Upload folder using huggingface_hub
Browse files- checkpoint-1049/config.json +32 -0
- checkpoint-1049/generation_config.json +7 -0
- checkpoint-1049/model.safetensors +3 -0
- checkpoint-1049/optimizer.pt +3 -0
- checkpoint-1049/rng_state.pth +3 -0
- checkpoint-1049/scheduler.pt +3 -0
- checkpoint-1049/trainer_state.json +761 -0
- checkpoint-1049/training_args.bin +3 -0
- checkpoint-2014/config.json +33 -0
- checkpoint-2014/generation_config.json +7 -0
- checkpoint-2014/model.safetensors +3 -0
- checkpoint-2014/optimizer.pt +3 -0
- checkpoint-2014/rng_state.pth +3 -0
- checkpoint-2014/scheduler.pt +3 -0
- checkpoint-2014/trainer_state.json +1440 -0
- checkpoint-2014/training_args.bin +3 -0
- checkpoint-3174/config.json +32 -0
- checkpoint-3174/generation_config.json +7 -0
- checkpoint-3174/model.safetensors +3 -0
- checkpoint-3174/optimizer.pt +3 -0
- checkpoint-3174/rng_state.pth +3 -0
- checkpoint-3174/scheduler.pt +3 -0
- checkpoint-3174/trainer_state.json +2252 -0
- checkpoint-3174/training_args.bin +3 -0
- checkpoint-504/config.json +32 -0
- checkpoint-504/generation_config.json +7 -0
- checkpoint-504/model.safetensors +3 -0
- checkpoint-504/optimizer.pt +3 -0
- checkpoint-504/rng_state.pth +3 -0
- checkpoint-504/scheduler.pt +3 -0
- checkpoint-504/trainer_state.json +383 -0
- checkpoint-504/training_args.bin +3 -0
- config.json +32 -0
- generation_config.json +7 -0
- model.safetensors +3 -0
- special_tokens_map.json +44 -0
- tokenizer.json +0 -0
- tokenizer.model +3 -0
- tokenizer_config.json +46 -0
- training_args.bin +3 -0
checkpoint-1049/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "h2oai/h2o-danube3-500m-chat",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"head_dim": 96,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1536,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 4096,
|
| 15 |
+
"max_position_embeddings": 8192,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 16,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 8,
|
| 21 |
+
"pad_token_id": 0,
|
| 22 |
+
"pretraining_tp": 1,
|
| 23 |
+
"rms_norm_eps": 1e-05,
|
| 24 |
+
"rope_scaling": null,
|
| 25 |
+
"rope_theta": 100000,
|
| 26 |
+
"sliding_window": null,
|
| 27 |
+
"tie_word_embeddings": false,
|
| 28 |
+
"torch_dtype": "float32",
|
| 29 |
+
"transformers_version": "4.46.1",
|
| 30 |
+
"use_cache": true,
|
| 31 |
+
"vocab_size": 32000
|
| 32 |
+
}
|
checkpoint-1049/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.46.1"
|
| 7 |
+
}
|
checkpoint-1049/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:16d72b96975b9bfe80b9e0a4dcda75e512328798da055cd9e7ecb00ab6af7404
|
| 3 |
+
size 2054379856
|
checkpoint-1049/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cf6ca9781911193a13c48baee0519c0ce5b7ce66c1923c0de4aa827320768bc9
|
| 3 |
+
size 4108850298
|
checkpoint-1049/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:386fcc8cc1089aade9450d86fb239ea3483f455fd2d78d8378645feecfec9d69
|
| 3 |
+
size 14244
|
checkpoint-1049/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:39fd759594e38f9a163215368d8bb30356ccaedff88016b3377c4458da4b7a2b
|
| 3 |
+
size 1064
|
checkpoint-1049/trainer_state.json
ADDED
|
@@ -0,0 +1,761 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 5000,
|
| 6 |
+
"global_step": 1049,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.009532888465204958,
|
| 13 |
+
"grad_norm": 9735.205078125,
|
| 14 |
+
"learning_rate": 6.000000000000001e-07,
|
| 15 |
+
"loss": 14.3342,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.019065776930409915,
|
| 20 |
+
"grad_norm": 12878.38671875,
|
| 21 |
+
"learning_rate": 2.2e-06,
|
| 22 |
+
"loss": 13.2398,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.028598665395614873,
|
| 27 |
+
"grad_norm": 7679.97412109375,
|
| 28 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 29 |
+
"loss": 11.5969,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.03813155386081983,
|
| 34 |
+
"grad_norm": 1830.318603515625,
|
| 35 |
+
"learning_rate": 5.2e-06,
|
| 36 |
+
"loss": 7.4151,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.047664442326024785,
|
| 41 |
+
"grad_norm": 1004.4345092773438,
|
| 42 |
+
"learning_rate": 7.2e-06,
|
| 43 |
+
"loss": 5.4187,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.057197330791229746,
|
| 48 |
+
"grad_norm": 588.986328125,
|
| 49 |
+
"learning_rate": 9.2e-06,
|
| 50 |
+
"loss": 4.4843,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0667302192564347,
|
| 55 |
+
"grad_norm": 599.4677734375,
|
| 56 |
+
"learning_rate": 1.1200000000000001e-05,
|
| 57 |
+
"loss": 3.5594,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.07626310772163966,
|
| 62 |
+
"grad_norm": 360.55419921875,
|
| 63 |
+
"learning_rate": 1.32e-05,
|
| 64 |
+
"loss": 2.8006,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08579599618684461,
|
| 69 |
+
"grad_norm": 78.43992614746094,
|
| 70 |
+
"learning_rate": 1.52e-05,
|
| 71 |
+
"loss": 1.9745,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.09532888465204957,
|
| 76 |
+
"grad_norm": 49.44837951660156,
|
| 77 |
+
"learning_rate": 1.7199999999999998e-05,
|
| 78 |
+
"loss": 1.1316,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.10486177311725453,
|
| 83 |
+
"grad_norm": 25.379404067993164,
|
| 84 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 85 |
+
"loss": 0.5633,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.11439466158245949,
|
| 90 |
+
"grad_norm": 57.260284423828125,
|
| 91 |
+
"learning_rate": 2.12e-05,
|
| 92 |
+
"loss": 0.3606,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.12392755004766444,
|
| 97 |
+
"grad_norm": 83.0460205078125,
|
| 98 |
+
"learning_rate": 2.32e-05,
|
| 99 |
+
"loss": 0.3603,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.1334604385128694,
|
| 104 |
+
"grad_norm": 91.75605773925781,
|
| 105 |
+
"learning_rate": 2.5200000000000003e-05,
|
| 106 |
+
"loss": 0.2501,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.14299332697807435,
|
| 111 |
+
"grad_norm": 84.04005432128906,
|
| 112 |
+
"learning_rate": 2.7200000000000004e-05,
|
| 113 |
+
"loss": 0.2193,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.15252621544327932,
|
| 118 |
+
"grad_norm": 28.945568084716797,
|
| 119 |
+
"learning_rate": 2.9199999999999998e-05,
|
| 120 |
+
"loss": 0.2095,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.16205910390848427,
|
| 125 |
+
"grad_norm": 9.935993194580078,
|
| 126 |
+
"learning_rate": 3.12e-05,
|
| 127 |
+
"loss": 0.1697,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.17159199237368922,
|
| 132 |
+
"grad_norm": 5.943849563598633,
|
| 133 |
+
"learning_rate": 3.32e-05,
|
| 134 |
+
"loss": 0.1945,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1811248808388942,
|
| 139 |
+
"grad_norm": 13.284988403320312,
|
| 140 |
+
"learning_rate": 3.52e-05,
|
| 141 |
+
"loss": 0.1817,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.19065776930409914,
|
| 146 |
+
"grad_norm": 42.371212005615234,
|
| 147 |
+
"learning_rate": 3.72e-05,
|
| 148 |
+
"loss": 0.5578,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2001906577693041,
|
| 153 |
+
"grad_norm": 17.454591751098633,
|
| 154 |
+
"learning_rate": 3.9200000000000004e-05,
|
| 155 |
+
"loss": 0.2339,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.20972354623450906,
|
| 160 |
+
"grad_norm": 248.91416931152344,
|
| 161 |
+
"learning_rate": 4.12e-05,
|
| 162 |
+
"loss": 0.1741,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.219256434699714,
|
| 167 |
+
"grad_norm": 1.8945605754852295,
|
| 168 |
+
"learning_rate": 4.32e-05,
|
| 169 |
+
"loss": 0.1669,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.22878932316491898,
|
| 174 |
+
"grad_norm": 2.4241456985473633,
|
| 175 |
+
"learning_rate": 4.52e-05,
|
| 176 |
+
"loss": 0.1693,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.23832221163012393,
|
| 181 |
+
"grad_norm": 1.1613476276397705,
|
| 182 |
+
"learning_rate": 4.72e-05,
|
| 183 |
+
"loss": 0.1331,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.24785510009532888,
|
| 188 |
+
"grad_norm": 2.931299924850464,
|
| 189 |
+
"learning_rate": 4.92e-05,
|
| 190 |
+
"loss": 0.1527,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.25738798856053385,
|
| 195 |
+
"grad_norm": 139.0654754638672,
|
| 196 |
+
"learning_rate": 5.1200000000000004e-05,
|
| 197 |
+
"loss": 0.1426,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.2669208770257388,
|
| 202 |
+
"grad_norm": 1.6811221837997437,
|
| 203 |
+
"learning_rate": 5.3200000000000006e-05,
|
| 204 |
+
"loss": 0.1377,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.27645376549094375,
|
| 209 |
+
"grad_norm": 2.773026466369629,
|
| 210 |
+
"learning_rate": 5.520000000000001e-05,
|
| 211 |
+
"loss": 0.1551,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.2859866539561487,
|
| 216 |
+
"grad_norm": 0.9235513210296631,
|
| 217 |
+
"learning_rate": 5.72e-05,
|
| 218 |
+
"loss": 0.1328,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.29551954242135364,
|
| 223 |
+
"grad_norm": 0.634608805179596,
|
| 224 |
+
"learning_rate": 5.92e-05,
|
| 225 |
+
"loss": 0.1203,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.30505243088655865,
|
| 230 |
+
"grad_norm": 1.2987741231918335,
|
| 231 |
+
"learning_rate": 6.12e-05,
|
| 232 |
+
"loss": 0.1127,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.3145853193517636,
|
| 237 |
+
"grad_norm": 1.1649960279464722,
|
| 238 |
+
"learning_rate": 6.32e-05,
|
| 239 |
+
"loss": 0.1228,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.32411820781696854,
|
| 244 |
+
"grad_norm": 1.1517277956008911,
|
| 245 |
+
"learning_rate": 6.52e-05,
|
| 246 |
+
"loss": 0.15,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.3336510962821735,
|
| 251 |
+
"grad_norm": 2.274184465408325,
|
| 252 |
+
"learning_rate": 6.720000000000001e-05,
|
| 253 |
+
"loss": 0.1121,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.34318398474737843,
|
| 258 |
+
"grad_norm": 0.8627381324768066,
|
| 259 |
+
"learning_rate": 6.92e-05,
|
| 260 |
+
"loss": 0.1171,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.35271687321258344,
|
| 265 |
+
"grad_norm": 1.2641953229904175,
|
| 266 |
+
"learning_rate": 7.12e-05,
|
| 267 |
+
"loss": 0.1437,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.3622497616777884,
|
| 272 |
+
"grad_norm": 14.235816955566406,
|
| 273 |
+
"learning_rate": 7.32e-05,
|
| 274 |
+
"loss": 0.1719,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.37178265014299333,
|
| 279 |
+
"grad_norm": 1.2613039016723633,
|
| 280 |
+
"learning_rate": 7.52e-05,
|
| 281 |
+
"loss": 0.1401,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.3813155386081983,
|
| 286 |
+
"grad_norm": 1.125999927520752,
|
| 287 |
+
"learning_rate": 7.72e-05,
|
| 288 |
+
"loss": 0.1176,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.3908484270734032,
|
| 293 |
+
"grad_norm": 2.6700384616851807,
|
| 294 |
+
"learning_rate": 7.920000000000001e-05,
|
| 295 |
+
"loss": 0.116,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.4003813155386082,
|
| 300 |
+
"grad_norm": 0.9148916602134705,
|
| 301 |
+
"learning_rate": 8.120000000000001e-05,
|
| 302 |
+
"loss": 0.1392,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.4099142040038132,
|
| 307 |
+
"grad_norm": 1.3875360488891602,
|
| 308 |
+
"learning_rate": 8.32e-05,
|
| 309 |
+
"loss": 0.168,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.4194470924690181,
|
| 314 |
+
"grad_norm": 0.6264955997467041,
|
| 315 |
+
"learning_rate": 8.52e-05,
|
| 316 |
+
"loss": 0.1484,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.42897998093422307,
|
| 321 |
+
"grad_norm": 1.0963646173477173,
|
| 322 |
+
"learning_rate": 8.72e-05,
|
| 323 |
+
"loss": 0.1457,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.438512869399428,
|
| 328 |
+
"grad_norm": 0.6666226387023926,
|
| 329 |
+
"learning_rate": 8.92e-05,
|
| 330 |
+
"loss": 0.1311,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.44804575786463297,
|
| 335 |
+
"grad_norm": 0.9684863686561584,
|
| 336 |
+
"learning_rate": 9.120000000000001e-05,
|
| 337 |
+
"loss": 0.1428,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.45757864632983797,
|
| 342 |
+
"grad_norm": 0.6623441576957703,
|
| 343 |
+
"learning_rate": 9.320000000000002e-05,
|
| 344 |
+
"loss": 0.1317,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.4671115347950429,
|
| 349 |
+
"grad_norm": 14.834782600402832,
|
| 350 |
+
"learning_rate": 9.52e-05,
|
| 351 |
+
"loss": 0.2165,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.47664442326024786,
|
| 356 |
+
"grad_norm": 2.777594566345215,
|
| 357 |
+
"learning_rate": 9.72e-05,
|
| 358 |
+
"loss": 0.2189,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.4861773117254528,
|
| 363 |
+
"grad_norm": 0.6071160435676575,
|
| 364 |
+
"learning_rate": 9.92e-05,
|
| 365 |
+
"loss": 0.1401,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.49571020019065776,
|
| 370 |
+
"grad_norm": 0.9390038251876831,
|
| 371 |
+
"learning_rate": 9.890710382513662e-05,
|
| 372 |
+
"loss": 0.2272,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.5052430886558628,
|
| 377 |
+
"grad_norm": 0.527932345867157,
|
| 378 |
+
"learning_rate": 9.708561020036431e-05,
|
| 379 |
+
"loss": 0.1311,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.5147759771210677,
|
| 384 |
+
"grad_norm": 0.7770779728889465,
|
| 385 |
+
"learning_rate": 9.5264116575592e-05,
|
| 386 |
+
"loss": 0.1278,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.5243088655862727,
|
| 391 |
+
"grad_norm": 0.5907878279685974,
|
| 392 |
+
"learning_rate": 9.344262295081968e-05,
|
| 393 |
+
"loss": 0.1331,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.5338417540514776,
|
| 398 |
+
"grad_norm": 0.8410826325416565,
|
| 399 |
+
"learning_rate": 9.162112932604736e-05,
|
| 400 |
+
"loss": 0.1274,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.5433746425166825,
|
| 405 |
+
"grad_norm": 7.286756992340088,
|
| 406 |
+
"learning_rate": 8.979963570127505e-05,
|
| 407 |
+
"loss": 0.108,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.5529075309818875,
|
| 412 |
+
"grad_norm": 1.711925745010376,
|
| 413 |
+
"learning_rate": 8.797814207650273e-05,
|
| 414 |
+
"loss": 0.1128,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.5624404194470924,
|
| 419 |
+
"grad_norm": 1.0234737396240234,
|
| 420 |
+
"learning_rate": 8.615664845173043e-05,
|
| 421 |
+
"loss": 0.1405,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.5719733079122974,
|
| 426 |
+
"grad_norm": 12.920223236083984,
|
| 427 |
+
"learning_rate": 8.433515482695811e-05,
|
| 428 |
+
"loss": 0.3122,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.5815061963775023,
|
| 433 |
+
"grad_norm": 1.8921475410461426,
|
| 434 |
+
"learning_rate": 8.25136612021858e-05,
|
| 435 |
