.3577122688293457,
+      "learning_rate": 6e-06,
+      "loss": 0.7677,
+      "step": 9
+    },
+    {
+      "epoch": 0.04085801838610827,
+      "grad_norm": 1.20220947265625,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7597,
+      "step": 10
+    },
+    {
+      "epoch": 0.0449438202247191,
+      "grad_norm": 0.9984002709388733,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.7527,
+      "step": 11
+    },
+    {
+      "epoch": 0.049029622063329927,
+      "grad_norm": 1.228557825088501,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7182,
+      "step": 12
+    },
+    {
+      "epoch": 0.05311542390194075,
+      "grad_norm": 1.1346293687820435,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.7141,
+      "step": 13
+    },
+    {
+      "epoch": 0.05720122574055159,
+      "grad_norm": 0.9792781472206116,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.596,
+      "step": 14
+    },
+    {
+      "epoch": 0.06128702757916241,
+      "grad_norm": 0.6987927556037903,
+      "learning_rate": 1e-05,
+      "loss": 0.6034,
+      "step": 15
+    },
+    {
+      "epoch": 0.06537282941777324,
+      "grad_norm": 0.8394439220428467,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.5871,
+      "step": 16
+    },
+    {
+      "epoch": 0.06945863125638406,
+      "grad_norm": 0.7268638014793396,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 0.5364,
+      "step": 17
+    },
+    {
+      "epoch": 0.0735444330949949,
+      "grad_norm": 0.6233928203582764,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5608,
+      "step": 18
+    },
+    {
+      "epoch": 0.07763023493360573,
+      "grad_norm": 0.5607686042785645,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 0.5334,
+      "step": 19
+    },
+    {
+      "epoch": 0.08171603677221655,
+      "grad_norm": 0.599319338798523,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.512,
+      "step": 20
+    },
+    {
+      "epoch": 0.08580183861082738,
+      "grad_norm": 0.5736873149871826,
+      "learning_rate": 1.4e-05,
+      "loss": 0.5177,
+      "step": 21
+    },
+    {
+      "epoch": 0.0898876404494382,
+      "grad_norm": 0.5210515260696411,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.4653,
+      "step": 22
+    },
+    {
+      "epoch": 0.09397344228804903,
+      "grad_norm": 0.47595059871673584,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 0.4719,
+      "step": 23
+    },
+    {
+      "epoch": 0.09805924412665985,
+      "grad_norm": 0.5215305685997009,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4388,
+      "step": 24
+    },
+    {
+      "epoch": 0.10214504596527069,
+      "grad_norm": 0.48963531851768494,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4377,
+      "step": 25
+    },
+    {
+      "epoch": 0.1062308478038815,
+      "grad_norm": 0.48373010754585266,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.4269,
+      "step": 26
+    },
+    {
+      "epoch": 0.11031664964249234,
+      "grad_norm": 0.47493812441825867,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4871,
+      "step": 27
+    },
+    {
+      "epoch": 0.11440245148110317,
+      "grad_norm": 0.47279101610183716,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.4321,
+      "step": 28
+    },
+    {
+      "epoch": 0.118488253319714,
+      "grad_norm": 0.45037031173706055,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 0.4572,
+      "step": 29
+    },
+    {
+      "epoch": 0.12257405515832483,
+      "grad_norm": 0.3955671787261963,
+      "learning_rate": 2e-05,
+      "loss": 0.4001,
+      "step": 30
+    },
+    {
+      "epoch": 0.12665985699693566,
+      "grad_norm": 0.4108459949493408,
+      "learning_rate": 1.999989986294826e-05,
+      "loss": 0.4116,
+      "step": 31
+    },
+    {
+      "epoch": 0.13074565883554648,
+      "grad_norm": 0.4507107436656952,
+      "learning_rate": 1.9999599453798523e-05,
+      "loss": 0.4379,
+      "step": 32
+    },
+    {
+      "epoch": 0.1348314606741573,
+      "grad_norm": 0.4269455671310425,
+      "learning_rate": 1.999909877856721e-05,
+      "loss": 0.3826,
+      "step": 33
+    },
+    {
+      "epoch": 0.13891726251276812,
+      "grad_norm": 0.4127860963344574,
+      "learning_rate": 1.9998397847281548e-05,
+      "loss": 0.4081,
+      "step": 34
+    },
+    {
+      "epoch": 0.14300306435137897,
+      "grad_norm": 0.4478752613067627,
+      "learning_rate": 1.9997496673979375e-05,
+      "loss": 0.4071,
+      "step": 35
+    },
+    {
+      "epoch": 0.1470888661899898,
+      "grad_norm": 0.39848199486732483,
+      "learning_rate": 1.9996395276708856e-05,
+      "loss": 0.3752,
+      "step": 36
+    },
+    {
+      "epoch": 0.1511746680286006,
+      "grad_norm": 0.3737429678440094,
+      "learning_rate": 1.999509367752813e-05,
+      "loss": 0.389,
+      "step": 37
+    },
+    {
+      "epoch": 0.15526046986721145,
+      "grad_norm": 0.43358317017555237,
+      "learning_rate": 1.9993591902504854e-05,
+      "loss": 0.3933,
+      "step": 38
+    },
+    {
+      "epoch": 0.15934627170582227,
+      "grad_norm": 0.4123345613479614,
+      "learning_rate": 1.9991889981715696e-05,
+      "loss": 0.3905,
+      "step": 39
+    },
+    {
+      "epoch": 0.1634320735444331,
+      "grad_norm": 0.4238924980163574,
+      "learning_rate": 1.9989987949245725e-05,
+      "loss": 0.408,
+      "step": 40
+    },
+    {
+      "epoch": 0.1675178753830439,
+      "grad_norm": 0.4302418828010559,
+      "learning_rate": 1.9987885843187717e-05,
+      "loss": 0.4099,
+      "step": 41
+    },
+    {
+      "epoch": 0.17160367722165476,
+      "grad_norm": 0.40162044763565063,
+      "learning_rate": 1.9985583705641418e-05,
+      "loss": 0.3784,
+      "step": 42
+    },
+    {
+      "epoch": 0.17568947906026558,
+      "grad_norm": 0.3754842281341553,
+      "learning_rate": 1.9983081582712684e-05,
+      "loss": 0.3389,
+      "step": 43
+    },
+    {
+      "epoch": 0.1797752808988764,
+      "grad_norm": 0.37256139516830444,
+      "learning_rate": 1.998037952451255e-05,
+      "loss": 0.3869,
+      "step": 44
+    },
+    {
+      "epoch": 0.18386108273748722,
+      "grad_norm": 0.4196374714374542,
+      "learning_rate": 1.9977477585156252e-05,
+      "loss": 0.3892,
+      "step": 45
+    },
+    {
+      "epoch": 0.18794688457609807,
+      "grad_norm": 0.4057196378707886,
+      "learning_rate": 1.9974375822762117e-05,
+      "loss": 0.3932,
+      "step": 46
+    },
+    {
+      "epoch": 0.1920326864147089,
+      "grad_norm": 0.3770217001438141,
+      "learning_rate": 1.9971074299450414e-05,
+      "loss": 0.3953,
+      "step": 47
+    },
+    {
+      "epoch": 0.1961184882533197,
+      "grad_norm": 0.3990303575992584,
+      "learning_rate": 1.9967573081342103e-05,
+      "loss": 0.389,
+      "step": 48
+    },
+    {
+      "epoch": 0.20020429009193055,
+      "grad_norm": 0.39290961623191833,
+      "learning_rate": 1.9963872238557516e-05,
+      "loss": 0.3719,
+      "step": 49
+    },
+    {
+      "epoch": 0.20429009193054137,
+      "grad_norm": 0.3699814975261688,
+      "learning_rate": 1.9959971845214953e-05,
+      "loss": 0.3672,
+      "step": 50
+    },
+    {
+      "epoch": 0.2083758937691522,
+      "grad_norm": 0.3796740472316742,
+      "learning_rate": 1.9955871979429188e-05,
+      "loss": 0.3748,
+      "step": 51
+    },
+    {
+      "epoch": 0.212461695607763,
+      "grad_norm": 0.38025254011154175,
+      "learning_rate": 1.9951572723309918e-05,
+      "loss": 0.3726,
+      "step": 52
+    },
+    {
+      "epoch": 0.21654749744637386,
+      "grad_norm": 0.3801882863044739,
+      "learning_rate": 1.9947074162960113e-05,
+      "loss": 0.3937,
+      "step": 53
+    },
+    {
+      "epoch": 0.22063329928498468,
+      "grad_norm": 0.4046827554702759,
+      "learning_rate": 1.9942376388474282e-05,
+      "loss": 0.3692,
+      "step": 54
+    },
+    {
+      "epoch": 0.2247191011235955,
+      "grad_norm": 0.4178898334503174,
+      "learning_rate": 1.993747949393668e-05,
+      "loss": 0.3929,
+      "step": 55
+    },
+    {
+      "epoch": 0.22880490296220635,
+      "grad_norm": 0.4095941483974457,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.3754,
+      "step": 56
+    },
+    {
+      "epoch": 0.23289070480081717,
+      "grad_norm": 0.42728450894355774,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 0.3653,
+      "step": 57
+    },
+    {
+      "epoch": 0.236976506639428,
+      "grad_norm": 0.4169493019580841,
+      "learning_rate": 1.9921595090661872e-05,
+      "loss": 0.3544,
+      "step": 58
+    },
+    {
+      "epoch": 0.2410623084780388,
+      "grad_norm": 0.4049960970878601,
+      "learning_rate": 1.991590273648702e-05,
+      "loss": 0.3574,
+      "step": 59
+    },
+    {
+      "epoch": 0.24514811031664965,
+      "grad_norm": 0.4269113540649414,
+      "learning_rate": 1.9910011792459086e-05,
+      "loss": 0.3675,
+      "step": 60
+    },
+    {
+      "epoch": 0.24923391215526047,
+      "grad_norm": 0.3787218928337097,
+      "learning_rate": 1.9903922376558432e-05,
+      "loss": 0.3559,
+      "step": 61
+    },
+    {
+      "epoch": 0.2533197139938713,
+      "grad_norm": 0.3840825855731964,
+      "learning_rate": 1.989763461074029e-05,
+      "loss": 0.3706,
+      "step": 62
+    },
+    {
+      "epoch": 0.2574055158324821,
+      "grad_norm": 0.44640985131263733,
+      "learning_rate": 1.989114862093232e-05,
+      "loss": 0.372,
+      "step": 63
+    },
+    {
+      "epoch": 0.26149131767109296,
+      "grad_norm": 0.3927050530910492,
+      "learning_rate": 1.9884464537032103e-05,
+      "loss": 0.3542,
+      "step": 64
+    },
+    {
+      "epoch": 0.26557711950970375,
+      "grad_norm": 0.40718400478363037,
+      "learning_rate": 1.9877582492904533e-05,
+      "loss": 0.3588,
+      "step": 65
+    },
+    {
+      "epoch": 0.2696629213483146,
+      "grad_norm": 0.3699505031108856,
+      "learning_rate": 1.9870502626379127e-05,
+      "loss": 0.3455,
+      "step": 66
+    },
+    {
+      "epoch": 0.27374872318692545,
+      "grad_norm": 0.3665759265422821,
+      "learning_rate": 1.9863225079247286e-05,
+      "loss": 0.3229,
+      "step": 67
+    },
+    {
+      "epoch": 0.27783452502553624,
+      "grad_norm": 0.43499717116355896,
+      "learning_rate": 1.985574999725943e-05,
+      "loss": 0.3687,
+      "step": 68
+    },
+    {
+      "epoch": 0.2819203268641471,
+      "grad_norm": 0.3771592080593109,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.3586,
+      "step": 69
+    },
+    {
+      "epoch": 0.28600612870275793,
+      "grad_norm": 0.3490198850631714,
+      "learning_rate": 1.9840207831494903e-05,
+      "loss": 0.3124,
+      "step": 70
+    },
+    {
+      "epoch": 0.2900919305413687,
+      "grad_norm": 0.4739793837070465,
+      "learning_rate": 1.983214105898757e-05,
+      "loss": 0.3464,
+      "step": 71
+    },
+    {
+      "epoch": 0.2941777323799796,
+      "grad_norm": 0.4140881896018982,
+      "learning_rate": 1.9823877374156647e-05,
+      "loss": 0.3341,
+      "step": 72
+    },
+    {
+      "epoch": 0.2982635342185904,
+      "grad_norm": 0.40869495272636414,
+      "learning_rate": 1.9815416942502346e-05,
+      "loss": 0.3294,
+      "step": 73
+    },
+    {
+      "epoch": 0.3023493360572012,
+      "grad_norm": 0.44560012221336365,
+      "learning_rate": 1.98067599334652e-05,
+      "loss": 0.4026,
+      "step": 74
+    },
+    {
+      "epoch": 0.30643513789581206,
+      "grad_norm": 0.4086117744445801,
+      "learning_rate": 1.979790652042268e-05,
+      "loss": 0.3767,
+      "step": 75
+    },
+    {
+      "epoch": 0.3105209397344229,
+      "grad_norm": 0.4267195761203766,
+      "learning_rate": 1.978885688068572e-05,
+      "loss": 0.339,
+      "step": 76
+    },
+    {
+      "epoch": 0.3146067415730337,
+      "grad_norm": 0.3377031087875366,
+      "learning_rate": 1.9779611195495177e-05,
+      "loss": 0.3315,
+      "step": 77
+    },
+    {
+      "epoch": 0.31869254341164455,
+      "grad_norm": 0.4165203273296356,
+      "learning_rate": 1.977016965001817e-05,
+      "loss": 0.3607,
+      "step": 78
+    },
+    {
+      "epoch": 0.32277834525025534,
+      "grad_norm": 0.4003439247608185,
+      "learning_rate": 1.976053243334442e-05,
+      "loss": 0.3519,
+      "step": 79
+    },
+    {
+      "epoch": 0.3268641470888662,