+
"loss": 0.1527,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.5910390848427073,
|
| 440 |
+
"grad_norm": 0.22176694869995117,
|
| 441 |
+
"learning_rate": 8.069216757741349e-05,
|
| 442 |
+
"loss": 0.0753,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.6005719733079123,
|
| 447 |
+
"grad_norm": 5.071296691894531,
|
| 448 |
+
"learning_rate": 7.887067395264117e-05,
|
| 449 |
+
"loss": 0.1564,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.6101048617731173,
|
| 454 |
+
"grad_norm": 0.5505660772323608,
|
| 455 |
+
"learning_rate": 7.704918032786885e-05,
|
| 456 |
+
"loss": 0.1145,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.6196377502383222,
|
| 461 |
+
"grad_norm": 10.267545700073242,
|
| 462 |
+
"learning_rate": 7.522768670309654e-05,
|
| 463 |
+
"loss": 0.1342,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.6291706387035272,
|
| 468 |
+
"grad_norm": 0.65236496925354,
|
| 469 |
+
"learning_rate": 7.340619307832422e-05,
|
| 470 |
+
"loss": 0.156,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.6387035271687321,
|
| 475 |
+
"grad_norm": 117.93495178222656,
|
| 476 |
+
"learning_rate": 7.158469945355192e-05,
|
| 477 |
+
"loss": 0.1338,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.6482364156339371,
|
| 482 |
+
"grad_norm": 0.8004688024520874,
|
| 483 |
+
"learning_rate": 6.97632058287796e-05,
|
| 484 |
+
"loss": 0.1319,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.657769304099142,
|
| 489 |
+
"grad_norm": 0.7219289541244507,
|
| 490 |
+
"learning_rate": 6.79417122040073e-05,
|
| 491 |
+
"loss": 0.1115,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.667302192564347,
|
| 496 |
+
"grad_norm": 0.35620227456092834,
|
| 497 |
+
"learning_rate": 6.612021857923498e-05,
|
| 498 |
+
"loss": 0.1471,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.6768350810295519,
|
| 503 |
+
"grad_norm": 1.245010495185852,
|
| 504 |
+
"learning_rate": 6.429872495446266e-05,
|
| 505 |
+
"loss": 0.1107,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.6863679694947569,
|
| 510 |
+
"grad_norm": 0.7738503217697144,
|
| 511 |
+
"learning_rate": 6.247723132969034e-05,
|
| 512 |
+
"loss": 0.1153,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.6959008579599618,
|
| 517 |
+
"grad_norm": 0.8265978693962097,
|
| 518 |
+
"learning_rate": 6.0655737704918034e-05,
|
| 519 |
+
"loss": 0.1653,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.7054337464251669,
|
| 524 |
+
"grad_norm": 0.6407246589660645,
|
| 525 |
+
"learning_rate": 5.8834244080145716e-05,
|
| 526 |
+
"loss": 0.1236,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.7149666348903718,
|
| 531 |
+
"grad_norm": 1.0173193216323853,
|
| 532 |
+
"learning_rate": 5.701275045537341e-05,
|
| 533 |
+
"loss": 0.1167,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.7244995233555768,
|
| 538 |
+
"grad_norm": 0.8038718104362488,
|
| 539 |
+
"learning_rate": 5.519125683060109e-05,
|
| 540 |
+
"loss": 0.1183,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.7340324118207817,
|
| 545 |
+
"grad_norm": 0.5294243097305298,
|
| 546 |
+
"learning_rate": 5.336976320582878e-05,
|
| 547 |
+
"loss": 0.1066,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.7435653002859867,
|
| 552 |
+
"grad_norm": 0.5142996311187744,
|
| 553 |
+
"learning_rate": 5.1548269581056475e-05,
|
| 554 |
+
"loss": 0.1085,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.7530981887511916,
|
| 559 |
+
"grad_norm": 0.5086462497711182,
|
| 560 |
+
"learning_rate": 4.9726775956284156e-05,
|
| 561 |
+
"loss": 0.1077,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.7626310772163966,
|
| 566 |
+
"grad_norm": 0.3155575394630432,
|
| 567 |
+
"learning_rate": 4.7905282331511844e-05,
|
| 568 |
+
"loss": 0.0976,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.7721639656816015,
|
| 573 |
+
"grad_norm": 0.6338363289833069,
|
| 574 |
+
"learning_rate": 4.6083788706739525e-05,
|
| 575 |
+
"loss": 0.104,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.7816968541468065,
|
| 580 |
+
"grad_norm": 0.6568697094917297,
|
| 581 |
+
"learning_rate": 4.426229508196721e-05,
|
| 582 |
+
"loss": 0.1092,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.7912297426120114,
|
| 587 |
+
"grad_norm": 0.5086967349052429,
|
| 588 |
+
"learning_rate": 4.24408014571949e-05,
|
| 589 |
+
"loss": 0.1209,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.8007626310772163,
|
| 594 |
+
"grad_norm": 0.466325044631958,
|
| 595 |
+
"learning_rate": 4.061930783242259e-05,
|
| 596 |
+
"loss": 0.1113,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.8102955195424214,
|
| 601 |
+
"grad_norm": 0.8668030500411987,
|
| 602 |
+
"learning_rate": 3.879781420765027e-05,
|
| 603 |
+
"loss": 0.1015,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.8198284080076264,
|
| 608 |
+
"grad_norm": 0.7336855530738831,
|
| 609 |
+
"learning_rate": 3.697632058287796e-05,
|
| 610 |
+
"loss": 0.0947,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.8293612964728313,
|
| 615 |
+
"grad_norm": 0.8070642948150635,
|
| 616 |
+
"learning_rate": 3.515482695810565e-05,
|
| 617 |
+
"loss": 0.1254,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.8388941849380362,
|
| 622 |
+
"grad_norm": 0.5929535031318665,
|
| 623 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 624 |
+
"loss": 0.0781,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.8484270734032412,
|
| 629 |
+
"grad_norm": 0.585108757019043,
|
| 630 |
+
"learning_rate": 3.1511839708561016e-05,
|
| 631 |
+
"loss": 0.0882,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.8579599618684461,
|
| 636 |
+
"grad_norm": 0.6924892067909241,
|
| 637 |
+
"learning_rate": 2.9690346083788704e-05,
|
| 638 |
+
"loss": 0.1026,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.8674928503336511,
|
| 643 |
+
"grad_norm": 0.968221127986908,
|
| 644 |
+
"learning_rate": 2.7868852459016392e-05,
|
| 645 |
+
"loss": 0.109,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.877025738798856,
|
| 650 |
+
"grad_norm": 0.5783460736274719,
|
| 651 |
+
"learning_rate": 2.604735883424408e-05,
|
| 652 |
+
"loss": 0.0762,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.886558627264061,
|
| 657 |
+
"grad_norm": 0.44456052780151367,
|
| 658 |
+
"learning_rate": 2.422586520947177e-05,
|
| 659 |
+
"loss": 0.0963,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.8960915157292659,
|
| 664 |
+
"grad_norm": 0.49880489706993103,
|
| 665 |
+
"learning_rate": 2.2404371584699453e-05,
|
| 666 |
+
"loss": 0.0988,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.9056244041944709,
|
| 671 |
+
"grad_norm": 0.5427205562591553,
|
| 672 |
+
"learning_rate": 2.058287795992714e-05,
|
| 673 |
+
"loss": 0.0909,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.9151572926596759,
|
| 678 |
+
"grad_norm": 0.5747605562210083,
|
| 679 |
+
"learning_rate": 1.8761384335154826e-05,
|
| 680 |
+
"loss": 0.0983,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.9246901811248809,
|
| 685 |
+
"grad_norm": 0.8023502230644226,
|
| 686 |
+
"learning_rate": 1.6939890710382514e-05,
|
| 687 |
+
"loss": 0.1178,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.9342230695900858,
|
| 692 |
+
"grad_norm": 0.5023919939994812,
|
| 693 |
+
"learning_rate": 1.51183970856102e-05,
|
| 694 |
+
"loss": 0.0837,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.9437559580552908,
|
| 699 |
+
"grad_norm": 0.5053642392158508,
|
| 700 |
+
"learning_rate": 1.3296903460837887e-05,
|
| 701 |
+
"loss": 0.0964,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.9532888465204957,
|
| 706 |
+
"grad_norm": 0.9289111495018005,
|
| 707 |
+
"learning_rate": 1.1475409836065575e-05,
|
| 708 |
+
"loss": 0.1087,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.9628217349857007,
|
| 713 |
+
"grad_norm": 0.6367629766464233,
|
| 714 |
+
"learning_rate": 9.653916211293261e-06,
|
| 715 |
+
"loss": 0.0936,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.9723546234509056,
|
| 720 |
+
"grad_norm": 0.7083348035812378,
|
| 721 |
+
"learning_rate": 7.832422586520947e-06,
|
| 722 |
+
"loss": 0.0905,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.9818875119161106,
|
| 727 |
+
"grad_norm": 0.9046676158905029,
|
| 728 |
+
"learning_rate": 6.010928961748634e-06,
|
| 729 |
+
"loss": 0.1157,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.9914204003813155,
|
| 734 |
+
"grad_norm": 0.5118803977966309,
|
| 735 |
+
"learning_rate": 4.189435336976321e-06,
|
| 736 |
+
"loss": 0.104,
|
| 737 |
+
"step": 1040
|
| 738 |
+
}
|
| 739 |
+
],
|
| 740 |
+
"logging_steps": 10,
|
| 741 |
+
"max_steps": 1049,
|
| 742 |
+
"num_input_tokens_seen": 0,
|
| 743 |
+
"num_train_epochs": 1,
|
| 744 |
+
"save_steps": 5000,
|
| 745 |
+
"stateful_callbacks": {
|
| 746 |
+
"TrainerControl": {
|
| 747 |
+
"args": {
|
| 748 |
+
"should_epoch_stop": false,
|
| 749 |
+
"should_evaluate": false,
|
| 750 |
+
"should_log": false,
|
| 751 |
+
"should_save": true,
|
| 752 |
+
"should_training_stop": true
|
| 753 |
+
},
|
| 754 |
+
"attributes": {}
|
| 755 |
+
}
|
| 756 |
+
},
|
| 757 |
+
"total_flos": 2.394667177161523e+16,
|
| 758 |
+
"train_batch_size": 1,
|
| 759 |
+
"trial_name": null,
|
| 760 |
+
"trial_params": null
|
| 761 |
+
}
|
checkpoint-1049/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b2067065596394160c2f1252866efba0676cb2ab1c658ff89d8bbbba76f5c504
|
| 3 |
+
size 5240
|
checkpoint-2014/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "appvoid/arco-2.1",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"head_dim": 96,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1536,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 4096,
|
| 15 |
+
"max_position_embeddings": 8192,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 16,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 8,
|
| 21 |
+
"pad_token_id": 0,
|
| 22 |
+
"pretraining_tp": 1,
|
| 23 |
+
"rms_norm_eps": 1e-05,
|
| 24 |
+
"rope_scaling": null,
|
| 25 |
+
"rope_theta": 100000,
|
| 26 |
+
"sliding_window": null,
|
| 27 |
+
"tie_word_embeddings": false,
|
| 28 |
+
"torch_dtype": "float32",
|
| 29 |
+
"transformers_version": "4.46.1",
|
| 30 |
+
"unsloth_version": "2024.7",
|
| 31 |
+
"use_cache": true,
|
| 32 |
+
"vocab_size": 32000
|
| 33 |
+
}
|
checkpoint-2014/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.46.1"
|
| 7 |
+
}
|
checkpoint-2014/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8ef37432361d17c0088e9082800949f4279746c6d5535eaa440181b70521e27a
|
| 3 |
+
size 2054379856
|
checkpoint-2014/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f0ffc19b2ad2f62e0453de5935ba7dbd9ecf518454de91e0e5d38d21c175b0e1
|
| 3 |
+
size 4108850298
|
checkpoint-2014/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ff264f99d31b522cc7e2a4eac9d38606d0c58a34c0adc74d71e0ca8b371dc36
|
| 3 |
+
size 14244
|
checkpoint-2014/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc768ed48ad99e4b065bdfee53e82828a0e480d833856e9522bc3f5885018e81
|
| 3 |
+
size 1064
|
checkpoint-2014/trainer_state.json
ADDED
|
@@ -0,0 +1,1440 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 5000,
|
| 6 |
+
"global_step": 2014,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.009930486593843098,
|
| 13 |
+
"grad_norm": Infinity,
|
| 14 |
+
"learning_rate": 0.0,
|
| 15 |
+
"loss": 14.711,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.019860973187686197,
|
| 20 |
+
"grad_norm": 4983.64990234375,
|
| 21 |
+
"learning_rate": 2.0000000000000002e-07,
|
| 22 |
+
"loss": 14.5984,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.029791459781529295,
|
| 27 |
+
"grad_norm": 19263.828125,
|
| 28 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 29 |
+
"loss": 14.0403,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.039721946375372394,
|
| 34 |
+
"grad_norm": 33094.21484375,
|
| 35 |
+
"learning_rate": 6.000000000000001e-07,
|
| 36 |
+
"loss": 12.7662,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04965243296921549,
|
| 41 |
+
"grad_norm": 92821.109375,
|
| 42 |
+
"learning_rate": 8.000000000000001e-07,
|
| 43 |
+
"loss": 11.4273,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.05958291956305859,
|
| 48 |
+
"grad_norm": 8011.658203125,
|
| 49 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 50 |
+
"loss": 10.0555,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.06951340615690169,
|
| 55 |
+
"grad_norm": 6365.61962890625,
|
| 56 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 57 |
+
"loss": 9.0921,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.07944389275074479,
|
| 62 |
+
"grad_norm": 398.2843933105469,
|
| 63 |
+
"learning_rate": 1.4000000000000001e-06,
|
| 64 |
+
"loss": 7.661,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08937437934458789,
|
| 69 |
+
"grad_norm": 158.12139892578125,
|
| 70 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 71 |
+
"loss": 6.3558,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.09930486593843098,
|
| 76 |
+
"grad_norm": 103.36075592041016,
|
| 77 |
+
"learning_rate": 1.8000000000000001e-06,
|
| 78 |
+
"loss": 5.9047,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.10923535253227408,
|
| 83 |
+
"grad_norm": 565.8269653320312,
|
| 84 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 85 |
+
"loss": 5.6256,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.11916583912611718,
|
| 90 |
+
"grad_norm": 414.0867919921875,
|
| 91 |
+
"learning_rate": 2.2e-06,
|
| 92 |
+
"loss": 5.3668,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.12909632571996027,
|
| 97 |
+
"grad_norm": 153.89100646972656,
|
| 98 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 99 |
+
"loss": 5.0483,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.13902681231380337,
|
| 104 |
+
"grad_norm": 149.46444702148438,
|
| 105 |
+
"learning_rate": 2.6e-06,
|
| 106 |
+
"loss": 4.8008,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.14895729890764647,
|
| 111 |
+
"grad_norm": 259.7363586425781,
|
| 112 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 113 |
+
"loss": 4.6652,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.15888778550148958,
|
| 118 |
+
"grad_norm": 575.7271728515625,
|
| 119 |
+
"learning_rate": 3e-06,
|
| 120 |
+
"loss": 4.4912,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.16881827209533268,
|
| 125 |
+
"grad_norm": 316.88525390625,
|
| 126 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 127 |
+
"loss": 4.2479,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.17874875868917578,
|
| 132 |
+
"grad_norm": 310.1224670410156,
|
| 133 |
+
"learning_rate": 3.4000000000000005e-06,
|
| 134 |
+
"loss": 4.0592,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.18867924528301888,
|
| 139 |
+
"grad_norm": 366.346435546875,
|
| 140 |
+
"learning_rate": 3.6000000000000003e-06,
|
| 141 |
+