+      "grad_norm": 0.39171072840690613,
+      "learning_rate": 1.9750699738482403e-05,
+      "loss": 0.3471,
+      "step": 80
+    },
+    {
+      "epoch": 0.33094994892747703,
+      "grad_norm": 0.3989802598953247,
+      "learning_rate": 1.9740671762355548e-05,
+      "loss": 0.3671,
+      "step": 81
+    },
+    {
+      "epoch": 0.3350357507660878,
+      "grad_norm": 0.4387929141521454,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.3639,
+      "step": 82
+    },
+    {
+      "epoch": 0.3350357507660878,
+      "eval_loss": 0.34231218695640564,
+      "eval_runtime": 5.7708,
+      "eval_samples_per_second": 13.69,
+      "eval_steps_per_second": 1.733,
+      "step": 82
+    },
+    {
+      "epoch": 0.3391215526046987,
+      "grad_norm": 0.40271735191345215,
+      "learning_rate": 1.972003077355183e-05,
+      "loss": 0.3691,
+      "step": 83
+    },
+    {
+      "epoch": 0.3432073544433095,
+      "grad_norm": 0.3957328200340271,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 0.3601,
+      "step": 84
+    },
+    {
+      "epoch": 0.3472931562819203,
+      "grad_norm": 0.36126795411109924,
+      "learning_rate": 1.9698611120467196e-05,
+      "loss": 0.3355,
+      "step": 85
+    },
+    {
+      "epoch": 0.35137895812053116,
+      "grad_norm": 0.4093964695930481,
+      "learning_rate": 1.9687609828609156e-05,
+      "loss": 0.3341,
+      "step": 86
+    },
+    {
+      "epoch": 0.355464759959142,
+      "grad_norm": 0.38025736808776855,
+      "learning_rate": 1.9676414519013782e-05,
+      "loss": 0.3636,
+      "step": 87
+    },
+    {
+      "epoch": 0.3595505617977528,
+      "grad_norm": 0.44078460335731506,
+      "learning_rate": 1.966502541589414e-05,
+      "loss": 0.317,
+      "step": 88
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.352769136428833,
+      "learning_rate": 1.965344274734447e-05,
+      "loss": 0.3545,
+      "step": 89
+    },
+    {
+      "epoch": 0.36772216547497444,
+      "grad_norm": 0.3688812255859375,
+      "learning_rate": 1.9641666745335626e-05,
+      "loss": 0.3249,
+      "step": 90
+    },
+    {
+      "epoch": 0.3718079673135853,
+      "grad_norm": 0.40612637996673584,
+      "learning_rate": 1.9629697645710432e-05,
+      "loss": 0.3352,
+      "step": 91
+    },
+    {
+      "epoch": 0.37589376915219613,
+      "grad_norm": 0.35389018058776855,
+      "learning_rate": 1.961753568817896e-05,
+      "loss": 0.3203,
+      "step": 92
+    },
+    {
+      "epoch": 0.3799795709908069,
+      "grad_norm": 0.38835206627845764,
+      "learning_rate": 1.9605181116313725e-05,
+      "loss": 0.3248,
+      "step": 93
+    },
+    {
+      "epoch": 0.3840653728294178,
+      "grad_norm": 0.38264191150665283,
+      "learning_rate": 1.9592634177544803e-05,
+      "loss": 0.3468,
+      "step": 94
+    },
+    {
+      "epoch": 0.3881511746680286,
+      "grad_norm": 0.37082910537719727,
+      "learning_rate": 1.957989512315489e-05,
+      "loss": 0.3602,
+      "step": 95
+    },
+    {
+      "epoch": 0.3922369765066394,
+      "grad_norm": 0.35148903727531433,
+      "learning_rate": 1.9566964208274254e-05,
+      "loss": 0.3505,
+      "step": 96
+    },
+    {
+      "epoch": 0.39632277834525026,
+      "grad_norm": 0.350971519947052,
+      "learning_rate": 1.9553841691875632e-05,
+      "loss": 0.3208,
+      "step": 97
+    },
+    {
+      "epoch": 0.4004085801838611,
+      "grad_norm": 0.36287498474121094,
+      "learning_rate": 1.9540527836769047e-05,
+      "loss": 0.3451,
+      "step": 98
+    },
+    {
+      "epoch": 0.4044943820224719,
+      "grad_norm": 0.4268699288368225,
+      "learning_rate": 1.9527022909596537e-05,
+      "loss": 0.3623,
+      "step": 99
+    },
+    {
+      "epoch": 0.40858018386108275,
+      "grad_norm": 0.3622118830680847,
+      "learning_rate": 1.951332718082682e-05,
+      "loss": 0.3219,
+      "step": 100
+    },
+    {
+      "epoch": 0.41266598569969354,
+      "grad_norm": 0.3629276752471924,
+      "learning_rate": 1.9499440924749878e-05,
+      "loss": 0.3565,
+      "step": 101
+    },
+    {
+      "epoch": 0.4167517875383044,
+      "grad_norm": 0.3923068940639496,
+      "learning_rate": 1.9485364419471454e-05,
+      "loss": 0.358,
+      "step": 102
+    },
+    {
+      "epoch": 0.42083758937691523,
+      "grad_norm": 0.378465861082077,
+      "learning_rate": 1.9471097946907506e-05,
+      "loss": 0.3443,
+      "step": 103
+    },
+    {
+      "epoch": 0.424923391215526,
+      "grad_norm": 0.41037172079086304,
+      "learning_rate": 1.9456641792778527e-05,
+      "loss": 0.3531,
+      "step": 104
+    },
+    {
+      "epoch": 0.4290091930541369,
+      "grad_norm": 0.38496220111846924,
+      "learning_rate": 1.9441996246603848e-05,
+      "loss": 0.3305,
+      "step": 105
+    },
+    {
+      "epoch": 0.4330949948927477,
+      "grad_norm": 0.37626922130584717,
+      "learning_rate": 1.9427161601695833e-05,
+      "loss": 0.3556,
+      "step": 106
+    },
+    {
+      "epoch": 0.4371807967313585,
+      "grad_norm": 0.3895152509212494,
+      "learning_rate": 1.9412138155154e-05,
+      "loss": 0.3363,
+      "step": 107
+    },
+    {
+      "epoch": 0.44126659856996936,
+      "grad_norm": 0.3873409032821655,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.3322,
+      "step": 108
+    },
+    {
+      "epoch": 0.4453524004085802,
+      "grad_norm": 0.3699762225151062,
+      "learning_rate": 1.9381526064466995e-05,
+      "loss": 0.3482,
+      "step": 109
+    },
+    {
+      "epoch": 0.449438202247191,
+      "grad_norm": 0.3172101080417633,
+      "learning_rate": 1.9365938033402715e-05,
+      "loss": 0.3266,
+      "step": 110
+    },
+    {
+      "epoch": 0.45352400408580185,
+      "grad_norm": 0.414472371339798,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 0.3372,
+      "step": 111
+    },
+    {
+      "epoch": 0.4576098059244127,
+      "grad_norm": 0.3786821663379669,
+      "learning_rate": 1.933419956076584e-05,
+      "loss": 0.3311,
+      "step": 112
+    },
+    {
+      "epoch": 0.4616956077630235,
+      "grad_norm": 0.3550058603286743,
+      "learning_rate": 1.9318049754832656e-05,
+      "loss": 0.3263,
+      "step": 113
+    },
+    {
+      "epoch": 0.46578140960163433,
+      "grad_norm": 0.3771400451660156,
+      "learning_rate": 1.9301713332493386e-05,
+      "loss": 0.3468,
+      "step": 114
+    },
+    {
+      "epoch": 0.4698672114402451,
+      "grad_norm": 0.3498581349849701,
+      "learning_rate": 1.9285190620924267e-05,
+      "loss": 0.3035,
+      "step": 115
+    },
+    {
+      "epoch": 0.473953013278856,
+      "grad_norm": 0.37467050552368164,
+      "learning_rate": 1.926848195103242e-05,
+      "loss": 0.3208,
+      "step": 116
+    },
+    {
+      "epoch": 0.4780388151174668,
+      "grad_norm": 0.37225377559661865,
+      "learning_rate": 1.925158765744924e-05,
+      "loss": 0.3073,
+      "step": 117
+    },
+    {
+      "epoch": 0.4821246169560776,
+      "grad_norm": 0.3326958119869232,
+      "learning_rate": 1.923450807852367e-05,
+      "loss": 0.3148,
+      "step": 118
+    },
+    {
+      "epoch": 0.48621041879468846,
+      "grad_norm": 0.3672332465648651,
+      "learning_rate": 1.9217243556315445e-05,
+      "loss": 0.3391,
+      "step": 119
+    },
+    {
+      "epoch": 0.4902962206332993,
+      "grad_norm": 0.3444240093231201,
+      "learning_rate": 1.9199794436588244e-05,
+      "loss": 0.3113,
+      "step": 120
+    },
+    {
+      "epoch": 0.4943820224719101,
+      "grad_norm": 0.3641497790813446,
+      "learning_rate": 1.9182161068802742e-05,
+      "loss": 0.3358,
+      "step": 121
+    },
+    {
+      "epoch": 0.49846782431052095,
+      "grad_norm": 0.40089380741119385,
+      "learning_rate": 1.916434380610963e-05,
+      "loss": 0.3312,
+      "step": 122
+    },
+    {
+      "epoch": 0.5025536261491318,
+      "grad_norm": 0.38641074299812317,
+      "learning_rate": 1.9146343005342546e-05,
+      "loss": 0.3322,
+      "step": 123
+    },
+    {
+      "epoch": 0.5066394279877426,
+      "grad_norm": 0.34743842482566833,
+      "learning_rate": 1.912815902701091e-05,
+      "loss": 0.3061,
+      "step": 124
+    },
+    {
+      "epoch": 0.5107252298263534,
+      "grad_norm": 0.3766823410987854,
+      "learning_rate": 1.9109792235292715e-05,
+      "loss": 0.3364,
+      "step": 125
+    },
+    {
+      "epoch": 0.5148110316649642,
+      "grad_norm": 0.33645695447921753,
+      "learning_rate": 1.909124299802724e-05,
+      "loss": 0.3307,
+      "step": 126
+    },
+    {
+      "epoch": 0.5188968335035751,
+      "grad_norm": 0.3658338189125061,
+      "learning_rate": 1.9072511686707663e-05,
+      "loss": 0.3223,
+      "step": 127
+    },
+    {
+      "epoch": 0.5229826353421859,
+      "grad_norm": 0.3340202867984772,
+      "learning_rate": 1.9053598676473656e-05,
+      "loss": 0.3317,
+      "step": 128
+    },
+    {
+      "epoch": 0.5270684371807968,
+      "grad_norm": 0.3270389437675476,
+      "learning_rate": 1.9034504346103825e-05,
+      "loss": 0.3353,
+      "step": 129
+    },
+    {
+      "epoch": 0.5311542390194075,
+      "grad_norm": 0.30215510725975037,
+      "learning_rate": 1.9015229078008163e-05,
+      "loss": 0.3023,
+      "step": 130
+    },
+    {
+      "epoch": 0.5352400408580184,
+      "grad_norm": 0.3734086751937866,
+      "learning_rate": 1.8995773258220374e-05,
+      "loss": 0.3189,
+      "step": 131
+    },
+    {
+      "epoch": 0.5393258426966292,
+      "grad_norm": 0.3713105320930481,
+      "learning_rate": 1.8976137276390145e-05,
+      "loss": 0.2995,
+      "step": 132
+    },
+    {
+      "epoch": 0.54341164453524,
+      "grad_norm": 0.3478635847568512,
+      "learning_rate": 1.8956321525775337e-05,
+      "loss": 0.3249,
+      "step": 133
+    },
+    {
+      "epoch": 0.5474974463738509,
+      "grad_norm": 0.3437729775905609,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.3306,
+      "step": 134
+    },
+    {
+      "epoch": 0.5515832482124617,
+      "grad_norm": 0.3874213397502899,
+      "learning_rate": 1.891615230921703e-05,
+      "loss": 0.3558,
+      "step": 135
+    },
+    {
+      "epoch": 0.5556690500510725,
+      "grad_norm": 0.7216417193412781,
+      "learning_rate": 1.8895799647758912e-05,
+      "loss": 0.3127,
+      "step": 136
+    },
+    {
+      "epoch": 0.5597548518896833,
+      "grad_norm": 0.3799072504043579,
+      "learning_rate": 1.8875268826470875e-05,
+      "loss": 0.3319,
+      "step": 137
+    },
+    {
+      "epoch": 0.5638406537282942,
+      "grad_norm": 0.36491042375564575,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 0.3296,
+      "step": 138
+    },
+    {
+      "epoch": 0.567926455566905,
+      "grad_norm": 0.3649715483188629,
+      "learning_rate": 1.8833674352681613e-05,
+      "loss": 0.3145,
+      "step": 139
+    },
+    {
+      "epoch": 0.5720122574055159,
+      "grad_norm": 0.37904301285743713,
+      "learning_rate": 1.881261153320999e-05,
+      "loss": 0.341,
+      "step": 140
+    },
+    {
+      "epoch": 0.5760980592441267,
+      "grad_norm": 0.3647158145904541,
+      "learning_rate": 1.879137221995095e-05,
+      "loss": 0.3412,
+      "step": 141
+    },
+    {
+      "epoch": 0.5801838610827375,
+      "grad_norm": 0.397286593914032,
+      "learning_rate": 1.8769956838272937e-05,
+      "loss": 0.3156,
+      "step": 142
+    },
+    {
+      "epoch": 0.5842696629213483,
+      "grad_norm": 0.3785179853439331,
+      "learning_rate": 1.8748365817070586e-05,
+      "loss": 0.3037,
+      "step": 143
+    },
+    {
+      "epoch": 0.5883554647599591,
+      "grad_norm": 0.40976065397262573,
+      "learning_rate": 1.8726599588756144e-05,
+      "loss": 0.3194,
+      "step": 144
+    },
+    {
+      "epoch": 0.59244126659857,
+      "grad_norm": 0.39624109864234924,
+      "learning_rate": 1.8704658589250795e-05,
+      "loss": 0.3442,
+      "step": 145
+    },
+    {
+      "epoch": 0.5965270684371808,
+      "grad_norm": 0.3463563919067383,
+      "learning_rate": 1.868254325797594e-05,
+      "loss": 0.3178,
+      "step": 146
+    },
+    {
+      "epoch": 0.6006128702757916,
+      "grad_norm": 0.4022400677204132,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.3175,
+      "step": 147
+    },
+    {
+      "epoch": 0.6046986721144024,
+      "grad_norm": 0.39606690406799316,