"loss": 3.8585,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.19860973187686196,
|
| 146 |
+
"grad_norm": 242.16476440429688,
|
| 147 |
+
"learning_rate": 3.8000000000000005e-06,
|
| 148 |
+
"loss": 3.6879,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.20854021847070506,
|
| 153 |
+
"grad_norm": 445.31646728515625,
|
| 154 |
+
"learning_rate": 4.000000000000001e-06,
|
| 155 |
+
"loss": 3.5381,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.21847070506454816,
|
| 160 |
+
"grad_norm": 345.4029235839844,
|
| 161 |
+
"learning_rate": 4.2000000000000004e-06,
|
| 162 |
+
"loss": 3.272,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.22840119165839126,
|
| 167 |
+
"grad_norm": 243.06161499023438,
|
| 168 |
+
"learning_rate": 4.4e-06,
|
| 169 |
+
"loss": 3.0486,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.23833167825223436,
|
| 174 |
+
"grad_norm": 237.27239990234375,
|
| 175 |
+
"learning_rate": 4.600000000000001e-06,
|
| 176 |
+
"loss": 2.7942,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.24826216484607747,
|
| 181 |
+
"grad_norm": 209.28985595703125,
|
| 182 |
+
"learning_rate": 4.800000000000001e-06,
|
| 183 |
+
"loss": 2.5742,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.25819265143992054,
|
| 188 |
+
"grad_norm": 458.2777099609375,
|
| 189 |
+
"learning_rate": 5e-06,
|
| 190 |
+
"loss": 2.5509,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.26812313803376364,
|
| 195 |
+
"grad_norm": 431.66522216796875,
|
| 196 |
+
"learning_rate": 5.2e-06,
|
| 197 |
+
"loss": 2.4819,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.27805362462760674,
|
| 202 |
+
"grad_norm": 314.18792724609375,
|
| 203 |
+
"learning_rate": 5.400000000000001e-06,
|
| 204 |
+
"loss": 2.2044,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.28798411122144985,
|
| 209 |
+
"grad_norm": 366.9264831542969,
|
| 210 |
+
"learning_rate": 5.600000000000001e-06,
|
| 211 |
+
"loss": 1.9759,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.29791459781529295,
|
| 216 |
+
"grad_norm": 148.99159240722656,
|
| 217 |
+
"learning_rate": 5.8e-06,
|
| 218 |
+
"loss": 1.7,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.30784508440913605,
|
| 223 |
+
"grad_norm": 324.4754638671875,
|
| 224 |
+
"learning_rate": 6e-06,
|
| 225 |
+
"loss": 1.5349,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.31777557100297915,
|
| 230 |
+
"grad_norm": 205.2728729248047,
|
| 231 |
+
"learning_rate": 6.200000000000001e-06,
|
| 232 |
+
"loss": 1.4228,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.32770605759682225,
|
| 237 |
+
"grad_norm": 312.1114807128906,
|
| 238 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 239 |
+
"loss": 1.3114,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.33763654419066536,
|
| 244 |
+
"grad_norm": 274.58221435546875,
|
| 245 |
+
"learning_rate": 6.600000000000001e-06,
|
| 246 |
+
"loss": 1.2638,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.34756703078450846,
|
| 251 |
+
"grad_norm": 239.2190399169922,
|
| 252 |
+
"learning_rate": 6.800000000000001e-06,
|
| 253 |
+
"loss": 1.031,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.35749751737835156,
|
| 258 |
+
"grad_norm": 208.7018585205078,
|
| 259 |
+
"learning_rate": 7e-06,
|
| 260 |
+
"loss": 0.92,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.36742800397219466,
|
| 265 |
+
"grad_norm": 237.0693817138672,
|
| 266 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 267 |
+
"loss": 1.665,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.37735849056603776,
|
| 272 |
+
"grad_norm": 208.263671875,
|
| 273 |
+
"learning_rate": 7.4e-06,
|
| 274 |
+
"loss": 0.7744,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3872889771598808,
|
| 279 |
+
"grad_norm": 184.25924682617188,
|
| 280 |
+
"learning_rate": 7.600000000000001e-06,
|
| 281 |
+
"loss": 0.809,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.3972194637537239,
|
| 286 |
+
"grad_norm": 146.69178771972656,
|
| 287 |
+
"learning_rate": 7.800000000000002e-06,
|
| 288 |
+
"loss": 0.7246,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.407149950347567,
|
| 293 |
+
"grad_norm": 207.02853393554688,
|
| 294 |
+
"learning_rate": 8.000000000000001e-06,
|
| 295 |
+
"loss": 0.7616,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.4170804369414101,
|
| 300 |
+
"grad_norm": 146.81492614746094,
|
| 301 |
+
"learning_rate": 8.2e-06,
|
| 302 |
+
"loss": 0.5971,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.4270109235352532,
|
| 307 |
+
"grad_norm": 135.36524963378906,
|
| 308 |
+
"learning_rate": 8.400000000000001e-06,
|
| 309 |
+
"loss": 1.2755,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.4369414101290963,
|
| 314 |
+
"grad_norm": 298.3485107421875,
|
| 315 |
+
"learning_rate": 8.6e-06,
|
| 316 |
+
"loss": 1.6886,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.4468718967229394,
|
| 321 |
+
"grad_norm": 240.9180908203125,
|
| 322 |
+
"learning_rate": 8.8e-06,
|
| 323 |
+
"loss": 0.8591,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.4568023833167825,
|
| 328 |
+
"grad_norm": 152.96194458007812,
|
| 329 |
+
"learning_rate": 9e-06,
|
| 330 |
+
"loss": 0.7211,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.4667328699106256,
|
| 335 |
+
"grad_norm": 5467.83447265625,
|
| 336 |
+
"learning_rate": 9.180000000000002e-06,
|
| 337 |
+
"loss": 1.4453,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.4766633565044687,
|
| 342 |
+
"grad_norm": 249.68312072753906,
|
| 343 |
+
"learning_rate": 9.38e-06,
|
| 344 |
+
"loss": 0.718,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.48659384309831183,
|
| 349 |
+
"grad_norm": 75.28675842285156,
|
| 350 |
+
"learning_rate": 9.58e-06,
|
| 351 |
+
"loss": 0.5454,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.49652432969215493,
|
| 356 |
+
"grad_norm": 98.99024963378906,
|
| 357 |
+
"learning_rate": 9.780000000000001e-06,
|
| 358 |
+
"loss": 0.5422,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.506454816285998,
|
| 363 |
+
"grad_norm": 97.877197265625,
|
| 364 |
+
"learning_rate": 9.980000000000001e-06,
|
| 365 |
+
"loss": 0.5424,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.5163853028798411,
|
| 370 |
+
"grad_norm": 58.41091537475586,
|
| 371 |
+
"learning_rate": 9.940554821664465e-06,
|
| 372 |
+
"loss": 0.4753,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.5263157894736842,
|
| 377 |
+
"grad_norm": 50.91957092285156,
|
| 378 |
+
"learning_rate": 9.87450462351387e-06,
|
| 379 |
+
"loss": 0.4657,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.5362462760675273,
|
| 384 |
+
"grad_norm": 66.574462890625,
|
| 385 |
+
"learning_rate": 9.808454425363277e-06,
|
| 386 |
+
"loss": 0.378,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.5461767626613704,
|
| 391 |
+
"grad_norm": 108.19845581054688,
|
| 392 |
+
"learning_rate": 9.742404227212683e-06,
|
| 393 |
+
"loss": 0.4171,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.5561072492552135,
|
| 398 |
+
"grad_norm": 44.33118438720703,
|
| 399 |
+
"learning_rate": 9.676354029062089e-06,
|
| 400 |
+
"loss": 0.4289,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.5660377358490566,
|
| 405 |
+
"grad_norm": 61.063758850097656,
|
| 406 |
+
"learning_rate": 9.610303830911495e-06,
|
| 407 |
+
"loss": 0.3511,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.5759682224428997,
|
| 412 |
+
"grad_norm": 91.71956634521484,
|
| 413 |
+
"learning_rate": 9.544253632760899e-06,
|
| 414 |
+
"loss": 0.4377,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.5858987090367428,
|
| 419 |
+
"grad_norm": 46.6739501953125,
|
| 420 |
+
"learning_rate": 9.478203434610305e-06,
|
| 421 |
+
"loss": 0.351,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.5958291956305859,
|
| 426 |
+
"grad_norm": 38.2111701965332,
|
| 427 |
+
"learning_rate": 9.41215323645971e-06,
|
| 428 |
+
"loss": 0.3556,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.605759682224429,
|
| 433 |
+
"grad_norm": 61.8499755859375,
|
| 434 |
+
"learning_rate": 9.346103038309115e-06,
|
| 435 |
+
"loss": 0.4027,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.6156901688182721,
|
| 440 |
+
"grad_norm": 219.7929229736328,
|
| 441 |
+
"learning_rate": 9.280052840158521e-06,
|
| 442 |
+
"loss": 0.4932,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.6256206554121152,
|
| 447 |
+
"grad_norm": 52.08991241455078,
|
| 448 |
+
"learning_rate": 9.214002642007927e-06,
|
| 449 |
+
"loss": 0.4588,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.6355511420059583,
|
| 454 |
+
"grad_norm": 28.5477294921875,
|
| 455 |
+
"learning_rate": 9.147952443857333e-06,
|
| 456 |
+
"loss": 0.3302,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.6454816285998014,
|
| 461 |
+
"grad_norm": 24.50048065185547,
|
| 462 |
+
"learning_rate": 9.081902245706737e-06,
|
| 463 |
+
"loss": 0.3144,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.6554121151936445,
|
| 468 |
+
"grad_norm": 59.86844253540039,
|
| 469 |
+
"learning_rate": 9.015852047556145e-06,
|
| 470 |
+
"loss": 0.3919,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.6653426017874876,
|
| 475 |
+
"grad_norm": 22.47603988647461,
|
| 476 |
+
"learning_rate": 8.949801849405549e-06,
|
| 477 |
+
"loss": 0.37,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.6752730883813307,
|
| 482 |
+
"grad_norm": 21.400466918945312,
|
| 483 |
+
"learning_rate": 8.883751651254955e-06,
|
| 484 |
+
"loss": 0.3293,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.6852035749751738,
|
| 489 |
+
"grad_norm": 16.669824600219727,
|
| 490 |
+
"learning_rate": 8.81770145310436e-06,
|
| 491 |
+
"loss": 0.2647,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.6951340615690169,
|
| 496 |
+
"grad_norm": 23.852384567260742,
|
| 497 |
+
"learning_rate": 8.751651254953765e-06,
|
| 498 |
+
"loss": 0.3246,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.70506454816286,
|
| 503 |
+
"grad_norm": 38.244258880615234,
|
| 504 |
+
"learning_rate": 8.685601056803171e-06,
|
| 505 |
+
"loss": 0.385,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.7149950347567031,
|
| 510 |
+
"grad_norm": 29.78006362915039,
|
| 511 |
+
"learning_rate": 8.619550858652577e-06,
|
| 512 |
+
"loss": 0.3071,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.7249255213505462,
|
| 517 |
+
"grad_norm": 24.519458770751953,
|
| 518 |
+
"learning_rate": 8.553500660501983e-06,
|
| 519 |
+
"loss": 0.3429,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.7348560079443893,
|
| 524 |
+
"grad_norm": 25.153671264648438,
|
| 525 |
+
"learning_rate": 8.487450462351387e-06,
|
| 526 |
+
"loss": 0.3843,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.7447864945382324,
|
| 531 |
+
"grad_norm": 27.24303436279297,
|
| 532 |
+
"learning_rate": 8.421400264200793e-06,
|
| 533 |
+
"loss": 0.369,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.7547169811320755,
|
| 538 |
+
"grad_norm": 14.806120872497559,
|
| 539 |
+
"learning_rate": 8.355350066050199e-06,
|
| 540 |
+
"loss": 0.301,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.7646474677259185,
|
| 545 |
+
"grad_norm": 22.724933624267578,
|
| 546 |
+
"learning_rate": 8.289299867899603e-06,
|
| 547 |
+
"loss": 0.3101,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.7745779543197616,
|
| 552 |
+
"grad_norm": 15.89284610748291,
|
| 553 |
+
"learning_rate": 8.22324966974901e-06,
|
| 554 |
+
"loss": 0.3665,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.7845084409136047,
|
| 559 |
+
"grad_norm": 87.12956237792969,
|
| 560 |
+
"learning_rate": 8.157199471598415e-06,
|
| 561 |
+
"loss": 0.4025,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.7944389275074478,
|
| 566 |
+
"grad_norm": 27.03380012512207,
|
| 567 |
+
"learning_rate": 8.091149273447821e-06,
|
| 568 |
+
"loss": 0.2919,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.8043694141012909,
|
| 573 |
+
"grad_norm": 27.622365951538086,
|
| 574 |
+
"learning_rate": 8.025099075297227e-06,
|
| 575 |
+
"loss": 0.3441,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.814299900695134,
|
| 580 |
+
"grad_norm": 32.42015838623047,
|
| 581 |
+
"learning_rate": 7.959048877146633e-06,
|
| 582 |
+
"loss": 0.3307,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.8242303872889771,
|
| 587 |
+
"grad_norm": 16.96957015991211,
|
| 588 |
+
"learning_rate": 7.892998678996037e-06,
|
| 589 |
+
"loss": 0.3224,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.8341608738828202,
|
| 594 |
+
"grad_norm": 12.882108688354492,
|
| 595 |
+
"learning_rate": 7.826948480845443e-06,
|
| 596 |
+
"loss": 0.2992,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.8440913604766633,
|
| 601 |
+
"grad_norm": 17.12813949584961,
|
| 602 |
+
"learning_rate": 7.760898282694849e-06,
|
| 603 |
+
"loss": 0.2783,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.8540218470705064,
|
| 608 |
+
"grad_norm": 57.04343795776367,
|
| 609 |
+
"learning_rate": 7.694848084544253e-06,
|
| 610 |
+
"loss": 0.4445,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.8639523336643495,
|
| 615 |
+
"grad_norm": 17.658185958862305,
|
| 616 |
+
"learning_rate": 7.628797886393659e-06,
|
| 617 |
+
"loss": 0.322,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.8738828202581926,
|
| 622 |
+
"grad_norm": 37.49673080444336,
|
| 623 |
+
"learning_rate": 7.562747688243065e-06,
|
| 624 |
+
"loss": 0.5781,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.8838133068520357,
|
| 629 |
+
"grad_norm": 22.256288528442383,
|
| 630 |
+
"learning_rate": 7.49669749009247e-06,
|
| 631 |
+
"loss": 0.3146,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.8937437934458788,
|
| 636 |
+
"grad_norm": 28.0767765045166,
|
| 637 |
+
"learning_rate": 7.430647291941876e-06,
|
| 638 |
+
"loss": 0.2795,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.9036742800397219,
|
| 643 |
+
"grad_norm": 26.734256744384766,
|
| 644 |
+
"learning_rate": 7.364597093791282e-06,
|
| 645 |
+
"loss": 0.286,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.913604766633565,
|
| 650 |
+
"grad_norm": 16.944440841674805,
|
| 651 |
+
"learning_rate": 7.298546895640688e-06,
|
| 652 |
+
"loss": 0.2396,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.9235352532274081,
|
| 657 |
+
"grad_norm": 11.377096176147461,
|
| 658 |
+
"learning_rate": 7.232496697490093e-06,
|
| 659 |
+
"loss": 0.3234,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.9334657398212513,
|
| 664 |
+
"grad_norm": 21.92275047302246,
|
| 665 |
+
"learning_rate": 7.166446499339499e-06,
|
| 666 |
+
"loss": 0.3054,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.9433962264150944,
|
| 671 |
+
"grad_norm": 11.469223976135254,
|
| 672 |
+
"learning_rate": 7.100396301188904e-06,
|
| 673 |
+
"loss": 0.2805,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.9533267130089375,
|
| 678 |
+
"grad_norm": 21.188453674316406,
|
| 679 |
+
"learning_rate": 7.034346103038309e-06,
|
| 680 |
+
"loss": 0.3169,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.9632571996027806,
|
| 685 |
+
"grad_norm": 14.333311080932617,
|
| 686 |
+
"learning_rate": 6.968295904887715e-06,
|
| 687 |
+
"loss": 0.2381,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.9731876861966237,
|
| 692 |
+
"grad_norm": 10.782018661499023,
|
| 693 |
+
"learning_rate": 6.90224570673712e-06,
|
| 694 |
+
"loss": 0.2914,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.9831181727904668,
|
| 699 |
+
"grad_norm": 12.117927551269531,
|
| 700 |
+
"learning_rate": 6.836195508586526e-06,
|
| 701 |
+
"loss": 0.3186,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.9930486593843099,
|
| 706 |
+
"grad_norm": 15.109565734863281,
|
| 707 |
+
"learning_rate": 6.7701453104359315e-06,
|
| 708 |
+
"loss": 0.2323,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 1.002979145978153,
|
| 713 |
+
"grad_norm": 27.17137336730957,
|
| 714 |
+
"learning_rate": 6.704095112285337e-06,
|
| 715 |
+
"loss": 0.2664,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 1.012909632571996,
|
| 720 |
+
"grad_norm": 17.559146881103516,
|
| 721 |
+
"learning_rate": 6.6380449141347425e-06,
|
| 722 |
+
"loss": 0.3117,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 1.0228401191658392,
|
| 727 |
+
"grad_norm": 18.243568420410156,
|
| 728 |
+
"learning_rate": 6.571994715984148e-06,
|
| 729 |
+
"loss": 0.2746,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.0327706057596822,
|
| 734 |
+
"grad_norm": 19.41541862487793,
|
| 735 |
+
"learning_rate": 6.505944517833554e-06,
|
| 736 |
+
"loss": 0.2757,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.0427010923535254,
|
| 741 |
+
"grad_norm": 23.584136962890625,
|
| 742 |
+
"learning_rate": 6.43989431968296e-06,
|
| 743 |
+
"loss": 0.3311,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.0526315789473684,
|
| 748 |
+
"grad_norm": 8.158661842346191,
|
| 749 |
+
"learning_rate": 6.3738441215323654e-06,
|
| 750 |
+
"loss": 0.2664,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.0625620655412116,
|
| 755 |
+
"grad_norm": 14.265728950500488,
|
| 756 |
+
"learning_rate": 6.3077939233817705e-06,
|
| 757 |
+
"loss": 0.3737,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.0724925521350546,
|
| 762 |
+
"grad_norm": 45.05762481689453,
|
| 763 |
+
"learning_rate": 6.2417437252311765e-06,
|
| 764 |
+
"loss": 0.2805,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.0824230387288978,
|
| 769 |
+
"grad_norm": 26.889284133911133,
|
| 770 |
+
"learning_rate": 6.1756935270805816e-06,
|
| 771 |
+
"loss": 0.3286,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.0923535253227408,
|
| 776 |
+
"grad_norm": 14.202479362487793,
|
| 777 |
+
"learning_rate": 6.1096433289299875e-06,
|
| 778 |
+
"loss": 0.2859,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.102284011916584,
|
| 783 |
+
"grad_norm": 14.509482383728027,
|
| 784 |
+
"learning_rate": 6.043593130779393e-06,
|
| 785 |
+
"loss": 0.2255,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.112214498510427,
|
| 790 |
+
"grad_norm": 17.728469848632812,
|
| 791 |
+
"learning_rate": 5.9775429326287985e-06,
|
| 792 |
+
"loss": 0.2592,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.1221449851042702,
|
| 797 |
+
"grad_norm": 15.352997779846191,
|
| 798 |
+
"learning_rate": 5.911492734478204e-06,
|
| 799 |
+
"loss": 0.2538,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 1.1320754716981132,
|
| 804 |
+
"grad_norm": 12.579967498779297,
|
| 805 |
+
"learning_rate": 5.845442536327609e-06,
|
| 806 |
+
"loss": 0.2803,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.1420059582919564,
|
| 811 |
+
"grad_norm": 9.809431076049805,
|
| 812 |
+
"learning_rate": 5.779392338177015e-06,
|
| 813 |
+
"loss": 0.2874,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 1.1519364448857994,
|
| 818 |
+
"grad_norm": 18.45332145690918,
|
| 819 |
+
"learning_rate": 5.7133421400264215e-06,
|
| 820 |
+
"loss": 0.3388,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.1618669314796426,
|
| 825 |
+
"grad_norm": 15.697205543518066,
|
| 826 |
+
"learning_rate": 5.6472919418758266e-06,
|
| 827 |
+
"loss": 0.6377,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 1.1717974180734856,
|
| 832 |
+
"grad_norm": 245.89466857910156,
|
| 833 |
+
"learning_rate": 5.581241743725232e-06,
|
| 834 |
+
"loss": 0.4898,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.1817279046673286,
|
| 839 |
+
"grad_norm": 13.679891586303711,
|
| 840 |
+
"learning_rate": 5.515191545574638e-06,
|
| 841 |
+
"loss": 0.3288,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.1916583912611718,
|
| 846 |
+
"grad_norm": 16.556507110595703,
|
| 847 |
+
"learning_rate": 5.449141347424043e-06,
|
| 848 |
+
"loss": 0.3122,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.201588877855015,
|
| 853 |
+
"grad_norm": 14.205398559570312,
|
| 854 |
+
"learning_rate": 5.383091149273449e-06,
|
| 855 |
+
"loss": 0.2981,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 1.211519364448858,
|
| 860 |
+
"grad_norm": 12.833841323852539,
|
| 861 |
+
"learning_rate": 5.317040951122854e-06,
|
| 862 |
+
"loss": 0.3071,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 1.221449851042701,
|
| 867 |
+
"grad_norm": 23.980009078979492,
|
| 868 |
+
"learning_rate": 5.250990752972259e-06,
|
| 869 |
+
"loss": 0.2632,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 1.2313803376365442,
|
| 874 |
+
"grad_norm": 17.329069137573242,
|
| 875 |
+
"learning_rate": 5.184940554821665e-06,
|
| 876 |
+
"loss": 0.3039,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 1.2413108242303874,
|
| 881 |
+
"grad_norm": 24.93738555908203,
|
| 882 |
+
"learning_rate": 5.11889035667107e-06,
|
| 883 |
+
"loss": 0.3467,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.2512413108242304,
|
| 888 |
+
"grad_norm": 14.432474136352539,
|
| 889 |
+
"learning_rate": 5.052840158520476e-06,
|
| 890 |
+
"loss": 0.2253,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.2611717974180734,
|
| 895 |
+
"grad_norm": 21.262495040893555,
|
| 896 |
+
"learning_rate": 4.986789960369882e-06,
|
| 897 |
+
"loss": 0.3464,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.2711022840119166,
|
| 902 |
+
"grad_norm": 16.39492416381836,
|
| 903 |
+
"learning_rate": 4.920739762219287e-06,
|
| 904 |
+
"loss": 0.2812,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.2810327706057598,
|
| 909 |
+
"grad_norm": 9.893121719360352,
|
| 910 |
+
"learning_rate": 4.854689564068693e-06,
|
| 911 |
+
"loss": 0.302,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.2909632571996028,
|
| 916 |
+
"grad_norm": 7.951263427734375,
|
| 917 |
+
"learning_rate": 4.788639365918098e-06,
|
| 918 |
+
"loss": 0.4157,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.3008937437934458,
|
| 923 |
+
"grad_norm": 8.177956581115723,
|
| 924 |
+
"learning_rate": 4.722589167767504e-06,
|
| 925 |
+
"loss": 0.206,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.310824230387289,
|
| 930 |
+
"grad_norm": 6.12276029586792,
|
| 931 |
+
"learning_rate": 4.65653896961691e-06,
|
| 932 |
+
"loss": 0.2143,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.320754716981132,
|
| 937 |
+
"grad_norm": 16.6575984954834,
|
| 938 |
+
"learning_rate": 4.590488771466315e-06,
|
| 939 |
+
"loss": 0.2677,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.3306852035749752,
|
| 944 |
+
"grad_norm": 8.128823280334473,
|
| 945 |
+
"learning_rate": 4.52443857331572e-06,
|
| 946 |
+
"loss": 0.2279,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.3406156901688182,
|
| 951 |
+
"grad_norm": 21.915355682373047,
|
| 952 |
+
"learning_rate": 4.458388375165126e-06,
|
| 953 |
+
"loss": 0.2724,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.3505461767626614,
|
| 958 |
+
"grad_norm": 18.645946502685547,
|
| 959 |
+
"learning_rate": 4.392338177014531e-06,
|
| 960 |
+
"loss": 0.2307,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 1.3604766633565044,
|
| 965 |
+
"grad_norm": 12.229573249816895,
|
| 966 |
+
"learning_rate": 4.326287978863937e-06,
|
| 967 |
+
"loss": 0.3043,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.3704071499503476,
|
| 972 |
+
"grad_norm": 21.465803146362305,
|
| 973 |
+
"learning_rate": 4.260237780713343e-06,
|
| 974 |
+
"loss": 0.3025,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1.3803376365441906,
|
| 979 |
+
"grad_norm": 12.59046459197998,
|
| 980 |
+
"learning_rate": 4.194187582562748e-06,
|
| 981 |
+
"loss": 0.3685,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 1.3902681231380338,
|
| 986 |
+
"grad_norm": 20.229293823242188,
|
| 987 |
+
"learning_rate": 4.128137384412154e-06,
|
| 988 |
+
"loss": 0.2627,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.4001986097318768,
|
| 993 |
+
"grad_norm": 5.609831809997559,
|
| 994 |
+
"learning_rate": 4.062087186261559e-06,
|
| 995 |
+
"loss": 0.3011,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.41012909632572,
|
| 1000 |
+
"grad_norm": 13.317000389099121,
|
| 1001 |
+
"learning_rate": 3.996036988110964e-06,
|
| 1002 |
+
"loss": 0.226,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.420059582919563,
|
| 1007 |
+
"grad_norm": 10.011391639709473,
|
| 1008 |
+
"learning_rate": 3.92998678996037e-06,
|
| 1009 |
+
"loss": 0.2295,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.4299900695134062,
|
| 1014 |
+
"grad_norm": 14.472879409790039,
|
| 1015 |
+
"learning_rate": 3.863936591809776e-06,
|
| 1016 |
+
"loss": 0.2577,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.4399205561072492,
|
| 1021 |
+
"grad_norm": 7.8662614822387695,
|
| 1022 |
+
"learning_rate": 3.7978863936591815e-06,
|
| 1023 |
+
"loss": 0.3256,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 1.4498510427010922,
|
| 1028 |
+
"grad_norm": 23.64836883544922,
|
| 1029 |
+
"learning_rate": 3.731836195508587e-06,
|
| 1030 |
+
"loss": 0.336,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.4597815292949354,
|
| 1035 |
+
"grad_norm": 4.431447982788086,
|
| 1036 |
+
"learning_rate": 3.665785997357992e-06,
|
| 1037 |
+
"loss": 0.3114,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 1.4697120158887786,
|
| 1042 |
+
"grad_norm": 37.42151641845703,
|
| 1043 |
+
"learning_rate": 3.5997357992073977e-06,
|
| 1044 |
+
"loss": 0.358,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 1.4796425024826216,
|
| 1049 |
+
"grad_norm": 9.819945335388184,
|
| 1050 |
+
"learning_rate": 3.533685601056803e-06,
|
| 1051 |
+
"loss": 0.2852,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.4895729890764646,
|
| 1056 |
+
"grad_norm": 7.768363952636719,
|
| 1057 |
+
"learning_rate": 3.467635402906209e-06,
|
| 1058 |
+
"loss": 0.2703,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.4995034756703078,
|
| 1063 |
+
"grad_norm": 22.80790901184082,
|
| 1064 |
+
"learning_rate": 3.4015852047556147e-06,
|
| 1065 |
+
"loss": 0.2419,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.509433962264151,
|
| 1070 |
+
"grad_norm": 14.61506175994873,
|
| 1071 |
+
"learning_rate": 3.33553500660502e-06,
|
| 1072 |
+
"loss": 0.2263,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.519364448857994,
|
| 1077 |
+
"grad_norm": 17.9895076751709,
|
| 1078 |
+
"learning_rate": 3.2694848084544257e-06,
|
| 1079 |
+
"loss": 0.3096,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.529294935451837,
|
| 1084 |
+
"grad_norm": 17.43570899963379,
|
| 1085 |
+
"learning_rate": 3.2034346103038312e-06,
|
| 1086 |
+
"loss": 0.2388,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 1.5392254220456802,
|
| 1091 |
+
"grad_norm": 24.771135330200195,
|
| 1092 |
+
"learning_rate": 3.1373844121532363e-06,
|
| 1093 |
+
"loss": 0.2859,
|
| 1094 |
+
"step": 1550
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.5491559086395235,
|
| 1098 |
+
"grad_norm": 14.774213790893555,
|
| 1099 |
+
"learning_rate": 3.0713342140026427e-06,
|
| 1100 |
+
"loss": 0.2478,
|
| 1101 |
+
"step": 1560
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.5590863952333665,
|
| 1105 |
+
"grad_norm": 9.036184310913086,
|
| 1106 |
+
"learning_rate": 3.0052840158520478e-06,
|
| 1107 |
+
"loss": 0.2538,
|
| 1108 |
+
"step": 1570
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 1.5690168818272094,
|
| 1112 |
+
"grad_norm": 15.238768577575684,
|
| 1113 |
+
"learning_rate": 2.9392338177014533e-06,
|
| 1114 |
+
"loss": 0.1995,
|
| 1115 |
+
"step": 1580
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.5789473684210527,
|
| 1119 |
+
"grad_norm": 7.666260242462158,
|
| 1120 |
+
"learning_rate": 2.873183619550859e-06,
|
| 1121 |
+
"loss": 0.2648,
|
| 1122 |
+
"step": 1590
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.5888778550148959,
|
| 1126 |
+
"grad_norm": 14.299159049987793,
|
| 1127 |
+
"learning_rate": 2.8071334214002643e-06,
|
| 1128 |
+
"loss": 0.2817,
|
| 1129 |
+
"step": 1600
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.5988083416087389,
|
| 1133 |
+
"grad_norm": 8.902689933776855,
|
| 1134 |
+
"learning_rate": 2.74108322324967e-06,
|
| 1135 |
+
"loss": 0.6261,
|
| 1136 |
+
"step": 1610
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.6087388282025818,
|
| 1140 |
+
"grad_norm": 11.452303886413574,
|
| 1141 |
+
"learning_rate": 2.6750330250990754e-06,
|
| 1142 |
+
"loss": 0.2743,
|
| 1143 |
+
"step": 1620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.618669314796425,
|
| 1147 |
+
"grad_norm": 8.431117057800293,
|
| 1148 |
+
"learning_rate": 2.6089828269484813e-06,
|
| 1149 |
+
"loss": 0.2472,
|
| 1150 |
+
"step": 1630
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.6285998013902683,
|
| 1154 |
+
"grad_norm": 13.621753692626953,
|
| 1155 |
+
"learning_rate": 2.542932628797887e-06,
|
| 1156 |
+
"loss": 0.3552,
|
| 1157 |
+
"step": 1640
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.6385302879841113,
|
| 1161 |
+
"grad_norm": 28.211891174316406,
|
| 1162 |
+
"learning_rate": 2.4768824306472924e-06,
|
| 1163 |
+
"loss": 0.2211,
|
| 1164 |
+
"step": 1650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.6484607745779543,
|
| 1168 |
+
"grad_norm": 9.634784698486328,
|
| 1169 |
+
"learning_rate": 2.4108322324966975e-06,
|
| 1170 |
+
"loss": 0.7658,
|
| 1171 |
+
"step": 1660
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.6583912611717975,
|
| 1175 |
+
"grad_norm": 10.460068702697754,
|
| 1176 |
+
"learning_rate": 2.3447820343461034e-06,
|
| 1177 |
+
"loss": 0.2257,
|
| 1178 |
+
"step": 1670
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.6683217477656405,
|
| 1182 |
+
"grad_norm": 8.11988639831543,
|
| 1183 |
+
"learning_rate": 2.278731836195509e-06,
|
| 1184 |
+
"loss": 0.2626,
|
| 1185 |
+
"step": 1680
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.6782522343594835,