+      "learning_rate": 1.8637791375251505e-05,
+      "loss": 0.314,
+      "step": 148
+    },
+    {
+      "epoch": 0.6087844739530133,
+      "grad_norm": 0.3561919331550598,
+      "learning_rate": 1.8615155720066247e-05,
+      "loss": 0.3324,
+      "step": 149
+    },
+    {
+      "epoch": 0.6128702757916241,
+      "grad_norm": 0.39145007729530334,
+      "learning_rate": 1.859234752562217e-05,
+      "loss": 0.3408,
+      "step": 150
+    },
+    {
+      "epoch": 0.616956077630235,
+      "grad_norm": 0.34467753767967224,
+      "learning_rate": 1.8569367248708343e-05,
+      "loss": 0.3181,
+      "step": 151
+    },
+    {
+      "epoch": 0.6210418794688458,
+      "grad_norm": 0.326642781496048,
+      "learning_rate": 1.8546215349560204e-05,
+      "loss": 0.3352,
+      "step": 152
+    },
+    {
+      "epoch": 0.6251276813074566,
+      "grad_norm": 0.3399414122104645,
+      "learning_rate": 1.8522892291850335e-05,
+      "loss": 0.324,
+      "step": 153
+    },
+    {
+      "epoch": 0.6292134831460674,
+      "grad_norm": 0.37066522240638733,
+      "learning_rate": 1.849939854267919e-05,
+      "loss": 0.3463,
+      "step": 154
+    },
+    {
+      "epoch": 0.6332992849846782,
+      "grad_norm": 0.34315940737724304,
+      "learning_rate": 1.847573457256571e-05,
+      "loss": 0.3385,
+      "step": 155
+    },
+    {
+      "epoch": 0.6373850868232891,
+      "grad_norm": 0.3131471574306488,
+      "learning_rate": 1.845190085543795e-05,
+      "loss": 0.3068,
+      "step": 156
+    },
+    {
+      "epoch": 0.6414708886618999,
+      "grad_norm": 0.31940385699272156,
+      "learning_rate": 1.8427897868623535e-05,
+      "loss": 0.3227,
+      "step": 157
+    },
+    {
+      "epoch": 0.6455566905005107,
+      "grad_norm": 0.3455657362937927,
+      "learning_rate": 1.840372609284013e-05,
+      "loss": 0.2897,
+      "step": 158
+    },
+    {
+      "epoch": 0.6496424923391215,
+      "grad_norm": 0.3038015067577362,
+      "learning_rate": 1.8379386012185813e-05,
+      "loss": 0.3217,
+      "step": 159
+    },
+    {
+      "epoch": 0.6537282941777324,
+      "grad_norm": 0.3192991614341736,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.3274,
+      "step": 160
+    },
+    {
+      "epoch": 0.6578140960163432,
+      "grad_norm": 0.3324792683124542,
+      "learning_rate": 1.8330202889500518e-05,
+      "loss": 0.3077,
+      "step": 161
+    },
+    {
+      "epoch": 0.6618998978549541,
+      "grad_norm": 0.3139905333518982,
+      "learning_rate": 1.8305360832480118e-05,
+      "loss": 0.2977,
+      "step": 162
+    },
+    {
+      "epoch": 0.6659856996935649,
+      "grad_norm": 0.3571060001850128,
+      "learning_rate": 1.8280352440590236e-05,
+      "loss": 0.345,
+      "step": 163
+    },
+    {
+      "epoch": 0.6700715015321757,
+      "grad_norm": 0.47001609206199646,
+      "learning_rate": 1.82551782146842e-05,
+      "loss": 0.303,
+      "step": 164
+    },
+    {
+      "epoch": 0.6700715015321757,
+      "eval_loss": 0.31244462728500366,
+      "eval_runtime": 5.7126,
+      "eval_samples_per_second": 13.829,
+      "eval_steps_per_second": 1.751,
+      "step": 164
+    },
+    {
+      "epoch": 0.6741573033707865,
+      "grad_norm": 0.3575729429721832,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 0.3568,
+      "step": 165
+    },
+    {
+      "epoch": 0.6782431052093973,
+      "grad_norm": 0.3167235553264618,
+      "learning_rate": 1.8204334280833005e-05,
+      "loss": 0.2982,
+      "step": 166
+    },
+    {
+      "epoch": 0.6823289070480082,
+      "grad_norm": 0.4739898145198822,
+      "learning_rate": 1.817866559116017e-05,
+      "loss": 0.3613,
+      "step": 167
+    },
+    {
+      "epoch": 0.686414708886619,
+      "grad_norm": 0.35848763585090637,
+      "learning_rate": 1.8152833103995443e-05,
+      "loss": 0.3438,
+      "step": 168
+    },
+    {
+      "epoch": 0.6905005107252298,
+      "grad_norm": 0.3687516450881958,
+      "learning_rate": 1.8126837336696645e-05,
+      "loss": 0.3328,
+      "step": 169
+    },
+    {
+      "epoch": 0.6945863125638406,
+      "grad_norm": 0.35989782214164734,
+      "learning_rate": 1.8100678809891668e-05,
+      "loss": 0.3279,
+      "step": 170
+    },
+    {
+      "epoch": 0.6986721144024515,
+      "grad_norm": 0.3408145606517792,
+      "learning_rate": 1.807435804746807e-05,
+      "loss": 0.3076,
+      "step": 171
+    },
+    {
+      "epoch": 0.7027579162410623,
+      "grad_norm": 0.3220236301422119,
+      "learning_rate": 1.8047875576562556e-05,
+      "loss": 0.3366,
+      "step": 172
+    },
+    {
+      "epoch": 0.7068437180796732,
+      "grad_norm": 0.3259022533893585,
+      "learning_rate": 1.802123192755044e-05,
+      "loss": 0.3052,
+      "step": 173
+    },
+    {
+      "epoch": 0.710929519918284,
+      "grad_norm": 0.33262839913368225,
+      "learning_rate": 1.7994427634035016e-05,
+      "loss": 0.3016,
+      "step": 174
+    },
+    {
+      "epoch": 0.7150153217568948,
+      "grad_norm": 0.33660101890563965,
+      "learning_rate": 1.796746323283686e-05,
+      "loss": 0.3356,
+      "step": 175
+    },
+    {
+      "epoch": 0.7191011235955056,
+      "grad_norm": 0.32823285460472107,
+      "learning_rate": 1.7940339263983112e-05,
+      "loss": 0.3216,
+      "step": 176
+    },
+    {
+      "epoch": 0.7231869254341164,
+      "grad_norm": 0.34937459230422974,
+      "learning_rate": 1.791305627069662e-05,
+      "loss": 0.3351,
+      "step": 177
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.34692323207855225,
+      "learning_rate": 1.7885614799385086e-05,
+      "loss": 0.3142,
+      "step": 178
+    },
+    {
+      "epoch": 0.7313585291113381,
+      "grad_norm": 0.3434458076953888,
+      "learning_rate": 1.785801539963012e-05,
+      "loss": 0.325,
+      "step": 179
+    },
+    {
+      "epoch": 0.7354443309499489,
+      "grad_norm": 0.30833402276039124,
+      "learning_rate": 1.7830258624176224e-05,
+      "loss": 0.2976,
+      "step": 180
+    },
+    {
+      "epoch": 0.7395301327885597,
+      "grad_norm": 0.318734735250473,
+      "learning_rate": 1.7802345028919728e-05,
+      "loss": 0.3127,
+      "step": 181
+    },
+    {
+      "epoch": 0.7436159346271706,
+      "grad_norm": 0.34027615189552307,
+      "learning_rate": 1.777427517289766e-05,
+      "loss": 0.316,
+      "step": 182
+    },
+    {
+      "epoch": 0.7477017364657814,
+      "grad_norm": 0.35010436177253723,
+      "learning_rate": 1.7746049618276545e-05,
+      "loss": 0.3036,
+      "step": 183
+    },
+    {
+      "epoch": 0.7517875383043923,
+      "grad_norm": 0.3359059691429138,
+      "learning_rate": 1.7717668930341152e-05,
+      "loss": 0.3076,
+      "step": 184
+    },
+    {
+      "epoch": 0.7558733401430031,
+      "grad_norm": 0.3144778311252594,
+      "learning_rate": 1.768913367748316e-05,
+      "loss": 0.3063,
+      "step": 185
+    },
+    {
+      "epoch": 0.7599591419816139,
+      "grad_norm": 0.34391850233078003,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.3295,
+      "step": 186
+    },
+    {
+      "epoch": 0.7640449438202247,
+      "grad_norm": 0.33601975440979004,
+      "learning_rate": 1.7631601766032337e-05,
+      "loss": 0.3086,
+      "step": 187
+    },
+    {
+      "epoch": 0.7681307456588355,
+      "grad_norm": 0.3327379524707794,
+      "learning_rate": 1.7602606259654704e-05,
+      "loss": 0.3164,
+      "step": 188
+    },
+    {
+      "epoch": 0.7722165474974464,
+      "grad_norm": 0.30492648482322693,
+      "learning_rate": 1.7573458492761802e-05,
+      "loss": 0.311,
+      "step": 189
+    },
+    {
+      "epoch": 0.7763023493360572,
+      "grad_norm": 0.33100536465644836,
+      "learning_rate": 1.7544159049107902e-05,
+      "loss": 0.3025,
+      "step": 190
+    },
+    {
+      "epoch": 0.780388151174668,
+      "grad_norm": 0.3391270637512207,
+      "learning_rate": 1.7514708515485002e-05,
+      "loss": 0.3348,
+      "step": 191
+    },
+    {
+      "epoch": 0.7844739530132788,
+      "grad_norm": 0.3125060796737671,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 0.31,
+      "step": 192
+    },
+    {
+      "epoch": 0.7885597548518897,
+      "grad_norm": 0.3132609724998474,
+      "learning_rate": 1.7455356540617988e-05,
+      "loss": 0.3204,
+      "step": 193
+    },
+    {
+      "epoch": 0.7926455566905005,
+      "grad_norm": 0.31536075472831726,
+      "learning_rate": 1.7425456288040236e-05,
+      "loss": 0.3202,
+      "step": 194
+    },
+    {
+      "epoch": 0.7967313585291114,
+      "grad_norm": 0.3190646171569824,
+      "learning_rate": 1.7395407322802374e-05,
+      "loss": 0.3075,
+      "step": 195
+    },
+    {
+      "epoch": 0.8008171603677222,
+      "grad_norm": 0.333128422498703,
+      "learning_rate": 1.736521024670737e-05,
+      "loss": 0.3088,
+      "step": 196
+    },
+    {
+      "epoch": 0.804902962206333,
+      "grad_norm": 0.3181339204311371,
+      "learning_rate": 1.733486566452446e-05,
+      "loss": 0.31,
+      "step": 197
+    },
+    {
+      "epoch": 0.8089887640449438,
+      "grad_norm": 0.3423081338405609,
+      "learning_rate": 1.7304374183977032e-05,
+      "loss": 0.3158,
+      "step": 198
+    },
+    {
+      "epoch": 0.8130745658835546,
+      "grad_norm": 0.32152727246284485,
+      "learning_rate": 1.7273736415730488e-05,
+      "loss": 0.3125,
+      "step": 199
+    },
+    {
+      "epoch": 0.8171603677221655,
+      "grad_norm": 0.3334408402442932,
+      "learning_rate": 1.7242952973379983e-05,
+      "loss": 0.3015,
+      "step": 200
+    },
+    {
+      "epoch": 0.8212461695607763,
+      "grad_norm": 0.32138195633888245,
+      "learning_rate": 1.7212024473438145e-05,
+      "loss": 0.3069,
+      "step": 201
+    },
+    {
+      "epoch": 0.8253319713993871,
+      "grad_norm": 0.3368797302246094,
+      "learning_rate": 1.7180951535322742e-05,
+      "loss": 0.3141,
+      "step": 202
+    },
+    {
+      "epoch": 0.8294177732379979,
+      "grad_norm": 0.3377224802970886,
+      "learning_rate": 1.7149734781344247e-05,
+      "loss": 0.3102,
+      "step": 203
+    },
+    {
+      "epoch": 0.8335035750766088,
+      "grad_norm": 0.3103748857975006,
+      "learning_rate": 1.7118374836693407e-05,
+      "loss": 0.3417,
+      "step": 204
+    },
+    {
+      "epoch": 0.8375893769152196,
+      "grad_norm": 0.299224853515625,
+      "learning_rate": 1.7086872329428702e-05,
+      "loss": 0.3059,
+      "step": 205
+    },
+    {
+      "epoch": 0.8416751787538305,
+      "grad_norm": 0.3273285925388336,
+      "learning_rate": 1.705522789046377e-05,
+      "loss": 0.298,
+      "step": 206
+    },
+    {
+      "epoch": 0.8457609805924413,
+      "grad_norm": 0.34890103340148926,
+      "learning_rate": 1.7023442153554776e-05,
+      "loss": 0.327,
+      "step": 207
+    },
+    {
+      "epoch": 0.849846782431052,
+      "grad_norm": 0.3465830087661743,
+      "learning_rate": 1.6991515755287715e-05,
+      "loss": 0.3457,
+      "step": 208
+    },
+    {
+      "epoch": 0.8539325842696629,
+      "grad_norm": 0.32012030482292175,
+      "learning_rate": 1.695944933506567e-05,
+      "loss": 0.2886,
+      "step": 209
+    },
+    {
+      "epoch": 0.8580183861082737,
+      "grad_norm": 0.31332501769065857,
+      "learning_rate": 1.6927243535095995e-05,
+      "loss": 0.3145,
+      "step": 210
+    },
+    {
+      "epoch": 0.8621041879468846,
+      "grad_norm": 0.31248798966407776,
+      "learning_rate": 1.6894899000377462e-05,
+      "loss": 0.3065,
+      "step": 211
+    },
+    {
+      "epoch": 0.8661899897854954,
+      "grad_norm": 0.32849758863449097,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.2961,
+      "step": 212
+    },
+    {
+      "epoch": 0.8702757916241062,
+      "grad_norm": 0.33984649181365967,
+      "learning_rate": 1.6829796320568416e-05,
+      "loss": 0.2948,
+      "step": 213
+    },
+    {
+      "epoch": 0.874361593462717,
+      "grad_norm": 0.33225297927856445,
+      "learning_rate": 1.6797039479315994e-05,
+      "loss": 0.3039,
+      "step": 214
+    },
+    {
+      "epoch": 0.8784473953013279,
+      "grad_norm": 0.32829374074935913,
+      "learning_rate": 1.6764146510964762e-05,
+      "loss": 0.3269,
+      "step": 215
+    },
+    {
+      "epoch": 0.8825331971399387,
+      "grad_norm": 0.35096362233161926,