|
| 1189 |
+
"grad_norm": 12.30396842956543,
|
| 1190 |
+
"learning_rate": 2.2126816380449144e-06,
|
| 1191 |
+
"loss": 0.2174,
|
| 1192 |
+
"step": 1690
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.6881827209533267,
|
| 1196 |
+
"grad_norm": 45.93663787841797,
|
| 1197 |
+
"learning_rate": 2.14663143989432e-06,
|
| 1198 |
+
"loss": 0.3282,
|
| 1199 |
+
"step": 1700
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.6981132075471699,
|
| 1203 |
+
"grad_norm": 15.4673490524292,
|
| 1204 |
+
"learning_rate": 2.0805812417437255e-06,
|
| 1205 |
+
"loss": 0.2126,
|
| 1206 |
+
"step": 1710
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.7080436941410129,
|
| 1210 |
+
"grad_norm": 13.173709869384766,
|
| 1211 |
+
"learning_rate": 2.014531043593131e-06,
|
| 1212 |
+
"loss": 0.2311,
|
| 1213 |
+
"step": 1720
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.7179741807348559,
|
| 1217 |
+
"grad_norm": 17.14373779296875,
|
| 1218 |
+
"learning_rate": 1.9484808454425365e-06,
|
| 1219 |
+
"loss": 0.2538,
|
| 1220 |
+
"step": 1730
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.727904667328699,
|
| 1224 |
+
"grad_norm": 13.05432415008545,
|
| 1225 |
+
"learning_rate": 1.882430647291942e-06,
|
| 1226 |
+
"loss": 0.3432,
|
| 1227 |
+
"step": 1740
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.7378351539225423,
|
| 1231 |
+
"grad_norm": 6.768345832824707,
|
| 1232 |
+
"learning_rate": 1.8163804491413476e-06,
|
| 1233 |
+
"loss": 0.2487,
|
| 1234 |
+
"step": 1750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.7477656405163853,
|
| 1238 |
+
"grad_norm": 7.953887939453125,
|
| 1239 |
+
"learning_rate": 1.750330250990753e-06,
|
| 1240 |
+
"loss": 0.2478,
|
| 1241 |
+
"step": 1760
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.7576961271102283,
|
| 1245 |
+
"grad_norm": 10.380583763122559,
|
| 1246 |
+
"learning_rate": 1.6842800528401588e-06,
|
| 1247 |
+
"loss": 0.2797,
|
| 1248 |
+
"step": 1770
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.7676266137040715,
|
| 1252 |
+
"grad_norm": 18.359220504760742,
|
| 1253 |
+
"learning_rate": 1.6182298546895641e-06,
|
| 1254 |
+
"loss": 0.3262,
|
| 1255 |
+
"step": 1780
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.7775571002979147,
|
| 1259 |
+
"grad_norm": 39.89724349975586,
|
| 1260 |
+
"learning_rate": 1.5521796565389696e-06,
|
| 1261 |
+
"loss": 0.2176,
|
| 1262 |
+
"step": 1790
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.7874875868917577,
|
| 1266 |
+
"grad_norm": 9.103048324584961,
|
| 1267 |
+
"learning_rate": 1.4861294583883754e-06,
|
| 1268 |
+
"loss": 0.2202,
|
| 1269 |
+
"step": 1800
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.7974180734856007,
|
| 1273 |
+
"grad_norm": 22.96454429626465,
|
| 1274 |
+
"learning_rate": 1.420079260237781e-06,
|
| 1275 |
+
"loss": 0.2874,
|
| 1276 |
+
"step": 1810
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.8073485600794439,
|
| 1280 |
+
"grad_norm": 16.194610595703125,
|
| 1281 |
+
"learning_rate": 1.3540290620871862e-06,
|
| 1282 |
+
"loss": 0.168,
|
| 1283 |
+
"step": 1820
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.817279046673287,
|
| 1287 |
+
"grad_norm": 7.695756435394287,
|
| 1288 |
+
"learning_rate": 1.287978863936592e-06,
|
| 1289 |
+
"loss": 0.2298,
|
| 1290 |
+
"step": 1830
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.82720953326713,
|
| 1294 |
+
"grad_norm": 13.192455291748047,
|
| 1295 |
+
"learning_rate": 1.2219286657859975e-06,
|
| 1296 |
+
"loss": 0.2309,
|
| 1297 |
+
"step": 1840
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.837140019860973,
|
| 1301 |
+
"grad_norm": 16.100967407226562,
|
| 1302 |
+
"learning_rate": 1.155878467635403e-06,
|
| 1303 |
+
"loss": 0.2933,
|
| 1304 |
+
"step": 1850
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.8470705064548163,
|
| 1308 |
+
"grad_norm": 4.693732261657715,
|
| 1309 |
+
"learning_rate": 1.0898282694848085e-06,
|
| 1310 |
+
"loss": 0.3162,
|
| 1311 |
+
"step": 1860
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.8570009930486595,
|
| 1315 |
+
"grad_norm": 15.807625770568848,
|
| 1316 |
+
"learning_rate": 1.023778071334214e-06,
|
| 1317 |
+
"loss": 0.2593,
|
| 1318 |
+
"step": 1870
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.8669314796425025,
|
| 1322 |
+
"grad_norm": 8.402117729187012,
|
| 1323 |
+
"learning_rate": 9.577278731836198e-07,
|
| 1324 |
+
"loss": 0.2331,
|
| 1325 |
+
"step": 1880
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.8768619662363455,
|
| 1329 |
+
"grad_norm": 8.832892417907715,
|
| 1330 |
+
"learning_rate": 8.916776750330252e-07,
|
| 1331 |
+
"loss": 0.2713,
|
| 1332 |
+
"step": 1890
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.8867924528301887,
|
| 1336 |
+
"grad_norm": 20.126842498779297,
|
| 1337 |
+
"learning_rate": 8.256274768824307e-07,
|
| 1338 |
+
"loss": 0.3002,
|
| 1339 |
+
"step": 1900
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 1.896722939424032,
|
| 1343 |
+
"grad_norm": 7.1585845947265625,
|
| 1344 |
+
"learning_rate": 7.595772787318363e-07,
|
| 1345 |
+
"loss": 0.2557,
|
| 1346 |
+
"step": 1910
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 1.906653426017875,
|
| 1350 |
+
"grad_norm": 9.247796058654785,
|
| 1351 |
+
"learning_rate": 6.935270805812417e-07,
|
| 1352 |
+
"loss": 0.2147,
|
| 1353 |
+
"step": 1920
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.916583912611718,
|
| 1357 |
+
"grad_norm": 19.770483016967773,
|
| 1358 |
+
"learning_rate": 6.274768824306473e-07,
|
| 1359 |
+
"loss": 0.3213,
|
| 1360 |
+
"step": 1930
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.9265143992055611,
|
| 1364 |
+
"grad_norm": 18.85097885131836,
|
| 1365 |
+
"learning_rate": 5.614266842800529e-07,
|
| 1366 |
+
"loss": 0.3408,
|
| 1367 |
+
"step": 1940
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.9364448857994043,
|
| 1371 |
+
"grad_norm": 24.601226806640625,
|
| 1372 |
+
"learning_rate": 4.953764861294584e-07,
|
| 1373 |
+
"loss": 0.2798,
|
| 1374 |
+
"step": 1950
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.9463753723932473,
|
| 1378 |
+
"grad_norm": 13.348860740661621,
|
| 1379 |
+
"learning_rate": 4.29326287978864e-07,
|
| 1380 |
+
"loss": 0.2631,
|
| 1381 |
+
"step": 1960
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 1.9563058589870903,
|
| 1385 |
+
"grad_norm": 9.122193336486816,
|
| 1386 |
+
"learning_rate": 3.6327608982826953e-07,
|
| 1387 |
+
"loss": 0.2529,
|
| 1388 |
+
"step": 1970
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.9662363455809335,
|
| 1392 |
+
"grad_norm": 20.501909255981445,
|
| 1393 |
+
"learning_rate": 2.9722589167767505e-07,
|
| 1394 |
+
"loss": 0.2518,
|
| 1395 |
+
"step": 1980
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.9761668321747765,
|
| 1399 |
+
"grad_norm": 9.407881736755371,
|
| 1400 |
+
"learning_rate": 2.311756935270806e-07,
|
| 1401 |
+
"loss": 0.2078,
|
| 1402 |
+
"step": 1990
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.9860973187686195,
|
| 1406 |
+
"grad_norm": 22.076322555541992,
|
| 1407 |
+
"learning_rate": 1.6512549537648615e-07,
|
| 1408 |
+
"loss": 0.2104,
|
| 1409 |
+
"step": 2000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 1.9960278053624627,
|
| 1413 |
+
"grad_norm": 18.52741241455078,
|
| 1414 |
+
"learning_rate": 9.907529722589168e-08,
|
| 1415 |
+
"loss": 0.2356,
|
| 1416 |
+
"step": 2010
|
| 1417 |
+
}
|
| 1418 |
+
],
|
| 1419 |
+
"logging_steps": 10,
|
| 1420 |
+
"max_steps": 2014,
|
| 1421 |
+
"num_input_tokens_seen": 0,
|
| 1422 |
+
"num_train_epochs": 2,
|
| 1423 |
+
"save_steps": 5000,
|
| 1424 |
+
"stateful_callbacks": {
|
| 1425 |
+
"TrainerControl": {
|
| 1426 |
+
"args": {
|
| 1427 |
+
"should_epoch_stop": false,
|
| 1428 |
+
"should_evaluate": false,
|
| 1429 |
+
"should_log": false,
|
| 1430 |
+
"should_save": true,
|
| 1431 |
+
"should_training_stop": true
|
| 1432 |
+
},
|
| 1433 |
+
"attributes": {}
|
| 1434 |
+
}
|
| 1435 |
+
},
|
| 1436 |
+
"total_flos": 4.4898226126848e+16,
|
| 1437 |
+
"train_batch_size": 1,
|
| 1438 |
+
"trial_name": null,
|
| 1439 |
+
"trial_params": null
|
| 1440 |
+
}
|
checkpoint-2014/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0db019dabdb6ded328cd1c47d011d738f7f010011b573a21899db2c47422815d
|
| 3 |
+
size 5240
|
checkpoint-3174/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "h2oai/h2o-danube3-500m-chat",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"head_dim": 96,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1536,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 4096,
|
| 15 |
+
"max_position_embeddings": 8192,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 16,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 8,
|
| 21 |
+
"pad_token_id": 0,
|
| 22 |
+
"pretraining_tp": 1,
|
| 23 |
+
"rms_norm_eps": 1e-05,
|
| 24 |
+
"rope_scaling": null,
|
| 25 |
+
"rope_theta": 100000,
|
| 26 |
+
"sliding_window": null,
|
| 27 |
+
"tie_word_embeddings": false,
|
| 28 |
+
"torch_dtype": "float32",
|
| 29 |
+
"transformers_version": "4.46.1",
|
| 30 |
+
"use_cache": true,
|
| 31 |
+
"vocab_size": 32000
|
| 32 |
+
}
|
checkpoint-3174/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.46.1"
|
| 7 |
+
}
|
checkpoint-3174/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e34da29d939d3c71535fdf34b3344bfa5bf1faadb8407db8583d575ec04d8328
|
| 3 |
+
size 2054379856
|
checkpoint-3174/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc016624318efefb8e43e35d19294e059873c3ac0f6c2fc065cac7fc8a4f011f
|
| 3 |
+
size 4108850298
|
checkpoint-3174/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9196a1e708bf24d6abba41cce3f8558820acc3e50f9394c5955e29eb41ffea3d
|
| 3 |
+
size 14244
|
checkpoint-3174/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e88160112b5c4d073ad5a2e1b9be7192702fa17d3708600efbd63ac9708d63c6
|
| 3 |
+
size 1064
|
checkpoint-3174/trainer_state.json
ADDED
|
@@ -0,0 +1,2252 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 5000,
|
| 6 |
+
"global_step": 3174,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.00945179584120983,
|
| 13 |
+
"grad_norm": 13739.59375,
|
| 14 |
+
"learning_rate": 6.000000000000001e-08,
|
| 15 |
+
"loss": 14.4504,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01890359168241966,
|
| 20 |
+
"grad_norm": 2555.72998046875,
|
| 21 |
+
"learning_rate": 2.6e-07,
|
| 22 |
+
"loss": 14.1577,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02835538752362949,
|
| 27 |
+
"grad_norm": 9887.001953125,
|
| 28 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 29 |
+
"loss": 13.151,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.03780718336483932,
|
| 34 |
+
"grad_norm": 9856.8603515625,
|
| 35 |
+
"learning_rate": 5.6e-07,
|
| 36 |
+
"loss": 13.2907,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04725897920604915,
|
| 41 |
+
"grad_norm": 5284.92724609375,
|
| 42 |
+
"learning_rate": 7.6e-07,
|
| 43 |
+
"loss": 12.378,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.05671077504725898,
|
| 48 |
+
"grad_norm": 3841.773681640625,
|
| 49 |
+
"learning_rate": 9.600000000000001e-07,
|
| 50 |
+
"loss": 9.7428,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0661625708884688,
|
| 55 |
+
"grad_norm": 5774.35205078125,
|
| 56 |
+
"learning_rate": 1.1600000000000001e-06,
|
| 57 |
+
"loss": 7.6229,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.07561436672967864,
|
| 62 |
+
"grad_norm": 2795.67626953125,
|
| 63 |
+
"learning_rate": 1.3600000000000001e-06,
|
| 64 |
+
"loss": 6.3023,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08506616257088846,
|
| 69 |
+
"grad_norm": 2087.883056640625,
|
| 70 |
+
"learning_rate": 1.56e-06,
|
| 71 |
+
"loss": 5.7153,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.0945179584120983,
|
| 76 |
+
"grad_norm": 3819.43310546875,
|
| 77 |
+
"learning_rate": 1.76e-06,
|
| 78 |
+
"loss": 5.3327,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.10396975425330812,
|
| 83 |
+
"grad_norm": 4049.7900390625,
|
| 84 |
+
"learning_rate": 1.9600000000000003e-06,
|
| 85 |
+
"loss": 4.9699,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.11342155009451796,
|
| 90 |
+
"grad_norm": 659.6953125,
|
| 91 |
+
"learning_rate": 2.16e-06,
|
| 92 |
+
"loss": 4.6378,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.12287334593572778,
|
| 97 |
+
"grad_norm": 951.3776245117188,
|
| 98 |
+
"learning_rate": 2.3600000000000003e-06,
|
| 99 |
+
"loss": 4.3035,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.1323251417769376,
|
| 104 |
+
"grad_norm": 820.498046875,
|
| 105 |
+
"learning_rate": 2.56e-06,
|
| 106 |
+
"loss": 4.0183,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.14177693761814744,
|
| 111 |
+
"grad_norm": 1631.3897705078125,
|
| 112 |
+
"learning_rate": 2.7600000000000003e-06,
|
| 113 |
+
"loss": 3.7,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.15122873345935728,
|
| 118 |
+
"grad_norm": 1278.76904296875,
|
| 119 |
+
"learning_rate": 2.96e-06,
|
| 120 |
+
"loss": 3.3597,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.16068052930056712,
|
| 125 |
+
"grad_norm": 831.0169067382812,
|
| 126 |
+
"learning_rate": 3.1600000000000002e-06,
|
| 127 |
+
"loss": 3.2386,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.17013232514177692,
|
| 132 |
+
"grad_norm": 1020.6453857421875,
|
| 133 |
+
"learning_rate": 3.3600000000000004e-06,
|
| 134 |
+
"loss": 2.9075,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.17958412098298676,
|
| 139 |
+
"grad_norm": 142.8883819580078,
|
| 140 |
+
"learning_rate": 3.5600000000000002e-06,
|
| 141 |
+
"loss": 2.6557,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.1890359168241966,
|
| 146 |
+
"grad_norm": 1896.7252197265625,
|
| 147 |
+
"learning_rate": 3.7600000000000004e-06,
|
| 148 |
+
"loss": 2.3728,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.19848771266540643,
|
| 153 |
+
"grad_norm": 356.39935302734375,
|
| 154 |
+
"learning_rate": 3.96e-06,
|
| 155 |
+
"loss": 2.1427,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.20793950850661624,
|
| 160 |
+
"grad_norm": 70.32272338867188,
|
| 161 |
+
"learning_rate": 4.16e-06,
|
| 162 |
+
"loss": 1.8594,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.21739130434782608,
|
| 167 |
+
"grad_norm": 184.2218475341797,
|
| 168 |
+
"learning_rate": 4.360000000000001e-06,
|
| 169 |
+
"loss": 1.5881,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.22684310018903592,
|
| 174 |
+
"grad_norm": 260.4848937988281,
|
| 175 |
+
"learning_rate": 4.56e-06,
|
| 176 |
+
"loss": 1.4066,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.23629489603024575,
|
| 181 |
+
"grad_norm": 455.8525085449219,
|
| 182 |
+
"learning_rate": 4.76e-06,
|
| 183 |
+
"loss": 1.0801,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.24574669187145556,
|
| 188 |
+
"grad_norm": 105.61797332763672,
|
| 189 |
+
"learning_rate": 4.960000000000001e-06,
|
| 190 |
+
"loss": 0.927,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.2551984877126654,
|
| 195 |
+
"grad_norm": 68.5740737915039,
|
| 196 |
+
"learning_rate": 5.1600000000000006e-06,
|
| 197 |
+
"loss": 0.7127,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.2646502835538752,
|
| 202 |
+
"grad_norm": 208.01817321777344,
|
| 203 |
+
"learning_rate": 5.36e-06,
|
| 204 |
+
"loss": 0.5245,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.2741020793950851,
|
| 209 |
+
"grad_norm": 69.70160675048828,
|
| 210 |
+
"learning_rate": 5.560000000000001e-06,
|
| 211 |
+
"loss": 0.4756,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.2835538752362949,
|
| 216 |
+
"grad_norm": 42.45973587036133,
|
| 217 |
+
"learning_rate": 5.76e-06,
|
| 218 |
+
"loss": 0.3668,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.29300567107750475,
|
| 223 |
+
"grad_norm": 14.809382438659668,
|
| 224 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 225 |
+
"loss": 0.3122,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.30245746691871456,
|
| 230 |
+
"grad_norm": 106.4734115600586,
|
| 231 |
+
"learning_rate": 6.16e-06,
|
| 232 |
+
"loss": 0.2707,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.31190926275992437,
|
| 237 |
+
"grad_norm": 248.66326904296875,
|
| 238 |
+
"learning_rate": 6.360000000000001e-06,
|
| 239 |
+
"loss": 0.2752,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.32136105860113423,
|
| 244 |
+
"grad_norm": 1199.41064453125,
|
| 245 |
+
"learning_rate": 6.560000000000001e-06,
|
| 246 |
+
"loss": 0.287,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.33081285444234404,
|
| 251 |
+
"grad_norm": 92.04813385009766,
|
| 252 |
+
"learning_rate": 6.760000000000001e-06,
|
| 253 |
+
"loss": 0.2436,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.34026465028355385,
|
| 258 |
+
"grad_norm": 42.634403228759766,
|
| 259 |
+
"learning_rate": 6.96e-06,
|
| 260 |
+
"loss": 0.2851,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.3497164461247637,
|
| 265 |
+
"grad_norm": 6404.83837890625,
|
| 266 |
+
"learning_rate": 7.16e-06,
|
| 267 |
+
"loss": 2.0508,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.3591682419659735,
|
| 272 |
+
"grad_norm": 109.61998748779297,
|
| 273 |
+
"learning_rate": 7.360000000000001e-06,
|
| 274 |
+
"loss": 1.6968,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3686200378071834,
|
| 279 |
+
"grad_norm": 92.08924102783203,
|
| 280 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 281 |
+
"loss": 0.2509,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.3780718336483932,
|
| 286 |
+
"grad_norm": 31.336467742919922,
|
| 287 |
+
"learning_rate": 7.76e-06,
|
| 288 |
+
"loss": 0.2022,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.387523629489603,
|
| 293 |
+
"grad_norm": 21.53646469116211,
|
| 294 |
+
"learning_rate": 7.960000000000002e-06,
|
| 295 |
+
"loss": 0.2043,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.39697542533081287,
|
| 300 |
+
"grad_norm": 103.8397216796875,
|
| 301 |
+
"learning_rate": 8.16e-06,
|
| 302 |
+
"loss": 0.1877,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.4064272211720227,
|
| 307 |
+
"grad_norm": 13.6918306350708,
|
| 308 |
+
"learning_rate": 8.36e-06,
|
| 309 |
+
"loss": 0.1542,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.4158790170132325,
|
| 314 |
+
"grad_norm": 564.9049072265625,
|
| 315 |
+
"learning_rate": 8.560000000000001e-06,
|
| 316 |
+
"loss": 0.1714,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.42533081285444235,
|
| 321 |
+
"grad_norm": 3.5834062099456787,
|
| 322 |
+
"learning_rate": 8.76e-06,
|
| 323 |
+
"loss": 0.1856,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.43478260869565216,
|
| 328 |
+
"grad_norm": 6.6105637550354,
|
| 329 |
+
"learning_rate": 8.96e-06,
|
| 330 |
+
"loss": 0.1631,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.444234404536862,
|
| 335 |
+
"grad_norm": 18.874755859375,
|
| 336 |
+
"learning_rate": 9.16e-06,
|
| 337 |
+
"loss": 0.1556,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.45368620037807184,
|
| 342 |
+
"grad_norm": 4.490022659301758,
|
| 343 |
+
"learning_rate": 9.360000000000002e-06,
|
| 344 |
+
"loss": 0.1196,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.46313799621928164,
|
| 349 |
+
"grad_norm": 647.4920654296875,
|
| 350 |
+
"learning_rate": 9.56e-06,
|
| 351 |
+
"loss": 0.1807,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.4725897920604915,
|
| 356 |
+
"grad_norm": 5.233212471008301,
|
| 357 |
+
"learning_rate": 9.760000000000001e-06,
|
| 358 |
+
"loss": 0.2179,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.4820415879017013,
|
| 363 |
+
"grad_norm": 10.719651222229004,
|
| 364 |
+
"learning_rate": 9.960000000000001e-06,
|
| 365 |
+
"loss": 0.1338,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.4914933837429111,
|
| 370 |
+
"grad_norm": 16.664508819580078,
|
| 371 |
+
"learning_rate": 9.970082273747197e-06,
|
| 372 |
+
"loss": 1.2648,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.500945179584121,
|
| 377 |
+
"grad_norm": 11.62501335144043,
|
| 378 |
+
"learning_rate": 9.93268511593119e-06,
|
| 379 |
+
"loss": 0.1396,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.5103969754253308,
|
| 384 |
+
"grad_norm": 139.4752960205078,
|
| 385 |
+
"learning_rate": 9.895287958115183e-06,
|
| 386 |
+
"loss": 0.1359,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.5198487712665406,
|
| 391 |
+
"grad_norm": 7.7168145179748535,
|
| 392 |
+
"learning_rate": 9.857890800299179e-06,
|
| 393 |
+
"loss": 0.1376,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.5293005671077504,
|
| 398 |
+
"grad_norm": 32.059261322021484,
|
| 399 |
+
"learning_rate": 9.820493642483172e-06,
|
| 400 |
+
"loss": 0.1389,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.5387523629489603,
|
| 405 |
+
"grad_norm": 7.257635116577148,
|
| 406 |
+
"learning_rate": 9.783096484667165e-06,
|
| 407 |
+
"loss": 0.1427,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.5482041587901701,
|
| 412 |
+
"grad_norm": 7.264402389526367,
|
| 413 |
+
"learning_rate": 9.74569932685116e-06,
|
| 414 |
+
"loss": 0.1254,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.55765595463138,
|
| 419 |
+
"grad_norm": 15.918863296508789,
|
| 420 |
+
"learning_rate": 9.708302169035154e-06,
|
| 421 |
+
"loss": 0.1375,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.5671077504725898,
|
| 426 |
+
"grad_norm": 3.28000807762146,
|
| 427 |
+
"learning_rate": 9.670905011219147e-06,
|
| 428 |
+
"loss": 0.1018,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.5765595463137996,
|
| 433 |
+
"grad_norm": 3.7656383514404297,
|
| 434 |
+
"learning_rate": 9.633507853403143e-06,
|
| 435 |
+
"loss": 0.1003,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.5860113421550095,
|
| 440 |
+
"grad_norm": 2.568873882293701,
|
| 441 |
+
"learning_rate": 9.596110695587136e-06,
|
| 442 |
+
"loss": 0.1004,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.5954631379962193,
|
| 447 |
+
"grad_norm": 95.7247314453125,
|
| 448 |
+
"learning_rate": 9.558713537771131e-06,
|
| 449 |
+
"loss": 0.1197,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.6049149338374291,
|
| 454 |
+
"grad_norm": 1.9285805225372314,
|
| 455 |
+
"learning_rate": 9.521316379955125e-06,
|
| 456 |
+
"loss": 0.1489,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.6143667296786389,
|
| 461 |
+
"grad_norm": 2368.34423828125,
|
| 462 |
+
"learning_rate": 9.483919222139118e-06,
|
| 463 |
+
"loss": 0.1099,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.6238185255198487,
|
| 468 |
+
"grad_norm": 127.53451538085938,
|
| 469 |
+
"learning_rate": 9.446522064323113e-06,
|
| 470 |
+
"loss": 0.2462,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.6332703213610587,
|
| 475 |
+
"grad_norm": 3.0768001079559326,
|
| 476 |
+
"learning_rate": 9.409124906507107e-06,
|
| 477 |
+
"loss": 0.8238,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.6427221172022685,
|
| 482 |
+
"grad_norm": 10.293018341064453,
|
| 483 |
+
"learning_rate": 9.3717277486911e-06,
|
| 484 |
+
"loss": 0.1209,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.6521739130434783,
|
| 489 |
+
"grad_norm": 2.587555408477783,
|
| 490 |
+
"learning_rate": 9.334330590875095e-06,
|
| 491 |
+
"loss": 0.1323,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.6616257088846881,
|
| 496 |
+
"grad_norm": 3.1364760398864746,
|
| 497 |
+
"learning_rate": 9.296933433059089e-06,
|
| 498 |
+
"loss": 0.1179,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.6710775047258979,
|
| 503 |
+
"grad_norm": 7.582152366638184,
|
| 504 |
+
"learning_rate": 9.259536275243082e-06,
|
| 505 |
+
"loss": 0.1037,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.6805293005671077,
|
| 510 |
+
"grad_norm": 1070.914794921875,
|
| 511 |
+
"learning_rate": 9.222139117427075e-06,
|
| 512 |
+
"loss": 0.106,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.6899810964083176,
|
| 517 |
+
"grad_norm": 24.98986053466797,
|
| 518 |
+
"learning_rate": 9.18474195961107e-06,
|
| 519 |
+
"loss": 0.1497,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.6994328922495274,
|
| 524 |
+
"grad_norm": 2.334745407104492,
|
| 525 |
+
"learning_rate": 9.147344801795064e-06,
|
| 526 |
+
"loss": 0.0928,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.7088846880907372,
|
| 531 |
+
"grad_norm": 48.43152618408203,
|
| 532 |
+
"learning_rate": 9.109947643979057e-06,
|
| 533 |
+
"loss": 0.0976,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.718336483931947,
|
| 538 |
+
"grad_norm": 12.193922996520996,
|
| 539 |
+
"learning_rate": 9.072550486163052e-06,
|
| 540 |
+
"loss": 0.1007,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.7277882797731569,
|
| 545 |
+
"grad_norm": 65.48289489746094,
|
| 546 |
+
"learning_rate": 9.035153328347046e-06,
|
| 547 |
+
"loss": 0.0961,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.7372400756143668,
|
| 552 |
+
"grad_norm": 9.52373218536377,
|
| 553 |
+
"learning_rate": 8.99775617053104e-06,
|
| 554 |
+
"loss": 0.094,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.7466918714555766,
|
| 559 |
+
"grad_norm": 31.735408782958984,
|
| 560 |
+
"learning_rate": 8.960359012715034e-06,
|
| 561 |
+
"loss": 0.0781,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.7561436672967864,
|
| 566 |
+
"grad_norm": 3.400327682495117,
|
| 567 |
+
"learning_rate": 8.922961854899028e-06,
|
| 568 |
+
"loss": 0.0797,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.7655954631379962,
|
| 573 |
+
"grad_norm": 523.4867553710938,
|
| 574 |
+
"learning_rate": 8.896783844427824e-06,
|
| 575 |
+
"loss": 0.7303,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.775047258979206,
|
| 580 |
+
"grad_norm": 13.026022911071777,
|
| 581 |
+
"learning_rate": 8.859386686611819e-06,
|
| 582 |
+
"loss": 0.1286,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.7844990548204159,
|
| 587 |
+
"grad_norm": 292.6473083496094,
|
| 588 |
+
"learning_rate": 8.821989528795813e-06,
|
| 589 |
+
"loss": 0.1002,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.7939508506616257,
|
| 594 |
+
"grad_norm": 10.678335189819336,
|
| 595 |
+
"learning_rate": 8.784592370979806e-06,
|
| 596 |
+
"loss": 0.5203,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.8034026465028355,
|
| 601 |
+
"grad_norm": 1.7254323959350586,
|
| 602 |
+
"learning_rate": 8.747195213163801e-06,
|
| 603 |
+
"loss": 0.0882,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.8128544423440454,
|
| 608 |
+
"grad_norm": 61.3662223815918,
|
| 609 |
+
"learning_rate": 8.709798055347795e-06,
|
| 610 |
+
"loss": 0.0789,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.8223062381852552,
|
| 615 |
+
"grad_norm": 9330.720703125,
|
| 616 |
+
"learning_rate": 8.672400897531788e-06,
|
| 617 |
+
"loss": 1.3957,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.831758034026465,
|
| 622 |
+
"grad_norm": 8887.767578125,
|
| 623 |
+
"learning_rate": 8.635003739715783e-06,
|
| 624 |
+
"loss": 2.697,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.8412098298676749,
|
| 629 |
+
"grad_norm": 14477.5322265625,
|
| 630 |
+
"learning_rate": 8.597606581899777e-06,
|
| 631 |
+
"loss": 1.646,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.8506616257088847,
|
| 636 |
+
"grad_norm": 2064.92724609375,
|
| 637 |
+
"learning_rate": 8.56020942408377e-06,
|
| 638 |
+
"loss": 1.0418,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.8601134215500945,
|
| 643 |
+
"grad_norm": 53.577369689941406,
|
| 644 |
+
"learning_rate": 8.522812266267765e-06,
|
| 645 |
+
"loss": 0.3661,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.8695652173913043,
|
| 650 |
+
"grad_norm": 11.63904094696045,
|
| 651 |
+
"learning_rate": 8.485415108451759e-06,
|
| 652 |
+
"loss": 0.1646,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.8790170132325141,
|
| 657 |
+
"grad_norm": 313.92529296875,
|
| 658 |
+
"learning_rate": 8.448017950635754e-06,
|
| 659 |
+
"loss": 0.2555,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.888468809073724,
|
| 664 |
+
"grad_norm": 27.883251190185547,
|
| 665 |
+
"learning_rate": 8.410620792819745e-06,
|
| 666 |
+
"loss": 0.0899,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.8979206049149339,
|
| 671 |
+
"grad_norm": 1.2697465419769287,
|
| 672 |
+
"learning_rate": 8.37322363500374e-06,
|
| 673 |
+
"loss": 0.093,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.9073724007561437,
|
| 678 |
+
"grad_norm": 658.786376953125,
|
| 679 |
+
"learning_rate": 8.335826477187734e-06,
|
| 680 |
+
"loss": 0.0879,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.9168241965973535,
|
| 685 |
+
"grad_norm": 12.427763938903809,
|
| 686 |
+
"learning_rate": 8.298429319371727e-06,
|
| 687 |
+
"loss": 0.068,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.9262759924385633,
|
| 692 |
+
"grad_norm": 2670.397216796875,
|
| 693 |
+
"learning_rate": 8.261032161555723e-06,
|
| 694 |
+
"loss": 0.7053,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.9357277882797732,
|
| 699 |
+
"grad_norm": 147.85740661621094,
|
| 700 |
+
"learning_rate": 8.223635003739716e-06,
|
| 701 |
+
"loss": 1.1022,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.945179584120983,
|
| 706 |
+
"grad_norm": 73.45378112792969,
|
| 707 |
+
"learning_rate": 8.18623784592371e-06,
|
| 708 |
+
"loss": 0.3059,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.9546313799621928,
|
| 713 |
+
"grad_norm": 7.717315196990967,
|
| 714 |
+
"learning_rate": 8.148840688107704e-06,
|
| 715 |
+
"loss": 0.1488,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.9640831758034026,
|
| 720 |
+
"grad_norm": 9.288606643676758,