+      "learning_rate": 1.67311180742757e-05,
+      "loss": 0.3142,
+      "step": 216
+    },
+    {
+      "epoch": 0.8866189989785496,
+      "grad_norm": 0.32767823338508606,
+      "learning_rate": 1.669795483072287e-05,
+      "loss": 0.3038,
+      "step": 217
+    },
+    {
+      "epoch": 0.8907048008171604,
+      "grad_norm": 0.3024498522281647,
+      "learning_rate": 1.6664657444480145e-05,
+      "loss": 0.2988,
+      "step": 218
+    },
+    {
+      "epoch": 0.8947906026557712,
+      "grad_norm": 0.3320610821247101,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 0.2902,
+      "step": 219
+    },
+    {
+      "epoch": 0.898876404494382,
+      "grad_norm": 0.3502938449382782,
+      "learning_rate": 1.6597662914039885e-05,
+      "loss": 0.3039,
+      "step": 220
+    },
+    {
+      "epoch": 0.9029622063329928,
+      "grad_norm": 0.3509136140346527,
+      "learning_rate": 1.65639671115693e-05,
+      "loss": 0.3114,
+      "step": 221
+    },
+    {
+      "epoch": 0.9070480081716037,
+      "grad_norm": 0.3239796459674835,
+      "learning_rate": 1.653013984983585e-05,
+      "loss": 0.297,
+      "step": 222
+    },
+    {
+      "epoch": 0.9111338100102145,
+      "grad_norm": 0.340690940618515,
+      "learning_rate": 1.6496181806312005e-05,
+      "loss": 0.3071,
+      "step": 223
+    },
+    {
+      "epoch": 0.9152196118488254,
+      "grad_norm": 0.342742919921875,
+      "learning_rate": 1.6462093661089432e-05,
+      "loss": 0.2908,
+      "step": 224
+    },
+    {
+      "epoch": 0.9193054136874361,
+      "grad_norm": 0.33802884817123413,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.3135,
+      "step": 225
+    },
+    {
+      "epoch": 0.923391215526047,
+      "grad_norm": 0.32046419382095337,
+      "learning_rate": 1.6393529798929103e-05,
+      "loss": 0.32,
+      "step": 226
+    },
+    {
+      "epoch": 0.9274770173646578,
+      "grad_norm": 0.32189199328422546,
+      "learning_rate": 1.635905545514795e-05,
+      "loss": 0.3277,
+      "step": 227
+    },
+    {
+      "epoch": 0.9315628192032687,
+      "grad_norm": 0.3596160113811493,
+      "learning_rate": 1.6324453755953772e-05,
+      "loss": 0.3115,
+      "step": 228
+    },
+    {
+      "epoch": 0.9356486210418795,
+      "grad_norm": 0.36959317326545715,
+      "learning_rate": 1.6289725394328998e-05,
+      "loss": 0.3255,
+      "step": 229
+    },
+    {
+      "epoch": 0.9397344228804902,
+      "grad_norm": 0.3319382071495056,
+      "learning_rate": 1.6254871065792776e-05,
+      "loss": 0.3413,
+      "step": 230
+    },
+    {
+      "epoch": 0.9438202247191011,
+      "grad_norm": 0.32328999042510986,
+      "learning_rate": 1.621989146838704e-05,
+      "loss": 0.3164,
+      "step": 231
+    },
+    {
+      "epoch": 0.947906026557712,
+      "grad_norm": 0.35541966557502747,
+      "learning_rate": 1.618478730266255e-05,
+      "loss": 0.3006,
+      "step": 232
+    },
+    {
+      "epoch": 0.9519918283963228,
+      "grad_norm": 0.31904998421669006,
+      "learning_rate": 1.6149559271664835e-05,
+      "loss": 0.2973,
+      "step": 233
+    },
+    {
+      "epoch": 0.9560776302349336,
+      "grad_norm": 0.30805569887161255,
+      "learning_rate": 1.6114208080920125e-05,
+      "loss": 0.2857,
+      "step": 234
+    },
+    {
+      "epoch": 0.9601634320735445,
+      "grad_norm": 0.3331436812877655,
+      "learning_rate": 1.607873443842122e-05,
+      "loss": 0.2957,
+      "step": 235
+    },
+    {
+      "epoch": 0.9642492339121552,
+      "grad_norm": 0.34486424922943115,
+      "learning_rate": 1.6043139054613326e-05,
+      "loss": 0.3332,
+      "step": 236
+    },
+    {
+      "epoch": 0.9683350357507661,
+      "grad_norm": 0.38448622822761536,
+      "learning_rate": 1.600742264237979e-05,
+      "loss": 0.3209,
+      "step": 237
+    },
+    {
+      "epoch": 0.9724208375893769,
+      "grad_norm": 0.3303643465042114,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.2967,
+      "step": 238
+    },
+    {
+      "epoch": 0.9765066394279878,
+      "grad_norm": 0.3266353905200958,
+      "learning_rate": 1.5935629596274345e-05,
+      "loss": 0.3328,
+      "step": 239
+    },
+    {
+      "epoch": 0.9805924412665986,
+      "grad_norm": 0.32337427139282227,
+      "learning_rate": 1.5899554400231233e-05,
+      "loss": 0.3077,
+      "step": 240
+    },
+    {
+      "epoch": 0.9846782431052093,
+      "grad_norm": 0.3424544632434845,
+      "learning_rate": 1.586336105139127e-05,
+      "loss": 0.3404,
+      "step": 241
+    },
+    {
+      "epoch": 0.9887640449438202,
+      "grad_norm": 0.3608018755912781,
+      "learning_rate": 1.5827050274613512e-05,
+      "loss": 0.305,
+      "step": 242
+    },
+    {
+      "epoch": 0.992849846782431,
+      "grad_norm": 0.3088077902793884,
+      "learning_rate": 1.579062279710879e-05,
+      "loss": 0.3021,
+      "step": 243
+    },
+    {
+      "epoch": 0.9969356486210419,
+      "grad_norm": 0.3329155743122101,
+      "learning_rate": 1.5754079348425137e-05,
+      "loss": 0.3141,
+      "step": 244
+    },
+    {
+      "epoch": 1.0040858018386107,
+      "grad_norm": 0.47268807888031006,
+      "learning_rate": 1.57174206604332e-05,
+      "loss": 0.5285,
+      "step": 245
+    },
+    {
+      "epoch": 1.0081716036772217,
+      "grad_norm": 0.19290198385715485,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 0.2298,
+      "step": 246
+    },
+    {
+      "epoch": 1.0081716036772217,
+      "eval_loss": 0.3009270131587982,
+      "eval_runtime": 5.657,
+      "eval_samples_per_second": 13.965,
+      "eval_steps_per_second": 1.768,
+      "step": 246
+    },
+    {
+      "epoch": 1.0122574055158324,
+      "grad_norm": 0.18387500941753387,
+      "learning_rate": 1.564376050553205e-05,
+      "loss": 0.2149,
+      "step": 247
+    },
+    {
+      "epoch": 1.0163432073544434,
+      "grad_norm": 0.22477443516254425,
+      "learning_rate": 1.560676051384499e-05,
+      "loss": 0.2182,
+      "step": 248
+    },
+    {
+      "epoch": 1.0204290091930541,
+      "grad_norm": 0.2133513242006302,
+      "learning_rate": 1.5569648233264395e-05,
+      "loss": 0.2251,
+      "step": 249
+    },
+    {
+      "epoch": 1.0245148110316649,
+      "grad_norm": 0.21625001728534698,
+      "learning_rate": 1.553242440705314e-05,
+      "loss": 0.2231,
+      "step": 250
+    },
+    {
+      "epoch": 1.0286006128702758,
+      "grad_norm": 0.21452628076076508,
+      "learning_rate": 1.5495089780708062e-05,
+      "loss": 0.2129,
+      "step": 251
+    },
+    {
+      "epoch": 1.0326864147088866,
+      "grad_norm": 0.22355499863624573,
+      "learning_rate": 1.5457645101945046e-05,
+      "loss": 0.2248,
+      "step": 252
+    },
+    {
+      "epoch": 1.0367722165474975,
+      "grad_norm": 0.247604101896286,
+      "learning_rate": 1.5420091120684042e-05,
+      "loss": 0.2473,
+      "step": 253
+    },
+    {
+      "epoch": 1.0408580183861083,
+      "grad_norm": 0.2052946239709854,
+      "learning_rate": 1.538242858903404e-05,
+      "loss": 0.2322,
+      "step": 254
+    },
+    {
+      "epoch": 1.0449438202247192,
+      "grad_norm": 0.23112209141254425,
+      "learning_rate": 1.5344658261278013e-05,
+      "loss": 0.2128,
+      "step": 255
+    },
+    {
+      "epoch": 1.04902962206333,
+      "grad_norm": 0.236256405711174,
+      "learning_rate": 1.530678089385782e-05,
+      "loss": 0.2062,
+      "step": 256
+    },
+    {
+      "epoch": 1.0531154239019407,
+      "grad_norm": 0.2142944186925888,
+      "learning_rate": 1.5268797245359035e-05,
+      "loss": 0.2222,
+      "step": 257
+    },
+    {
+      "epoch": 1.0572012257405516,
+      "grad_norm": 0.21305693686008453,
+      "learning_rate": 1.5230708076495777e-05,
+      "loss": 0.2192,
+      "step": 258
+    },
+    {
+      "epoch": 1.0612870275791624,
+      "grad_norm": 0.20170477032661438,
+      "learning_rate": 1.519251415009546e-05,
+      "loss": 0.2107,
+      "step": 259
+    },
+    {
+      "epoch": 1.0653728294177733,
+      "grad_norm": 0.22234131395816803,
+      "learning_rate": 1.5154216231083522e-05,
+      "loss": 0.2298,
+      "step": 260
+    },
+    {
+      "epoch": 1.069458631256384,
+      "grad_norm": 0.21675555408000946,
+      "learning_rate": 1.5115815086468103e-05,
+      "loss": 0.2296,
+      "step": 261
+    },
+    {
+      "epoch": 1.0735444330949948,
+      "grad_norm": 0.24653805792331696,
+      "learning_rate": 1.507731148532468e-05,
+      "loss": 0.2209,
+      "step": 262
+    },
+    {
+      "epoch": 1.0776302349336058,
+      "grad_norm": 0.21882732212543488,
+      "learning_rate": 1.5038706198780673e-05,
+      "loss": 0.2239,
+      "step": 263
+    },
+    {
+      "epoch": 1.0817160367722165,
+      "grad_norm": 0.19836413860321045,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2286,
+      "step": 264
+    },
+    {
+      "epoch": 1.0858018386108275,
+      "grad_norm": 0.22002094984054565,
+      "learning_rate": 1.496119366416759e-05,
+      "loss": 0.2261,
+      "step": 265
+    },
+    {
+      "epoch": 1.0898876404494382,
+      "grad_norm": 0.22356298565864563,
+      "learning_rate": 1.492228796847385e-05,
+      "loss": 0.2165,
+      "step": 266
+    },
+    {
+      "epoch": 1.093973442288049,
+      "grad_norm": 0.19260676205158234,
+      "learning_rate": 1.4883283692099114e-05,
+      "loss": 0.217,
+      "step": 267
+    },
+    {
+      "epoch": 1.09805924412666,
+      "grad_norm": 0.2038644701242447,
+      "learning_rate": 1.4844181616198028e-05,
+      "loss": 0.2234,
+      "step": 268
+    },
+    {
+      "epoch": 1.1021450459652706,
+      "grad_norm": 0.24655131995677948,
+      "learning_rate": 1.4804982523883915e-05,
+      "loss": 0.2316,
+      "step": 269
+    },
+    {
+      "epoch": 1.1062308478038816,
+      "grad_norm": 0.2111218273639679,
+      "learning_rate": 1.4765687200213079e-05,
+      "loss": 0.2258,
+      "step": 270
+    },
+    {
+      "epoch": 1.1103166496424923,
+      "grad_norm": 0.20191776752471924,
+      "learning_rate": 1.4726296432169095e-05,
+      "loss": 0.2168,
+      "step": 271
+    },
+    {
+      "epoch": 1.1144024514811033,
+      "grad_norm": 0.20933739840984344,
+      "learning_rate": 1.4686811008647037e-05,
+      "loss": 0.2141,
+      "step": 272
+    },
+    {
+      "epoch": 1.118488253319714,
+      "grad_norm": 0.21544787287712097,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 0.2278,
+      "step": 273
+    },
+    {
+      "epoch": 1.1225740551583248,
+      "grad_norm": 0.21159909665584564,
+      "learning_rate": 1.4607559360211688e-05,
+      "loss": 0.2131,
+      "step": 274
+    },
+    {
+      "epoch": 1.1266598569969357,
+      "grad_norm": 0.20316191017627716,
+      "learning_rate": 1.456779472250368e-05,
+      "loss": 0.2154,
+      "step": 275
+    },
+    {
+      "epoch": 1.1307456588355465,
+      "grad_norm": 0.19834713637828827,
+      "learning_rate": 1.4527938603696376e-05,
+      "loss": 0.2181,
+      "step": 276
+    },
+    {
+      "epoch": 1.1348314606741572,
+      "grad_norm": 0.19414545595645905,
+      "learning_rate": 1.4487991802004625e-05,
+      "loss": 0.2198,
+      "step": 277
+    },
+    {
+      "epoch": 1.1389172625127681,
+      "grad_norm": 0.19380849599838257,
+      "learning_rate": 1.4447955117459414e-05,
+      "loss": 0.2219,
+      "step": 278
+    },
+    {
+      "epoch": 1.1430030643513789,
+      "grad_norm": 0.20117753744125366,
+      "learning_rate": 1.4407829351891858e-05,
+      "loss": 0.2195,
+      "step": 279
+    },
+    {
+      "epoch": 1.1470888661899898,
+      "grad_norm": 0.20845960080623627,
+      "learning_rate": 1.436761530891713e-05,
+      "loss": 0.2331,
+      "step": 280
+    },
+    {
+      "epoch": 1.1511746680286006,
+      "grad_norm": 0.20118166506290436,
+      "learning_rate": 1.4327313793918362e-05,
+      "loss": 0.2106,
+      "step": 281
+    },
+    {
+      "epoch": 1.1552604698672115,
+      "grad_norm": 0.2053423970937729,
+      "learning_rate": 1.4286925614030542e-05,
+      "loss": 0.2118,
+      "step": 282
+    },
+    {
+      "epoch": 1.1593462717058223,
+      "grad_norm": 0.2187694013118744,
+      "learning_rate": 1.4246451578124321e-05,
+      "loss": 0.2227,
+      "step": 283
+    },
+    {
+      "epoch": 1.163432073544433,
+      "grad_norm": 0.20455628633499146,