|
| 721 |
+
"learning_rate": 8.111443530291698e-06,
|
| 722 |
+
"loss": 0.1161,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.9735349716446124,
|
| 727 |
+
"grad_norm": 12.23811149597168,
|
| 728 |
+
"learning_rate": 8.074046372475693e-06,
|
| 729 |
+
"loss": 0.0874,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.9829867674858223,
|
| 734 |
+
"grad_norm": 3.111706256866455,
|
| 735 |
+
"learning_rate": 8.036649214659686e-06,
|
| 736 |
+
"loss": 0.1212,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.9924385633270322,
|
| 741 |
+
"grad_norm": 7.152887344360352,
|
| 742 |
+
"learning_rate": 7.99925205684368e-06,
|
| 743 |
+
"loss": 0.1249,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.001890359168242,
|
| 748 |
+
"grad_norm": 1.6718932390213013,
|
| 749 |
+
"learning_rate": 7.961854899027675e-06,
|
| 750 |
+
"loss": 0.1261,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.011342155009452,
|
| 755 |
+
"grad_norm": 1.6129039525985718,
|
| 756 |
+
"learning_rate": 7.924457741211668e-06,
|
| 757 |
+
"loss": 0.0767,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.0207939508506616,
|
| 762 |
+
"grad_norm": 10.476449966430664,
|
| 763 |
+
"learning_rate": 7.887060583395662e-06,
|
| 764 |
+
"loss": 0.0995,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.0302457466918715,
|
| 769 |
+
"grad_norm": 5.244194030761719,
|
| 770 |
+
"learning_rate": 7.849663425579657e-06,
|
| 771 |
+
"loss": 0.1115,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.0396975425330812,
|
| 776 |
+
"grad_norm": 3.1006393432617188,
|
| 777 |
+
"learning_rate": 7.81226626776365e-06,
|
| 778 |
+
"loss": 0.0878,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 1.0491493383742911,
|
| 783 |
+
"grad_norm": 8.808783531188965,
|
| 784 |
+
"learning_rate": 7.774869109947646e-06,
|
| 785 |
+
"loss": 0.105,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.0586011342155008,
|
| 790 |
+
"grad_norm": 5.994989395141602,
|
| 791 |
+
"learning_rate": 7.737471952131639e-06,
|
| 792 |
+
"loss": 0.1088,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.0680529300567108,
|
| 797 |
+
"grad_norm": 2.1118083000183105,
|
| 798 |
+
"learning_rate": 7.700074794315632e-06,
|
| 799 |
+
"loss": 0.1051,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 1.0775047258979207,
|
| 804 |
+
"grad_norm": 5.780063629150391,
|
| 805 |
+
"learning_rate": 7.662677636499628e-06,
|
| 806 |
+
"loss": 0.0792,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.0869565217391304,
|
| 811 |
+
"grad_norm": 0.76413893699646,
|
| 812 |
+
"learning_rate": 7.625280478683621e-06,
|
| 813 |
+
"loss": 0.1034,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 1.0964083175803403,
|
| 818 |
+
"grad_norm": 1.0356695652008057,
|
| 819 |
+
"learning_rate": 7.587883320867615e-06,
|
| 820 |
+
"loss": 0.0698,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 1.10586011342155,
|
| 825 |
+
"grad_norm": 1.114787220954895,
|
| 826 |
+
"learning_rate": 7.550486163051609e-06,
|
| 827 |
+
"loss": 0.0847,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 1.11531190926276,
|
| 832 |
+
"grad_norm": 45.184532165527344,
|
| 833 |
+
"learning_rate": 7.513089005235603e-06,
|
| 834 |
+
"loss": 0.1104,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.1247637051039698,
|
| 839 |
+
"grad_norm": 1.0391864776611328,
|
| 840 |
+
"learning_rate": 7.475691847419597e-06,
|
| 841 |
+
"loss": 0.0829,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.1342155009451795,
|
| 846 |
+
"grad_norm": 5.533662796020508,
|
| 847 |
+
"learning_rate": 7.4382946896035916e-06,
|
| 848 |
+
"loss": 0.0853,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.1436672967863895,
|
| 853 |
+
"grad_norm": 4.5780229568481445,
|
| 854 |
+
"learning_rate": 7.400897531787585e-06,
|
| 855 |
+
"loss": 0.0656,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 1.1531190926275992,
|
| 860 |
+
"grad_norm": 1.4483156204223633,
|
| 861 |
+
"learning_rate": 7.363500373971578e-06,
|
| 862 |
+
"loss": 0.0748,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 1.162570888468809,
|
| 867 |
+
"grad_norm": 1.0575523376464844,
|
| 868 |
+
"learning_rate": 7.326103216155573e-06,
|
| 869 |
+
"loss": 0.0761,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 1.172022684310019,
|
| 874 |
+
"grad_norm": 1.2220410108566284,
|
| 875 |
+
"learning_rate": 7.288706058339566e-06,
|
| 876 |
+
"loss": 0.0885,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 1.1814744801512287,
|
| 881 |
+
"grad_norm": 9.184976577758789,
|
| 882 |
+
"learning_rate": 7.25130890052356e-06,
|
| 883 |
+
"loss": 0.0747,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.1909262759924386,
|
| 888 |
+
"grad_norm": 2.5088539123535156,
|
| 889 |
+
"learning_rate": 7.213911742707555e-06,
|
| 890 |
+
"loss": 0.0878,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.2003780718336483,
|
| 895 |
+
"grad_norm": 0.895652711391449,
|
| 896 |
+
"learning_rate": 7.176514584891548e-06,
|
| 897 |
+
"loss": 0.0648,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.2098298676748582,
|
| 902 |
+
"grad_norm": 3.6389918327331543,
|
| 903 |
+
"learning_rate": 7.139117427075542e-06,
|
| 904 |
+
"loss": 0.069,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.2192816635160681,
|
| 909 |
+
"grad_norm": 1.7553330659866333,
|
| 910 |
+
"learning_rate": 7.101720269259537e-06,
|
| 911 |
+
"loss": 0.085,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.2287334593572778,
|
| 916 |
+
"grad_norm": 1.7131526470184326,
|
| 917 |
+
"learning_rate": 7.064323111443531e-06,
|
| 918 |
+
"loss": 0.0821,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.2381852551984878,
|
| 923 |
+
"grad_norm": 1.947082281112671,
|
| 924 |
+
"learning_rate": 7.026925953627524e-06,
|
| 925 |
+
"loss": 0.079,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.2476370510396975,
|
| 930 |
+
"grad_norm": 0.9147713780403137,
|
| 931 |
+
"learning_rate": 6.989528795811519e-06,
|
| 932 |
+
"loss": 0.0854,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 1.2570888468809074,
|
| 937 |
+
"grad_norm": 1.334012746810913,
|
| 938 |
+
"learning_rate": 6.952131637995513e-06,
|
| 939 |
+
"loss": 0.1571,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.2665406427221173,
|
| 944 |
+
"grad_norm": 1.3452980518341064,
|
| 945 |
+
"learning_rate": 6.914734480179507e-06,
|
| 946 |
+
"loss": 0.0861,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.275992438563327,
|
| 951 |
+
"grad_norm": 0.8652889132499695,
|
| 952 |
+
"learning_rate": 6.877337322363501e-06,
|
| 953 |
+
"loss": 0.0755,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.285444234404537,
|
| 958 |
+
"grad_norm": 0.6752623319625854,
|
| 959 |
+
"learning_rate": 6.839940164547495e-06,
|
| 960 |
+
"loss": 0.0671,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 1.2948960302457466,
|
| 965 |
+
"grad_norm": 2.3604907989501953,
|
| 966 |
+
"learning_rate": 6.802543006731489e-06,
|
| 967 |
+
"loss": 0.0675,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.3043478260869565,
|
| 972 |
+
"grad_norm": 84.10884094238281,
|
| 973 |
+
"learning_rate": 6.7651458489154835e-06,
|
| 974 |
+
"loss": 0.1184,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 1.3137996219281662,
|
| 979 |
+
"grad_norm": 584.1233520507812,
|
| 980 |
+
"learning_rate": 6.727748691099477e-06,
|
| 981 |
+
"loss": 0.1766,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 1.3232514177693762,
|
| 986 |
+
"grad_norm": 5.4759202003479,
|
| 987 |
+
"learning_rate": 6.690351533283471e-06,
|
| 988 |
+
"loss": 0.2643,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.332703213610586,
|
| 993 |
+
"grad_norm": 8.900699615478516,
|
| 994 |
+
"learning_rate": 6.6529543754674655e-06,
|
| 995 |
+
"loss": 0.27,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.3421550094517958,
|
| 1000 |
+
"grad_norm": 31.19775390625,
|
| 1001 |
+
"learning_rate": 6.615557217651459e-06,
|
| 1002 |
+
"loss": 0.1707,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.3516068052930057,
|
| 1007 |
+
"grad_norm": 2.57696270942688,
|
| 1008 |
+
"learning_rate": 6.578160059835453e-06,
|
| 1009 |
+
"loss": 0.1375,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.3610586011342156,
|
| 1014 |
+
"grad_norm": 2.712918996810913,
|
| 1015 |
+
"learning_rate": 6.5407629020194475e-06,
|
| 1016 |
+
"loss": 0.1319,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 1.3705103969754253,
|
| 1021 |
+
"grad_norm": 2.8474884033203125,
|
| 1022 |
+
"learning_rate": 6.503365744203442e-06,
|
| 1023 |
+
"loss": 0.0871,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 1.3799621928166352,
|
| 1028 |
+
"grad_norm": 3.187256336212158,
|
| 1029 |
+
"learning_rate": 6.465968586387435e-06,
|
| 1030 |
+
"loss": 0.0876,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.389413988657845,
|
| 1035 |
+
"grad_norm": 1.1946818828582764,
|
| 1036 |
+
"learning_rate": 6.4285714285714295e-06,
|
| 1037 |
+
"loss": 0.0732,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 1.3988657844990549,
|
| 1042 |
+
"grad_norm": 2.49619722366333,
|
| 1043 |
+
"learning_rate": 6.391174270755424e-06,
|
| 1044 |
+
"loss": 0.0858,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 1.4083175803402646,
|
| 1049 |
+
"grad_norm": 1.9434871673583984,
|
| 1050 |
+
"learning_rate": 6.353777112939416e-06,
|
| 1051 |
+
"loss": 0.1049,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.4177693761814745,
|
| 1056 |
+
"grad_norm": 1.6459031105041504,
|
| 1057 |
+
"learning_rate": 6.316379955123411e-06,
|
| 1058 |
+
"loss": 0.0706,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.4272211720226844,
|
| 1063 |
+
"grad_norm": 2.103151798248291,
|
| 1064 |
+
"learning_rate": 6.278982797307405e-06,
|
| 1065 |
+
"loss": 0.0871,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.436672967863894,
|
| 1070 |
+
"grad_norm": 1.688470721244812,
|
| 1071 |
+
"learning_rate": 6.241585639491398e-06,
|
| 1072 |
+
"loss": 0.0662,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.446124763705104,
|
| 1077 |
+
"grad_norm": 1.6121007204055786,
|
| 1078 |
+
"learning_rate": 6.204188481675393e-06,
|
| 1079 |
+
"loss": 0.0702,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.455576559546314,
|
| 1084 |
+
"grad_norm": 1.3256118297576904,
|
| 1085 |
+
"learning_rate": 6.166791323859387e-06,
|
| 1086 |
+
"loss": 0.0867,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 1.4650283553875236,
|
| 1091 |
+
"grad_norm": 2.683123826980591,
|
| 1092 |
+
"learning_rate": 6.129394166043381e-06,
|
| 1093 |
+
"loss": 0.0658,
|
| 1094 |
+
"step": 1550
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.4744801512287333,
|
| 1098 |
+
"grad_norm": 11.660983085632324,
|
| 1099 |
+
"learning_rate": 6.091997008227375e-06,
|
| 1100 |
+
"loss": 0.0785,
|
| 1101 |
+
"step": 1560
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.4839319470699432,
|
| 1105 |
+
"grad_norm": 1.9295110702514648,
|
| 1106 |
+
"learning_rate": 6.054599850411369e-06,
|
| 1107 |
+
"loss": 0.085,
|
| 1108 |
+
"step": 1570
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 1.4933837429111532,
|
| 1112 |
+
"grad_norm": 0.9210951924324036,
|
| 1113 |
+
"learning_rate": 6.017202692595363e-06,
|
| 1114 |
+
"loss": 0.0974,
|
| 1115 |
+
"step": 1580
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.5028355387523629,
|
| 1119 |
+
"grad_norm": 2.4897303581237793,
|
| 1120 |
+
"learning_rate": 5.979805534779357e-06,
|
| 1121 |
+
"loss": 0.0708,
|
| 1122 |
+
"step": 1590
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.5122873345935728,
|
| 1126 |
+
"grad_norm": 2.9634382724761963,
|
| 1127 |
+
"learning_rate": 5.942408376963351e-06,
|
| 1128 |
+
"loss": 0.075,
|
| 1129 |
+
"step": 1600
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.5217391304347827,
|
| 1133 |
+
"grad_norm": 3.2284462451934814,
|
| 1134 |
+
"learning_rate": 5.905011219147345e-06,
|
| 1135 |
+
"loss": 0.1634,
|
| 1136 |
+
"step": 1610
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.5311909262759924,
|
| 1140 |
+
"grad_norm": 2.838763475418091,
|
| 1141 |
+
"learning_rate": 5.867614061331339e-06,
|
| 1142 |
+
"loss": 0.1089,
|
| 1143 |
+
"step": 1620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.5406427221172023,
|
| 1147 |
+
"grad_norm": 1.9930697679519653,
|
| 1148 |
+
"learning_rate": 5.830216903515334e-06,
|
| 1149 |
+
"loss": 0.0653,
|
| 1150 |
+
"step": 1630
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.5500945179584122,
|
| 1154 |
+
"grad_norm": 1.8953261375427246,
|
| 1155 |
+
"learning_rate": 5.792819745699327e-06,
|
| 1156 |
+
"loss": 0.1015,
|
| 1157 |
+
"step": 1640
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.559546313799622,
|
| 1161 |
+
"grad_norm": 0.7137608528137207,
|
| 1162 |
+
"learning_rate": 5.755422587883321e-06,
|
| 1163 |
+
"loss": 0.0608,
|
| 1164 |
+
"step": 1650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.5689981096408316,
|
| 1168 |
+
"grad_norm": 1.0345617532730103,
|
| 1169 |
+
"learning_rate": 5.718025430067316e-06,
|
| 1170 |
+
"loss": 0.0767,
|
| 1171 |
+
"step": 1660
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.5784499054820416,
|
| 1175 |
+
"grad_norm": 1.0763232707977295,
|
| 1176 |
+
"learning_rate": 5.680628272251309e-06,
|
| 1177 |
+
"loss": 0.089,
|
| 1178 |
+
"step": 1670
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.5879017013232515,
|
| 1182 |
+
"grad_norm": 13.364884376525879,
|
| 1183 |
+
"learning_rate": 5.643231114435303e-06,
|
| 1184 |
+
"loss": 0.0671,
|
| 1185 |
+
"step": 1680
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.5973534971644612,
|
| 1189 |
+
"grad_norm": 3.817563772201538,
|
| 1190 |
+
"learning_rate": 5.605833956619298e-06,
|
| 1191 |
+
"loss": 0.1104,
|
| 1192 |
+
"step": 1690
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.606805293005671,
|
| 1196 |
+
"grad_norm": 1.828808307647705,
|
| 1197 |
+
"learning_rate": 5.568436798803292e-06,
|
| 1198 |
+
"loss": 0.0953,
|
| 1199 |
+
"step": 1700
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.616257088846881,
|
| 1203 |
+
"grad_norm": 3.8507823944091797,
|
| 1204 |
+
"learning_rate": 5.531039640987285e-06,
|
| 1205 |
+
"loss": 0.0838,
|
| 1206 |
+
"step": 1710
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.6257088846880907,
|
| 1210 |
+
"grad_norm": 3.9542222023010254,
|
| 1211 |
+
"learning_rate": 5.49364248317128e-06,
|
| 1212 |
+
"loss": 0.0792,
|
| 1213 |
+
"step": 1720
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.6351606805293004,
|
| 1217 |
+
"grad_norm": 2.827173948287964,
|
| 1218 |
+
"learning_rate": 5.456245325355274e-06,
|
| 1219 |
+
"loss": 0.0869,
|
| 1220 |
+
"step": 1730
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.6446124763705106,
|
| 1224 |
+
"grad_norm": 8.706243515014648,
|
| 1225 |
+
"learning_rate": 5.418848167539268e-06,