+      "learning_rate": 1.4205892496789816e-05,
+      "loss": 0.223,
+      "step": 284
+    },
+    {
+      "epoch": 1.167517875383044,
+      "grad_norm": 0.1947728395462036,
+      "learning_rate": 1.4165249182320401e-05,
+      "loss": 0.2045,
+      "step": 285
+    },
+    {
+      "epoch": 1.1716036772216547,
+      "grad_norm": 0.1993764191865921,
+      "learning_rate": 1.4124522448696407e-05,
+      "loss": 0.2111,
+      "step": 286
+    },
+    {
+      "epoch": 1.1756894790602657,
+      "grad_norm": 0.20859940350055695,
+      "learning_rate": 1.4083713111568841e-05,
+      "loss": 0.2138,
+      "step": 287
+    },
+    {
+      "epoch": 1.1797752808988764,
+      "grad_norm": 0.2091362327337265,
+      "learning_rate": 1.404282198824305e-05,
+      "loss": 0.2233,
+      "step": 288
+    },
+    {
+      "epoch": 1.1838610827374871,
+      "grad_norm": 0.21919092535972595,
+      "learning_rate": 1.4001849897662337e-05,
+      "loss": 0.2004,
+      "step": 289
+    },
+    {
+      "epoch": 1.187946884576098,
+      "grad_norm": 0.21487314999103546,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.2116,
+      "step": 290
+    },
+    {
+      "epoch": 1.1920326864147088,
+      "grad_norm": 0.19989344477653503,
+      "learning_rate": 1.3919666098600753e-05,
+      "loss": 0.1993,
+      "step": 291
+    },
+    {
+      "epoch": 1.1961184882533198,
+      "grad_norm": 0.22701528668403625,
+      "learning_rate": 1.387845603604855e-05,
+      "loss": 0.2261,
+      "step": 292
+    },
+    {
+      "epoch": 1.2002042900919305,
+      "grad_norm": 0.20408843457698822,
+      "learning_rate": 1.3837168298065798e-05,
+      "loss": 0.2001,
+      "step": 293
+    },
+    {
+      "epoch": 1.2042900919305413,
+      "grad_norm": 0.19744303822517395,
+      "learning_rate": 1.3795803711538966e-05,
+      "loss": 0.2216,
+      "step": 294
+    },
+    {
+      "epoch": 1.2083758937691522,
+      "grad_norm": 0.19279907643795013,
+      "learning_rate": 1.37543631048936e-05,
+      "loss": 0.2086,
+      "step": 295
+    },
+    {
+      "epoch": 1.212461695607763,
+      "grad_norm": 0.20809710025787354,
+      "learning_rate": 1.3712847308077737e-05,
+      "loss": 0.2192,
+      "step": 296
+    },
+    {
+      "epoch": 1.216547497446374,
+      "grad_norm": 0.20847246050834656,
+      "learning_rate": 1.3671257152545277e-05,
+      "loss": 0.191,
+      "step": 297
+    },
+    {
+      "epoch": 1.2206332992849847,
+      "grad_norm": 0.20084427297115326,
+      "learning_rate": 1.3629593471239328e-05,
+      "loss": 0.2143,
+      "step": 298
+    },
+    {
+      "epoch": 1.2247191011235956,
+      "grad_norm": 0.21558277308940887,
+      "learning_rate": 1.3587857098575534e-05,
+      "loss": 0.2075,
+      "step": 299
+    },
+    {
+      "epoch": 1.2288049029622063,
+      "grad_norm": 0.20744889974594116,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 0.2204,
+      "step": 300
+    },
+    {
+      "epoch": 1.232890704800817,
+      "grad_norm": 0.22505250573158264,
+      "learning_rate": 1.350416962409934e-05,
+      "loss": 0.2217,
+      "step": 301
+    },
+    {
+      "epoch": 1.236976506639428,
+      "grad_norm": 0.1689983457326889,
+      "learning_rate": 1.346222019833033e-05,
+      "loss": 0.2111,
+      "step": 302
+    },
+    {
+      "epoch": 1.2410623084780388,
+      "grad_norm": 0.208964005112648,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.2033,
+      "step": 303
+    },
+    {
+      "epoch": 1.2451481103166497,
+      "grad_norm": 0.19528621435165405,
+      "learning_rate": 1.3378114170405473e-05,
+      "loss": 0.2101,
+      "step": 304
+    },
+    {
+      "epoch": 1.2492339121552605,
+      "grad_norm": 0.21318338811397552,
+      "learning_rate": 1.3335959252675566e-05,
+      "loss": 0.2175,
+      "step": 305
+    },
+    {
+      "epoch": 1.2533197139938714,
+      "grad_norm": 0.196754589676857,
+      "learning_rate": 1.3293737524320798e-05,
+      "loss": 0.2186,
+      "step": 306
+    },
+    {
+      "epoch": 1.2574055158324822,
+      "grad_norm": 0.20901250839233398,
+      "learning_rate": 1.3251449830933052e-05,
+      "loss": 0.207,
+      "step": 307
+    },
+    {
+      "epoch": 1.261491317671093,
+      "grad_norm": 0.1850128173828125,
+      "learning_rate": 1.3209097019425317e-05,
+      "loss": 0.2126,
+      "step": 308
+    },
+    {
+      "epoch": 1.2655771195097039,
+      "grad_norm": 0.19966518878936768,
+      "learning_rate": 1.3166679938014728e-05,
+      "loss": 0.2229,
+      "step": 309
+    },
+    {
+      "epoch": 1.2696629213483146,
+      "grad_norm": 0.19698403775691986,
+      "learning_rate": 1.3124199436205575e-05,
+      "loss": 0.202,
+      "step": 310
+    },
+    {
+      "epoch": 1.2737487231869253,
+      "grad_norm": 0.19038128852844238,
+      "learning_rate": 1.3081656364772308e-05,
+      "loss": 0.2102,
+      "step": 311
+    },
+    {
+      "epoch": 1.2778345250255363,
+      "grad_norm": 0.21141470968723297,
+      "learning_rate": 1.303905157574247e-05,
+      "loss": 0.2131,
+      "step": 312
+    },
+    {
+      "epoch": 1.281920326864147,
+      "grad_norm": 0.19535590708255768,
+      "learning_rate": 1.2996385922379657e-05,
+      "loss": 0.2014,
+      "step": 313
+    },
+    {
+      "epoch": 1.286006128702758,
+      "grad_norm": 0.21494199335575104,
+      "learning_rate": 1.2953660259166413e-05,
+      "loss": 0.221,
+      "step": 314
+    },
+    {
+      "epoch": 1.2900919305413687,
+      "grad_norm": 0.20786255598068237,
+      "learning_rate": 1.291087544178713e-05,
+      "loss": 0.21,
+      "step": 315
+    },
+    {
+      "epoch": 1.2941777323799797,
+      "grad_norm": 0.2005888670682907,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.2142,
+      "step": 316
+    },
+    {
+      "epoch": 1.2982635342185904,
+      "grad_norm": 0.20467688143253326,
+      "learning_rate": 1.2825131773174371e-05,
+      "loss": 0.2115,
+      "step": 317
+    },
+    {
+      "epoch": 1.3023493360572012,
+      "grad_norm": 0.20571133494377136,
+      "learning_rate": 1.2782174639164528e-05,
+      "loss": 0.1892,
+      "step": 318
+    },
+    {
+      "epoch": 1.3064351378958121,
+      "grad_norm": 0.19496022164821625,
+      "learning_rate": 1.2739161785401525e-05,
+      "loss": 0.2074,
+      "step": 319
+    },
+    {
+      "epoch": 1.3105209397344229,
+      "grad_norm": 0.18286117911338806,
+      "learning_rate": 1.269609407332144e-05,
+      "loss": 0.2112,
+      "step": 320
+    },
+    {
+      "epoch": 1.3146067415730336,
+      "grad_norm": 0.1806853860616684,
+      "learning_rate": 1.2652972365459008e-05,
+      "loss": 0.2261,
+      "step": 321
+    },
+    {
+      "epoch": 1.3186925434116445,
+      "grad_norm": 0.22207584977149963,
+      "learning_rate": 1.2609797525430374e-05,
+      "loss": 0.2246,
+      "step": 322
+    },
+    {
+      "epoch": 1.3227783452502553,
+      "grad_norm": 0.19839420914649963,
+      "learning_rate": 1.2566570417915769e-05,
+      "loss": 0.2022,
+      "step": 323
+    },
+    {
+      "epoch": 1.3268641470888662,
+      "grad_norm": 0.21665321290493011,
+      "learning_rate": 1.2523291908642219e-05,
+      "loss": 0.2234,
+      "step": 324
+    },
+    {
+      "epoch": 1.330949948927477,
+      "grad_norm": 0.2002275288105011,
+      "learning_rate": 1.2479962864366186e-05,
+      "loss": 0.2089,
+      "step": 325
+    },
+    {
+      "epoch": 1.335035750766088,
+      "grad_norm": 0.22598706185817719,
+      "learning_rate": 1.243658415285622e-05,
+      "loss": 0.2098,
+      "step": 326
+    },
+    {
+      "epoch": 1.3391215526046987,
+      "grad_norm": 0.2339962273836136,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 0.2385,
+      "step": 327
+    },
+    {
+      "epoch": 1.3432073544433094,
+      "grad_norm": 0.21430878341197968,
+      "learning_rate": 1.2349681204164823e-05,
+      "loss": 0.2219,
+      "step": 328
+    },
+    {
+      "epoch": 1.3432073544433094,
+      "eval_loss": 0.31022945046424866,
+      "eval_runtime": 5.6917,
+      "eval_samples_per_second": 13.88,
+      "eval_steps_per_second": 1.757,
+      "step": 328
+    },
+    {
+      "epoch": 1.3472931562819204,
+      "grad_norm": 0.23316442966461182,
+      "learning_rate": 1.2306158707424402e-05,
+      "loss": 0.2149,
+      "step": 329
+    },
+    {
+      "epoch": 1.351378958120531,
+      "grad_norm": 0.2072836458683014,
+      "learning_rate": 1.2262590024297226e-05,
+      "loss": 0.2143,
+      "step": 330
+    },
+    {
+      "epoch": 1.355464759959142,
+      "grad_norm": 0.18483592569828033,
+      "learning_rate": 1.2218976027351177e-05,
+      "loss": 0.2119,
+      "step": 331
+    },
+    {
+      "epoch": 1.3595505617977528,
+      "grad_norm": 0.2679055333137512,
+      "learning_rate": 1.2175317590061676e-05,
+      "loss": 0.2135,
+      "step": 332
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.23311381042003632,
+      "learning_rate": 1.2131615586794162e-05,
+      "loss": 0.2211,
+      "step": 333
+    },
+    {
+      "epoch": 1.3677221654749745,
+      "grad_norm": 0.20625115931034088,
+      "learning_rate": 1.2087870892786588e-05,
+      "loss": 0.2247,
+      "step": 334
+    },
+    {
+      "epoch": 1.3718079673135852,
+      "grad_norm": 0.19579333066940308,
+      "learning_rate": 1.2044084384131891e-05,
+      "loss": 0.2033,
+      "step": 335
+    },
+    {
+      "epoch": 1.3758937691521962,
+      "grad_norm": 0.20389385521411896,
+      "learning_rate": 1.2000256937760446e-05,
+      "loss": 0.1935,
+      "step": 336
+    },
+    {
+      "epoch": 1.379979570990807,
+      "grad_norm": 0.19781693816184998,
+      "learning_rate": 1.1956389431422508e-05,
+      "loss": 0.2225,
+      "step": 337
+    },
+    {
+      "epoch": 1.3840653728294177,
+      "grad_norm": 0.19394108653068542,
+      "learning_rate": 1.1912482743670624e-05,
+      "loss": 0.199,
+      "step": 338
+    },
+    {
+      "epoch": 1.3881511746680286,
+      "grad_norm": 0.20754311978816986,
+      "learning_rate": 1.1868537753842052e-05,
+      "loss": 0.2127,
+      "step": 339
+    },
+    {
+      "epoch": 1.3922369765066394,
+      "grad_norm": 0.20178458094596863,
+      "learning_rate": 1.1824555342041129e-05,
+      "loss": 0.222,
+      "step": 340
+    },
+    {
+      "epoch": 1.3963227783452503,
+      "grad_norm": 0.21634997427463531,
+      "learning_rate": 1.1780536389121668e-05,
+      "loss": 0.2201,
+      "step": 341
+    },
+    {
+      "epoch": 1.400408580183861,
+      "grad_norm": 0.22160372138023376,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.2137,
+      "step": 342
+    },
+    {
+      "epoch": 1.404494382022472,
+      "grad_norm": 0.2184736281633377,
+      "learning_rate": 1.1692392386983837e-05,
+      "loss": 0.2247,
+      "step": 343
+    },
+    {
+      "epoch": 1.4085801838610827,
+      "grad_norm": 0.19715434312820435,
+      "learning_rate": 1.1648269103061567e-05,
+      "loss": 0.2119,
+      "step": 344
+    },
+    {
+      "epoch": 1.4126659856996935,
+      "grad_norm": 0.2113262563943863,
+      "learning_rate": 1.1604112808577603e-05,
+      "loss": 0.2108,
+      "step": 345
+    },
+    {
+      "epoch": 1.4167517875383044,
+      "grad_norm": 0.22849951684474945,
+      "learning_rate": 1.155992438786818e-05,
+      "loss": 0.2193,
+      "step": 346
+    },
+    {
+      "epoch": 1.4208375893769152,
+      "grad_norm": 0.20339028537273407,
+      "learning_rate": 1.1515704725912926e-05,
+      "loss": 0.2041,
+      "step": 347
+    },
+    {
+      "epoch": 1.424923391215526,
+      "grad_norm": 0.21201005578041077,
+      "learning_rate": 1.1471454708317163e-05,
+      "loss": 0.2105,
+      "step": 348
+    },
+    {
+      "epoch": 1.4290091930541369,
+      "grad_norm": 0.22068007290363312,
+      "learning_rate": 1.1427175221294145e-05,
+      "loss": 0.2198,
+      "step": 349
+    },
+    {
+      "epoch": 1.4330949948927478,
+      "grad_norm": 0.2174069881439209,
+      "learning_rate": 1.1382867151647333e-05,
+      "loss": 0.2093,
+      "step": 350
+    },
+    {
+      "epoch": 1.4371807967313586,
+      "grad_norm": 0.193963423371315,
+      "learning_rate": 1.1338531386752618e-05,
+      "loss": 0.2068,
+      "step": 351
+    },
+    {
+      "epoch": 1.4412665985699693,
+      "grad_norm": 0.20459222793579102,
+      "learning_rate": 1.1294168814540554e-05,
+      "loss": 0.1933,
+      "step": 352
+    },
+    {
+      "epoch": 1.4453524004085803,
+      "grad_norm": 0.22635479271411896,
+      "learning_rate": 1.1249780323478585e-05,
+      "loss": 0.2065,
+      "step": 353
+    },
+    {
+      "epoch": 1.449438202247191,
+      "grad_norm": 0.194682314991951,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 0.1927,
+      "step": 354
+    },
+    {
+      "epoch": 1.4535240040858017,
+      "grad_norm": 0.21696004271507263,
+      "learning_rate": 1.1160929141252303e-05,
+      "loss": 0.2278,
+      "step": 355
+    },
+    {
+      "epoch": 1.4576098059244127,
+      "grad_norm": 0.21389201283454895,
+      "learning_rate": 1.1116468229547079e-05,
+      "loss": 0.2002,
+      "step": 356
+    },
+    {
+      "epoch": 1.4616956077630234,
+      "grad_norm": 0.2095150500535965,
+      "learning_rate": 1.107198495787448e-05,
+      "loss": 0.2068,
+      "step": 357
+    },
+    {
+      "epoch": 1.4657814096016344,
+      "grad_norm": 0.17831739783287048,
+      "learning_rate": 1.1027480217119245e-05,
+      "loss": 0.197,
+      "step": 358
+    },
+    {
+      "epoch": 1.4698672114402451,
+      "grad_norm": 0.20202742516994476,
+      "learning_rate": 1.0982954898596072e-05,
+      "loss": 0.2045,
+      "step": 359
+    },
+    {
+      "epoch": 1.473953013278856,
+      "grad_norm": 0.22862711548805237,
+      "learning_rate": 1.0938409894031793e-05,
+      "loss": 0.2325,
+      "step": 360
+    },
+    {
+      "epoch": 1.4780388151174668,
+      "grad_norm": 0.1705763190984726,
+      "learning_rate": 1.0893846095547493e-05,
+      "loss": 0.1981,
+      "step": 361
+    },
+    {
+      "epoch": 1.4821246169560776,
+      "grad_norm": 0.18474915623664856,
+      "learning_rate": 1.084926439564065e-05,
+      "loss": 0.2056,
+      "step": 362
+    },
+    {
+      "epoch": 1.4862104187946885,
+      "grad_norm": 0.20779083669185638,
+      "learning_rate": 1.0804665687167262e-05,
+      "loss": 0.2144,
+      "step": 363
+    },
+    {
+      "epoch": 1.4902962206332993,
+      "grad_norm": 0.21080774068832397,
+      "learning_rate": 1.0760050863323961e-05,
+      "loss": 0.2129,
+      "step": 364
+    },
+    {
+      "epoch": 1.49438202247191,
+      "grad_norm": 0.2220941036939621,
+      "learning_rate": 1.0715420817630137e-05,
+      "loss": 0.2143,
+      "step": 365
+    },
+    {
+      "epoch": 1.498467824310521,
+      "grad_norm": 0.2133949249982834,
+      "learning_rate": 1.0670776443910024e-05,
+      "loss": 0.2124,
+      "step": 366
+    },
+    {
+      "epoch": 1.502553626149132,
+      "grad_norm": 0.18696293234825134,
+      "learning_rate": 1.062611863627482e-05,
+      "loss": 0.1942,
+      "step": 367
+    },
+    {
+      "epoch": 1.5066394279877426,
+      "grad_norm": 0.23111994564533234,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.2029,
+      "step": 368
+    },
+    {
+      "epoch": 1.5107252298263534,
+      "grad_norm": 0.22670774161815643,
+      "learning_rate": 1.0536766297031216e-05,
+      "loss": 0.2142,
+      "step": 369
+    },
+    {
+      "epoch": 1.5148110316649643,
+      "grad_norm": 0.2128949612379074,
+      "learning_rate": 1.0492073554918782e-05,
+      "loss": 0.2159,
+      "step": 370
+    },
+    {
+      "epoch": 1.518896833503575,
+      "grad_norm": 0.22443941235542297,
+      "learning_rate": 1.0447370957847343e-05,
+      "loss": 0.2142,
+      "step": 371
+    },
+    {
+      "epoch": 1.5229826353421858,
+      "grad_norm": 0.2178109586238861,
+      "learning_rate": 1.0402659401094154e-05,
+      "loss": 0.2087,
+      "step": 372
+    },
+    {
+      "epoch": 1.5270684371807968,
+      "grad_norm": 0.20667575299739838,
+      "learning_rate": 1.0357939780115906e-05,
+      "loss": 0.2106,
+      "step": 373
+    },
+    {
+      "epoch": 1.5311542390194075,
+      "grad_norm": 0.22103561460971832,
+      "learning_rate": 1.0313212990530804e-05,
+      "loss": 0.2052,
+      "step": 374
+    },
+    {
+      "epoch": 1.5352400408580182,
+      "grad_norm": 0.24241849780082703,
+      "learning_rate": 1.0268479928100615e-05,
+      "loss": 0.2166,
+      "step": 375
+    },
+    {
+      "epoch": 1.5393258426966292,
+      "grad_norm": 0.2349155843257904,
+      "learning_rate": 1.0223741488712732e-05,
+      "loss": 0.2065,
+      "step": 376
+    },
+    {
+      "epoch": 1.5434116445352402,
+      "grad_norm": 0.21443332731723785,
+      "learning_rate": 1.0178998568362243e-05,
+      "loss": 0.215,
+      "step": 377
+    },
+    {
+      "epoch": 1.547497446373851,
+      "grad_norm": 0.20891189575195312,
+      "learning_rate": 1.0134252063133976e-05,
+      "loss": 0.2057,
+      "step": 378
+    },
+    {
+      "epoch": 1.5515832482124616,
+      "grad_norm": 0.1952524334192276,
+      "learning_rate": 1.0089502869184549e-05,
+      "loss": 0.2008,
+      "step": 379
+    },
+    {
+      "epoch": 1.5556690500510726,
+      "grad_norm": 0.21196359395980835,
+      "learning_rate": 1.0044751882724436e-05,
+      "loss": 0.2076,
+      "step": 380
+    },
+    {
+      "epoch": 1.5597548518896833,
+      "grad_norm": 0.2183002382516861,
+      "learning_rate": 1e-05,
+      "loss": 0.2136,
+      "step": 381
+    },
+    {
+      "epoch": 1.563840653728294,
+      "grad_norm": 0.18016640841960907,
+      "learning_rate": 9.955248117275566e-06,
+      "loss": 0.215,
+      "step": 382
+    },
+    {
+      "epoch": 1.567926455566905,
+      "grad_norm": 0.2017180472612381,
+      "learning_rate": 9.910497130815454e-06,
+      "loss": 0.2215,
+      "step": 383
+    },
+    {
+      "epoch": 1.572012257405516,
+      "grad_norm": 0.19466502964496613,
+      "learning_rate": 9.865747936866027e-06,
+      "loss": 0.2062,
+      "step": 384
+    },
+    {
+      "epoch": 1.5760980592441267,
+      "grad_norm": 0.20503178238868713,
+      "learning_rate": 9.821001431637759e-06,
+      "loss": 0.2043,
+      "step": 385
+    },
+    {
+      "epoch": 1.5801838610827375,
+      "grad_norm": 0.2414986789226532,
+      "learning_rate": 9.776258511287271e-06,
+      "loss": 0.2063,
+      "step": 386
+    },
+    {
+      "epoch": 1.5842696629213484,
+      "grad_norm": 0.21953481435775757,
+      "learning_rate": 9.73152007189939e-06,
+      "loss": 0.2246,
+      "step": 387
+    },
+    {
+      "epoch": 1.5883554647599591,
+      "grad_norm": 0.20504049956798553,
+      "learning_rate": 9.6867870094692e-06,
+      "loss": 0.1999,
+      "step": 388
+    },
+    {
+      "epoch": 1.5924412665985699,
+      "grad_norm": 0.17664040625095367,
+      "learning_rate": 9.642060219884096e-06,
+      "loss": 0.1965,
+      "step": 389
+    },
+    {
+      "epoch": 1.5965270684371808,
+      "grad_norm": 0.2129695564508438,
+      "learning_rate": 9.597340598905851e-06,
+      "loss": 0.2218,
+      "step": 390
+    },
+    {
+      "epoch": 1.6006128702757916,
+      "grad_norm": 0.20206192135810852,
+      "learning_rate": 9.55262904215266e-06,
+      "loss": 0.2126,
+      "step": 391
+    },
+    {
+      "epoch": 1.6046986721144023,
+      "grad_norm": 0.19526894390583038,
+      "learning_rate": 9.50792644508122e-06,
+      "loss": 0.2032,
+      "step": 392
+    },
+    {
+      "epoch": 1.6087844739530133,
+      "grad_norm": 0.22161059081554413,
+      "learning_rate": 9.463233702968784e-06,
+      "loss": 0.2228,
+      "step": 393
+    },
+    {
+      "epoch": 1.6128702757916242,
+      "grad_norm": 0.22896061837673187,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.2149,
+      "step": 394
+    },
+    {
+      "epoch": 1.616956077630235,
+      "grad_norm": 0.19969528913497925,
+      "learning_rate": 9.373881363725182e-06,
+      "loss": 0.2059,
+      "step": 395
+    },
+    {
+      "epoch": 1.6210418794688457,
+      "grad_norm": 0.19778963923454285,
+      "learning_rate": 9.329223556089976e-06,
+      "loss": 0.1946,
+      "step": 396
+    },
+    {
+      "epoch": 1.6251276813074567,
+      "grad_norm": 0.21220386028289795,
+      "learning_rate": 9.284579182369868e-06,
+      "loss": 0.2047,
+      "step": 397
+    },
+    {
+      "epoch": 1.6292134831460674,
+      "grad_norm": 0.20156344771385193,
+      "learning_rate": 9.239949136676042e-06,
+      "loss": 0.2057,
+      "step": 398
+    },
+    {
+      "epoch": 1.6332992849846781,
+      "grad_norm": 0.21823804080486298,
+      "learning_rate": 9.195334312832742e-06,
+      "loss": 0.1971,
+      "step": 399
+    },
+    {
+      "epoch": 1.637385086823289,
+      "grad_norm": 0.1959310621023178,
+      "learning_rate": 9.15073560435935e-06,
+      "loss": 0.1925,
+      "step": 400
+    },
+    {
+      "epoch": 1.6414708886619,
+      "grad_norm": 0.18495595455169678,
+      "learning_rate": 9.10615390445251e-06,
+      "loss": 0.1966,
+      "step": 401
+    },
+    {
+      "epoch": 1.6455566905005106,
+      "grad_norm": 0.21264636516571045,
+      "learning_rate": 9.061590105968208e-06,
+      "loss": 0.2295,
+      "step": 402
+    },
+    {
+      "epoch": 1.6496424923391215,
+      "grad_norm": 0.19887572526931763,
+      "learning_rate": 9.01704510140393e-06,
+      "loss": 0.2005,
+      "step": 403
+    },
+    {
+      "epoch": 1.6537282941777325,
+      "grad_norm": 0.22931122779846191,
+      "learning_rate": 8.97251978288076e-06,
+      "loss": 0.2066,
+      "step": 404
+    },
+    {
+      "epoch": 1.6578140960163432,
+      "grad_norm": 0.20790022611618042,
+      "learning_rate": 8.928015042125523e-06,
+      "loss": 0.2174,
+      "step": 405
+    },
+    {
+      "epoch": 1.661899897854954,
+      "grad_norm": 0.18505467474460602,
+      "learning_rate": 8.883531770452924e-06,
+      "loss": 0.1917,
+      "step": 406
+    },
+    {
+      "epoch": 1.665985699693565,
+      "grad_norm": 0.22820204496383667,
+      "learning_rate": 8.839070858747697e-06,
+      "loss": 0.2176,
+      "step": 407
+    },
+    {
+      "epoch": 1.6700715015321757,
+      "grad_norm": 0.20349323749542236,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 0.2168,
+      "step": 408
+    },
+    {
+      "epoch": 1.6741573033707864,
+      "grad_norm": 0.21869666874408722,
+      "learning_rate": 8.750219676521417e-06,
+      "loss": 0.2074,
+      "step": 409
+    },
+    {
+      "epoch": 1.6782431052093973,
+      "grad_norm": 0.1885787397623062,
+      "learning_rate": 8.705831185459446e-06,
+      "loss": 0.196,
+      "step": 410
+    },
+    {
+      "epoch": 1.6782431052093973,
+      "eval_loss": 0.3017351031303406,
+      "eval_runtime": 5.6615,
+      "eval_samples_per_second": 13.954,
+      "eval_steps_per_second": 1.766,
+      "step": 410
+    },
+    {
+      "epoch": 1.6823289070480083,
+      "grad_norm": 0.19935843348503113,
+      "learning_rate": 8.661468613247387e-06,
+      "loss": 0.2205,
+      "step": 411
+    },
+    {
+      "epoch": 1.686414708886619,
+      "grad_norm": 0.21857048571109772,
+      "learning_rate": 8.617132848352672e-06,
+      "loss": 0.2153,
+      "step": 412
+    },
+    {
+      "epoch": 1.6905005107252298,
+      "grad_norm": 0.22006161510944366,
+      "learning_rate": 8.572824778705858e-06,
+      "loss": 0.207,
+      "step": 413
+    },
+    {
+      "epoch": 1.6945863125638407,
+      "grad_norm": 0.18598918616771698,
+      "learning_rate": 8.528545291682839e-06,
+      "loss": 0.2124,
+      "step": 414
+    },
+    {
+      "epoch": 1.6986721144024515,
+      "grad_norm": 0.20595507323741913,
+      "learning_rate": 8.484295274087077e-06,
+      "loss": 0.1969,
+      "step": 415
+    },
+    {
+      "epoch": 1.7027579162410622,
+      "grad_norm": 0.20170485973358154,
+      "learning_rate": 8.440075612131823e-06,
+      "loss": 0.2135,
+      "step": 416
+    },
+    {
+      "epoch": 1.7068437180796732,
+      "grad_norm": 0.177800714969635,
+      "learning_rate": 8.395887191422397e-06,
+      "loss": 0.1942,
+      "step": 417
+    },
+    {
+      "epoch": 1.7109295199182841,
+      "grad_norm": 0.1993665099143982,
+      "learning_rate": 8.351730896938438e-06,
+      "loss": 0.2128,
+      "step": 418
+    },
+    {
+      "epoch": 1.7150153217568946,
+      "grad_norm": 0.20131590962409973,
+      "learning_rate": 8.307607613016166e-06,
+      "loss": 0.2119,
+      "step": 419
+    },
+    {
+      "epoch": 1.7191011235955056,
+      "grad_norm": 0.21999968588352203,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.2053,
+      "step": 420
+    },
+    {
+      "epoch": 1.7231869254341166,
+      "grad_norm": 0.2145904302597046,
+      "learning_rate": 8.219463610878336e-06,
+      "loss": 0.2163,
+      "step": 421
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 0.19223150610923767,
+      "learning_rate": 8.175444657958875e-06,
+      "loss": 0.1933,
+      "step": 422
+    },
+    {
+      "epoch": 1.731358529111338,
+      "grad_norm": 0.19847947359085083,
+      "learning_rate": 8.131462246157953e-06,
+      "loss": 0.2091,
+      "step": 423
+    },
+    {
+      "epoch": 1.735444330949949,
+      "grad_norm": 0.2146064192056656,
+      "learning_rate": 8.087517256329376e-06,
+      "loss": 0.1981,
+      "step": 424
+    },
+    {
+      "epoch": 1.7395301327885597,
+      "grad_norm": 0.18579524755477905,
+      "learning_rate": 8.043610568577497e-06,
+      "loss": 0.2158,
+      "step": 425
+    },
+    {
+      "epoch": 1.7436159346271705,
+      "grad_norm": 0.2061241865158081,
+      "learning_rate": 7.999743062239557e-06,
+      "loss": 0.1977,
+      "step": 426
+    },
+    {
+      "epoch": 1.7477017364657814,
+      "grad_norm": 0.2085464894771576,
+      "learning_rate": 7.95591561586811e-06,
+      "loss": 0.2048,
+      "step": 427
+    },
+    {
+      "epoch": 1.7517875383043924,
+      "grad_norm": 0.18300579488277435,
+      "learning_rate": 7.912129107213417e-06,
+      "loss": 0.1924,
+      "step": 428
+    },
+    {
+      "epoch": 1.7558733401430031,
+      "grad_norm": 0.20025411248207092,
+      "learning_rate": 7.868384413205842e-06,
+      "loss": 0.2064,
+      "step": 429
+    },
+    {
+      "epoch": 1.7599591419816139,
+      "grad_norm": 0.20789940655231476,
+      "learning_rate": 7.824682409938328e-06,
+      "loss": 0.2078,
+      "step": 430
+    },
+    {
+      "epoch": 1.7640449438202248,
+      "grad_norm": 0.20518545806407928,
+      "learning_rate": 7.781023972648826e-06,
+      "loss": 0.2144,
+      "step": 431
+    },
+    {
+      "epoch": 1.7681307456588355,
+      "grad_norm": 0.20950093865394592,
+      "learning_rate": 7.73740997570278e-06,
+      "loss": 0.2045,
+      "step": 432
+    },
+    {
+      "epoch": 1.7722165474974463,
+      "grad_norm": 0.19912225008010864,
+      "learning_rate": 7.6938412925756e-06,
+      "loss": 0.2015,
+      "step": 433
+    },
+    {
+      "epoch": 1.7763023493360572,
+      "grad_norm": 0.23430877923965454,
+      "learning_rate": 7.650318795835179e-06,
+      "loss": 0.2262,
+      "step": 434
+    },
+    {
+      "epoch": 1.780388151174668,
+      "grad_norm": 0.19112354516983032,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.2059,
+      "step": 435
+    },
+    {
+      "epoch": 1.7844739530132787,
+      "grad_norm": 0.20528246462345123,
+      "learning_rate": 7.563415847143782e-06,
+      "loss": 0.1966,
+      "step": 436
+    },
+    {
+      "epoch": 1.7885597548518897,
+      "grad_norm": 0.19889222085475922,
+      "learning_rate": 7.520037135633817e-06,
+      "loss": 0.2039,
+      "step": 437
+    },
+    {
+      "epoch": 1.7926455566905006,
+      "grad_norm": 0.23072609305381775,
+      "learning_rate": 7.476708091357783e-06,
+      "loss": 0.2142,
+      "step": 438
+    },
+    {
+      "epoch": 1.7967313585291114,
+      "grad_norm": 0.24155430495738983,
+      "learning_rate": 7.433429582084233e-06,
+      "loss": 0.2224,
+      "step": 439
+    },
+    {
+      "epoch": 1.800817160367722,
+      "grad_norm": 0.20754969120025635,
+      "learning_rate": 7.39020247456963e-06,
+      "loss": 0.2087,
+      "step": 440
+    },
+    {
+      "epoch": 1.804902962206333,
+      "grad_norm": 0.20180056989192963,
+      "learning_rate": 7.347027634540993e-06,
+      "loss": 0.2136,
+      "step": 441
+    },
+    {
+      "epoch": 1.8089887640449438,
+      "grad_norm": 0.1921422928571701,
+      "learning_rate": 7.303905926678565e-06,
+      "loss": 0.2002,
+      "step": 442
+    },
+    {
+      "epoch": 1.8130745658835545,
+      "grad_norm": 0.19783474504947662,
+      "learning_rate": 7.260838214598475e-06,
+      "loss": 0.1969,
+      "step": 443
+    },
+    {
+      "epoch": 1.8171603677221655,
+      "grad_norm": 0.19810006022453308,
+      "learning_rate": 7.217825360835475e-06,
+      "loss": 0.1953,
+      "step": 444
+    },
+    {
+      "epoch": 1.8212461695607765,
+      "grad_norm": 0.20617030560970306,
+      "learning_rate": 7.174868226825631e-06,
+      "loss": 0.2057,
+      "step": 445
+    },
+    {
+      "epoch": 1.825331971399387,
+      "grad_norm": 0.20550931990146637,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.2023,
+      "step": 446
+    },
+    {
+      "epoch": 1.829417773237998,
+      "grad_norm": 0.20049825310707092,
+      "learning_rate": 7.089124558212872e-06,
+      "loss": 0.2069,
+      "step": 447
+    },
+    {
+      "epoch": 1.8335035750766089,
+      "grad_norm": 0.20448057353496552,
+      "learning_rate": 7.04633974083359e-06,
+      "loss": 0.2117,
+      "step": 448
+    },
+    {
+      "epoch": 1.8375893769152196,
+      "grad_norm": 0.20932386815547943,
+      "learning_rate": 7.003614077620348e-06,
+      "loss": 0.2075,
+      "step": 449
+    },
+    {
+      "epoch": 1.8416751787538304,
+      "grad_norm": 0.20885604619979858,
+      "learning_rate": 6.960948424257532e-06,
+      "loss": 0.2224,
+      "step": 450
+    },
+    {
+      "epoch": 1.8457609805924413,
+      "grad_norm": 0.20506371557712555,
+      "learning_rate": 6.918343635227694e-06,
+      "loss": 0.2022,
+      "step": 451
+    },
+    {
+      "epoch": 1.849846782431052,
+      "grad_norm": 0.20515413582324982,
+      "learning_rate": 6.8758005637944245e-06,
+      "loss": 0.1968,
+      "step": 452
+    },
+    {
+      "epoch": 1.8539325842696628,
+      "grad_norm": 0.20033307373523712,
+      "learning_rate": 6.833320061985278e-06,
+      "loss": 0.1958,
+      "step": 453
+    },
+    {
+      "epoch": 1.8580183861082737,
+      "grad_norm": 0.18687546253204346,
+      "learning_rate": 6.7909029805746855e-06,
+      "loss": 0.2085,
+      "step": 454
+    },
+    {
+      "epoch": 1.8621041879468847,
+      "grad_norm": 0.20156648755073547,
+      "learning_rate": 6.7485501690669495e-06,
+      "loss": 0.2031,
+      "step": 455
+    },
+    {
+      "epoch": 1.8661899897854954,
+      "grad_norm": 0.1979530155658722,
+      "learning_rate": 6.706262475679205e-06,
+      "loss": 0.2061,
+      "step": 456
+    },
+    {
+      "epoch": 1.8702757916241062,
+      "grad_norm": 0.20796051621437073,
+      "learning_rate": 6.664040747324437e-06,
+      "loss": 0.1968,
+      "step": 457
+    },
+    {
+      "epoch": 1.8743615934627171,
+      "grad_norm": 0.19492687284946442,
+      "learning_rate": 6.62188582959453e-06,
+      "loss": 0.2145,
+      "step": 458
+    },
+    {
+      "epoch": 1.8784473953013279,
+      "grad_norm": 0.21463002264499664,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.195,
+      "step": 459
+    },
+    {
+      "epoch": 1.8825331971399386,
+      "grad_norm": 0.1875930279493332,
+      "learning_rate": 6.537779801669677e-06,
+      "loss": 0.1987,
+      "step": 460
+    },
+    {
+      "epoch": 1.8866189989785496,
+      "grad_norm": 0.22397062182426453,
+      "learning_rate": 6.495830375900665e-06,
+      "loss": 0.2237,
+      "step": 461
+    },
+    {
+      "epoch": 1.8907048008171605,
+      "grad_norm": 0.21807466447353363,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.21,
+      "step": 462
+    },
+    {
+      "epoch": 1.894790602655771,
+      "grad_norm": 0.2611398994922638,
+      "learning_rate": 6.41214290142447e-06,
+      "loss": 0.1902,
+      "step": 463
+    },
+    {
+      "epoch": 1.898876404494382,
+      "grad_norm": 0.1971069723367691,
+      "learning_rate": 6.370406528760675e-06,
+      "loss": 0.1914,
+      "step": 464
+    },
+    {
+      "epoch": 1.902962206332993,
+      "grad_norm": 0.2033158540725708,
+      "learning_rate": 6.3287428474547256e-06,
+      "loss": 0.2088,
+      "step": 465
+    },
+    {
+      "epoch": 1.9070480081716037,
+      "grad_norm": 0.18615838885307312,
+      "learning_rate": 6.287152691922264e-06,
+      "loss": 0.1832,
+      "step": 466
+    },
+    {
+      "epoch": 1.9111338100102144,
+      "grad_norm": 0.2230215072631836,
+      "learning_rate": 6.245636895106403e-06,
+      "loss": 0.216,
+      "step": 467
+    },
+    {
+      "epoch": 1.9152196118488254,
+      "grad_norm": 0.18996527791023254,
+      "learning_rate": 6.204196288461037e-06,
+      "loss": 0.1968,
+      "step": 468
+    },
+    {
+      "epoch": 1.9193054136874361,
+      "grad_norm": 0.2129172682762146,
+      "learning_rate": 6.162831701934203e-06,
+      "loss": 0.204,
+      "step": 469
+    },
+    {
+      "epoch": 1.9233912155260469,
+      "grad_norm": 0.20386871695518494,
+      "learning_rate": 6.121543963951453e-06,
+      "loss": 0.1933,
+      "step": 470
+    },
+    {
+      "epoch": 1.9274770173646578,
+      "grad_norm": 0.1936495006084442,
+      "learning_rate": 6.080333901399252e-06,
+      "loss": 0.1869,
+      "step": 471
+    },
+    {
+      "epoch": 1.9315628192032688,
+      "grad_norm": 0.200775146484375,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.1922,
+      "step": 472
+    },
+    {
+      "epoch": 1.9356486210418795,
+      "grad_norm": 0.19721902906894684,
+      "learning_rate": 5.998150102337665e-06,
+      "loss": 0.1946,
+      "step": 473
+    },
+    {
+      "epoch": 1.9397344228804902,
+      "grad_norm": 0.22053122520446777,
+      "learning_rate": 5.957178011756952e-06,
+      "loss": 0.2083,
+      "step": 474
+    },
+    {
+      "epoch": 1.9438202247191012,
+      "grad_norm": 0.20996588468551636,
+      "learning_rate": 5.9162868884311596e-06,
+      "loss": 0.2069,
+      "step": 475
+    },
+    {
+      "epoch": 1.947906026557712,
+      "grad_norm": 0.2055818736553192,
+      "learning_rate": 5.875477551303596e-06,
+      "loss": 0.2082,
+      "step": 476
+    },
+    {
+      "epoch": 1.9519918283963227,
+      "grad_norm": 0.18722133338451385,
+      "learning_rate": 5.834750817679606e-06,
+      "loss": 0.1904,
+      "step": 477
+    },
+    {
+      "epoch": 1.9560776302349336,
+      "grad_norm": 0.2298428863286972,
+      "learning_rate": 5.794107503210187e-06,
+      "loss": 0.2141,
+      "step": 478
+    },
+    {
+      "epoch": 1.9601634320735446,
+      "grad_norm": 0.20707903802394867,
+      "learning_rate": 5.753548421875686e-06,
+      "loss": 0.2166,
+      "step": 479
+    },
+    {
+      "epoch": 1.9642492339121551,
+      "grad_norm": 0.21654275059700012,
+      "learning_rate": 5.713074385969457e-06,
+      "loss": 0.207,
+      "step": 480
+    },
+    {
+      "epoch": 1.968335035750766,
+      "grad_norm": 0.1973462849855423,
+      "learning_rate": 5.672686206081638e-06,
+      "loss": 0.1936,
+      "step": 481
+    },
+    {
+      "epoch": 1.972420837589377,
+      "grad_norm": 0.2014143168926239,
+      "learning_rate": 5.632384691082874e-06,
+      "loss": 0.2031,
+      "step": 482
+    },
+    {
+      "epoch": 1.9765066394279878,
+      "grad_norm": 0.19778530299663544,
+      "learning_rate": 5.5921706481081405e-06,
+      "loss": 0.1984,
+      "step": 483
+    },
+    {
+      "epoch": 1.9805924412665985,
+      "grad_norm": 0.20483283698558807,
+      "learning_rate": 5.55204488254059e-06,
+      "loss": 0.1892,
+      "step": 484
+    },
+    {
+      "epoch": 1.9846782431052095,
+      "grad_norm": 0.18423794209957123,
+      "learning_rate": 5.512008197995379e-06,
+      "loss": 0.1954,
+      "step": 485
+    },
+    {
+      "epoch": 1.9887640449438202,
+      "grad_norm": 0.19180485606193542,
+      "learning_rate": 5.47206139630363e-06,
+      "loss": 0.1983,
+      "step": 486
+    },
+    {
+      "epoch": 1.992849846782431,
+      "grad_norm": 0.21696338057518005,
+      "learning_rate": 5.432205277496327e-06,
+      "loss": 0.2151,
+      "step": 487
+    },
+    {
+      "epoch": 1.996935648621042,
+      "grad_norm": 0.20143908262252808,
+      "learning_rate": 5.3924406397883174e-06,
+      "loss": 0.2001,
+      "step": 488
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 732,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 244,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.255680921752044e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

3b-mb_qwen/checkpoint-488/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adb0d71cc227e5931c5bf7b539200b82417ba4790efbcda6702c240d70a34d6a
+size 10744

3b-mb_qwen/checkpoint-488/zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)

3b-mb_qwen/checkpoint-507/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

3b-mb_qwen/checkpoint-507/config.json

@@ -0,0 +1,28 @@
+{
+  "_name_or_path": "Qwen/Qwen2.5-3B-Instruct",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 70,
+  "model_type": "qwen2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.48.1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151665
+}

3b-mb_qwen/checkpoint-507/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.48.1"
+}

3b-mb_qwen/checkpoint-507/latest

@@ -0,0 +1 @@
+global_step505

3b-mb_qwen/checkpoint-507/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c63c90852fa3fa4280db2cd535d3288d97103797c36bc01f6b86838774637395
+size 4956450288

3b-mb_qwen/checkpoint-507/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5abba5a27427c5628dcedab5d617b319036407f2fc964f81ba71cfb4a973b178
+size 1835586736

3b-mb_qwen/checkpoint-507/model.safetensors.index.json

@@ -0,0 +1,442 @@
+{
+  "metadata": {
+    "total_size": 6791987200
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00002-of-00002.safetensors",
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.v_proj.bias": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors"
+  }
+}

3b-mb_qwen/checkpoint-507/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f3803bff3f596c03b55881de967a825b5734e4a581739164f9cb9e7fd1aee89
+size 14512

3b-mb_qwen/checkpoint-507/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d768a04b798e2ca42effbe096b8e4481f32a402a9125a2ced390586dab8eb29e
+size 14512

3b-mb_qwen/checkpoint-507/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7cc083c6282a60be998cf859b1a6c559bc7ea7e1edebd39819cd91f2b32e45e
+size 1064

3b-mb_qwen/checkpoint-507/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}