|
| 1226 |
+
"loss": 0.0782,
|
| 1227 |
+
"step": 1740
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.6540642722117203,
|
| 1231 |
+
"grad_norm": 2.825563430786133,
|
| 1232 |
+
"learning_rate": 5.381451009723262e-06,
|
| 1233 |
+
"loss": 0.1127,
|
| 1234 |
+
"step": 1750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.66351606805293,
|
| 1238 |
+
"grad_norm": 152.21490478515625,
|
| 1239 |
+
"learning_rate": 5.344053851907256e-06,
|
| 1240 |
+
"loss": 0.1218,
|
| 1241 |
+
"step": 1760
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.6729678638941399,
|
| 1245 |
+
"grad_norm": 6.626716613769531,
|
| 1246 |
+
"learning_rate": 5.306656694091249e-06,
|
| 1247 |
+
"loss": 0.0866,
|
| 1248 |
+
"step": 1770
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.6824196597353498,
|
| 1252 |
+
"grad_norm": 1.9379291534423828,
|
| 1253 |
+
"learning_rate": 5.269259536275243e-06,
|
| 1254 |
+
"loss": 0.0829,
|
| 1255 |
+
"step": 1780
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.6918714555765595,
|
| 1259 |
+
"grad_norm": 18.214658737182617,
|
| 1260 |
+
"learning_rate": 5.231862378459237e-06,
|
| 1261 |
+
"loss": 0.0873,
|
| 1262 |
+
"step": 1790
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.7013232514177694,
|
| 1266 |
+
"grad_norm": 17.1606502532959,
|
| 1267 |
+
"learning_rate": 5.194465220643231e-06,
|
| 1268 |
+
"loss": 0.1232,
|
| 1269 |
+
"step": 1800
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.7107750472589793,
|
| 1273 |
+
"grad_norm": 142.15260314941406,
|
| 1274 |
+
"learning_rate": 5.157068062827225e-06,
|
| 1275 |
+
"loss": 0.1019,
|
| 1276 |
+
"step": 1810
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.720226843100189,
|
| 1280 |
+
"grad_norm": 124.0704345703125,
|
| 1281 |
+
"learning_rate": 5.119670905011219e-06,
|
| 1282 |
+
"loss": 0.4851,
|
| 1283 |
+
"step": 1820
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.7296786389413987,
|
| 1287 |
+
"grad_norm": 51.26754379272461,
|
| 1288 |
+
"learning_rate": 5.082273747195213e-06,
|
| 1289 |
+
"loss": 0.394,
|
| 1290 |
+
"step": 1830
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.7391304347826086,
|
| 1294 |
+
"grad_norm": 56.5087890625,
|
| 1295 |
+
"learning_rate": 5.044876589379208e-06,
|
| 1296 |
+
"loss": 0.1282,
|
| 1297 |
+
"step": 1840
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.7485822306238186,
|
| 1301 |
+
"grad_norm": 25.078203201293945,
|
| 1302 |
+
"learning_rate": 5.007479431563201e-06,
|
| 1303 |
+
"loss": 0.0931,
|
| 1304 |
+
"step": 1850
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.7580340264650283,
|
| 1308 |
+
"grad_norm": 1.4061124324798584,
|
| 1309 |
+
"learning_rate": 4.970082273747195e-06,
|
| 1310 |
+
"loss": 0.0745,
|
| 1311 |
+
"step": 1860
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.7674858223062382,
|
| 1315 |
+
"grad_norm": 16.42283058166504,
|
| 1316 |
+
"learning_rate": 4.93268511593119e-06,
|
| 1317 |
+
"loss": 0.0939,
|
| 1318 |
+
"step": 1870
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.776937618147448,
|
| 1322 |
+
"grad_norm": 2.5964503288269043,
|
| 1323 |
+
"learning_rate": 4.895287958115184e-06,
|
| 1324 |
+
"loss": 0.0865,
|
| 1325 |
+
"step": 1880
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.7863894139886578,
|
| 1329 |
+
"grad_norm": 4.37824821472168,
|
| 1330 |
+
"learning_rate": 4.857890800299177e-06,
|
| 1331 |
+
"loss": 0.0872,
|
| 1332 |
+
"step": 1890
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.7958412098298677,
|
| 1336 |
+
"grad_norm": 3.0010366439819336,
|
| 1337 |
+
"learning_rate": 4.820493642483172e-06,
|
| 1338 |
+
"loss": 0.1027,
|
| 1339 |
+
"step": 1900
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 1.8052930056710776,
|
| 1343 |
+
"grad_norm": 9.359098434448242,
|
| 1344 |
+
"learning_rate": 4.783096484667166e-06,
|
| 1345 |
+
"loss": 0.1273,
|
| 1346 |
+
"step": 1910
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 1.8147448015122873,
|
| 1350 |
+
"grad_norm": 1.1968307495117188,
|
| 1351 |
+
"learning_rate": 4.74569932685116e-06,
|
| 1352 |
+
"loss": 0.1062,
|
| 1353 |
+
"step": 1920
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.824196597353497,
|
| 1357 |
+
"grad_norm": 8.978023529052734,
|
| 1358 |
+
"learning_rate": 4.708302169035154e-06,
|
| 1359 |
+
"loss": 0.073,
|
| 1360 |
+
"step": 1930
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.833648393194707,
|
| 1364 |
+
"grad_norm": 4.380856513977051,
|
| 1365 |
+
"learning_rate": 4.670905011219148e-06,
|
| 1366 |
+
"loss": 0.0807,
|
| 1367 |
+
"step": 1940
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.8431001890359169,
|
| 1371 |
+
"grad_norm": 1.6301466226577759,
|
| 1372 |
+
"learning_rate": 4.633507853403142e-06,
|
| 1373 |
+
"loss": 0.069,
|
| 1374 |
+
"step": 1950
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.8525519848771266,
|
| 1378 |
+
"grad_norm": 1.8033560514450073,
|
| 1379 |
+
"learning_rate": 4.5961106955871356e-06,
|
| 1380 |
+
"loss": 0.1026,
|
| 1381 |
+
"step": 1960
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 1.8620037807183365,
|
| 1385 |
+
"grad_norm": 1.4674009084701538,
|
| 1386 |
+
"learning_rate": 4.55871353777113e-06,
|
| 1387 |
+
"loss": 0.0716,
|
| 1388 |
+
"step": 1970
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.8714555765595464,
|
| 1392 |
+
"grad_norm": 1.8336230516433716,
|
| 1393 |
+
"learning_rate": 4.521316379955123e-06,
|
| 1394 |
+
"loss": 0.0804,
|
| 1395 |
+
"step": 1980
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.8809073724007561,
|
| 1399 |
+
"grad_norm": 1.2159379720687866,
|
| 1400 |
+
"learning_rate": 4.4839192221391176e-06,
|
| 1401 |
+
"loss": 0.097,
|
| 1402 |
+
"step": 1990
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.8903591682419658,
|
| 1406 |
+
"grad_norm": 4.437017440795898,
|
| 1407 |
+
"learning_rate": 4.446522064323112e-06,
|
| 1408 |
+
"loss": 0.0833,
|
| 1409 |
+
"step": 2000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 1.899810964083176,
|
| 1413 |
+
"grad_norm": 1.8714925050735474,
|
| 1414 |
+
"learning_rate": 4.409124906507105e-06,
|
| 1415 |
+
"loss": 0.0764,
|
| 1416 |
+
"step": 2010
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 1.9092627599243857,
|
| 1420 |
+
"grad_norm": 1.631150722503662,
|
| 1421 |
+
"learning_rate": 4.3717277486910996e-06,
|
| 1422 |
+
"loss": 0.0827,
|
| 1423 |
+
"step": 2020
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 1.9187145557655954,
|
| 1427 |
+
"grad_norm": 1.145902395248413,
|
| 1428 |
+
"learning_rate": 4.334330590875094e-06,
|
| 1429 |
+
"loss": 0.0575,
|
| 1430 |
+
"step": 2030
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 1.9281663516068053,
|
| 1434 |
+
"grad_norm": 1.2010490894317627,
|
| 1435 |
+
"learning_rate": 4.296933433059088e-06,
|
| 1436 |
+
"loss": 0.086,
|
| 1437 |
+
"step": 2040
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.9376181474480152,
|
| 1441 |
+
"grad_norm": 1.181751012802124,
|
| 1442 |
+
"learning_rate": 4.2595362752430815e-06,
|
| 1443 |
+
"loss": 0.085,
|
| 1444 |
+
"step": 2050
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 1.947069943289225,
|
| 1448 |
+
"grad_norm": 1.8226746320724487,
|
| 1449 |
+
"learning_rate": 4.222139117427076e-06,
|
| 1450 |
+
"loss": 0.081,
|
| 1451 |
+
"step": 2060
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 1.9565217391304348,
|
| 1455 |
+
"grad_norm": 2.128484010696411,
|
| 1456 |
+
"learning_rate": 4.18474195961107e-06,
|
| 1457 |
+
"loss": 0.0924,
|
| 1458 |
+
"step": 2070
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 1.9659735349716447,
|
| 1462 |
+
"grad_norm": 3.3163959980010986,
|
| 1463 |
+
"learning_rate": 4.147344801795064e-06,
|
| 1464 |
+
"loss": 0.0878,
|
| 1465 |
+
"step": 2080
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 1.9754253308128544,
|
| 1469 |
+
"grad_norm": 3.319842576980591,
|
| 1470 |
+
"learning_rate": 4.109947643979058e-06,
|
| 1471 |
+
"loss": 0.0771,
|
| 1472 |
+
"step": 2090
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 1.9848771266540641,
|
| 1476 |
+
"grad_norm": 1.4262229204177856,
|
| 1477 |
+
"learning_rate": 4.072550486163052e-06,
|
| 1478 |
+
"loss": 0.0701,
|
| 1479 |
+
"step": 2100
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.994328922495274,
|
| 1483 |
+
"grad_norm": 0.9300116896629333,
|
| 1484 |
+
"learning_rate": 4.035153328347046e-06,
|
| 1485 |
+
"loss": 0.071,
|
| 1486 |
+
"step": 2110
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 2.003780718336484,
|
| 1490 |
+
"grad_norm": 3.044982671737671,
|
| 1491 |
+
"learning_rate": 3.99775617053104e-06,
|
| 1492 |
+
"loss": 0.093,
|
| 1493 |
+
"step": 2120
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 2.0132325141776937,
|
| 1497 |
+
"grad_norm": 1.7930673360824585,
|
| 1498 |
+
"learning_rate": 3.960359012715034e-06,
|
| 1499 |
+
"loss": 0.0543,
|
| 1500 |
+
"step": 2130
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 2.022684310018904,
|
| 1504 |
+
"grad_norm": 1.6980185508728027,
|
| 1505 |
+
"learning_rate": 3.9229618548990275e-06,
|
| 1506 |
+
"loss": 0.0611,
|
| 1507 |
+
"step": 2140
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 2.0321361058601135,
|
| 1511 |
+
"grad_norm": 1.0368515253067017,
|
| 1512 |
+
"learning_rate": 3.885564697083022e-06,
|
| 1513 |
+
"loss": 0.047,
|
| 1514 |
+
"step": 2150
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 2.041587901701323,
|
| 1518 |
+
"grad_norm": 3.497382640838623,
|
| 1519 |
+
"learning_rate": 3.848167539267016e-06,
|
| 1520 |
+
"loss": 0.0695,
|
| 1521 |
+
"step": 2160
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 2.051039697542533,
|
| 1525 |
+
"grad_norm": 3.369415521621704,
|
| 1526 |
+
"learning_rate": 3.81077038145101e-06,
|
| 1527 |
+
"loss": 0.1011,
|
| 1528 |
+
"step": 2170
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 2.060491493383743,
|
| 1532 |
+
"grad_norm": 1.3048640489578247,
|
| 1533 |
+
"learning_rate": 3.773373223635004e-06,
|
| 1534 |
+
"loss": 0.1194,
|
| 1535 |
+
"step": 2180
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 2.0699432892249527,
|
| 1539 |
+
"grad_norm": 1.1383615732192993,
|
| 1540 |
+
"learning_rate": 3.735976065818998e-06,
|
| 1541 |
+
"loss": 0.0582,
|
| 1542 |
+
"step": 2190
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 2.0793950850661624,
|
| 1546 |
+
"grad_norm": 1.8952641487121582,
|
| 1547 |
+
"learning_rate": 3.6985789080029923e-06,
|
| 1548 |
+
"loss": 0.0564,
|
| 1549 |
+
"step": 2200
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 2.0888468809073726,
|
| 1553 |
+
"grad_norm": 0.8446295261383057,
|
| 1554 |
+
"learning_rate": 3.661181750186986e-06,
|
| 1555 |
+
"loss": 0.0501,
|
| 1556 |
+
"step": 2210
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 2.0982986767485823,
|
| 1560 |
+
"grad_norm": 1.8624228239059448,
|
| 1561 |
+
"learning_rate": 3.62378459237098e-06,
|
| 1562 |
+
"loss": 0.0668,
|
| 1563 |
+
"step": 2220
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 2.107750472589792,
|
| 1567 |
+
"grad_norm": 0.6371021866798401,
|
| 1568 |
+
"learning_rate": 3.5863874345549743e-06,
|
| 1569 |
+
"loss": 0.1011,
|
| 1570 |
+
"step": 2230
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 2.1172022684310017,
|
| 1574 |
+
"grad_norm": 0.8320020437240601,
|
| 1575 |
+
"learning_rate": 3.548990276738968e-06,
|
| 1576 |
+
"loss": 0.0577,
|
| 1577 |
+
"step": 2240
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 2.126654064272212,
|
| 1581 |
+
"grad_norm": 1.212850570678711,
|
| 1582 |
+
"learning_rate": 3.5115931189229625e-06,
|
| 1583 |
+
"loss": 0.0515,
|
| 1584 |
+
"step": 2250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 2.1361058601134215,
|
| 1588 |
+
"grad_norm": 1.5608317852020264,
|
| 1589 |
+
"learning_rate": 3.474195961106956e-06,
|
| 1590 |
+
"loss": 0.0549,
|
| 1591 |
+
"step": 2260
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 2.145557655954631,
|
| 1595 |
+
"grad_norm": 1.6277782917022705,
|
| 1596 |
+
"learning_rate": 3.4367988032909497e-06,
|
| 1597 |
+
"loss": 0.0775,
|
| 1598 |
+
"step": 2270
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 2.1550094517958414,
|
| 1602 |
+
"grad_norm": 0.8742659091949463,
|
| 1603 |
+
"learning_rate": 3.399401645474944e-06,
|
| 1604 |
+
"loss": 0.0702,
|
| 1605 |
+
"step": 2280
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 2.164461247637051,
|
| 1609 |
+
"grad_norm": 2.2621870040893555,
|
| 1610 |
+
"learning_rate": 3.362004487658938e-06,
|
| 1611 |
+
"loss": 0.3185,
|
| 1612 |
+
"step": 2290
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 2.1739130434782608,
|
| 1616 |
+
"grad_norm": 3.3026936054229736,
|
| 1617 |
+
"learning_rate": 3.324607329842932e-06,
|
| 1618 |
+
"loss": 0.0706,
|
| 1619 |
+
"step": 2300
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 2.183364839319471,
|
| 1623 |
+
"grad_norm": 1.3657283782958984,
|
| 1624 |
+
"learning_rate": 3.287210172026926e-06,
|
| 1625 |
+
"loss": 0.074,
|
| 1626 |
+
"step": 2310
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 2.1928166351606806,
|
| 1630 |
+
"grad_norm": 2.3479464054107666,
|
| 1631 |
+
"learning_rate": 3.2498130142109203e-06,
|
| 1632 |
+
"loss": 0.0859,
|
| 1633 |
+
"step": 2320
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 2.2022684310018903,
|
| 1637 |
+
"grad_norm": 1.2391554117202759,
|
| 1638 |
+
"learning_rate": 3.212415856394914e-06,
|
| 1639 |
+
"loss": 0.0505,
|
| 1640 |
+
"step": 2330
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 2.2117202268431,
|
| 1644 |
+
"grad_norm": 2.2902774810791016,
|
| 1645 |
+
"learning_rate": 3.1750186985789084e-06,
|
| 1646 |
+
"loss": 0.0659,
|
| 1647 |
+
"step": 2340
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 2.22117202268431,
|
| 1651 |
+
"grad_norm": 1.3776533603668213,
|
| 1652 |
+
"learning_rate": 3.1376215407629023e-06,
|
| 1653 |
+
"loss": 0.0591,
|
| 1654 |
+
"step": 2350
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 2.23062381852552,
|
| 1658 |
+
"grad_norm": 2.1550002098083496,
|
| 1659 |
+
"learning_rate": 3.1002243829468966e-06,
|
| 1660 |
+
"loss": 0.0797,
|
| 1661 |
+
"step": 2360
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 2.2400756143667295,
|
| 1665 |
+
"grad_norm": 1.9426394701004028,
|
| 1666 |
+
"learning_rate": 3.0628272251308904e-06,
|
| 1667 |
+
"loss": 0.0986,
|
| 1668 |
+
"step": 2370
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 2.2495274102079397,
|
| 1672 |
+
"grad_norm": 1.2983957529067993,
|
| 1673 |
+
"learning_rate": 3.0254300673148847e-06,
|
| 1674 |
+
"loss": 0.0454,
|
| 1675 |
+
"step": 2380
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 2.2589792060491494,
|
| 1679 |
+
"grad_norm": 1.7152113914489746,
|
| 1680 |
+
"learning_rate": 2.9880329094988786e-06,
|
| 1681 |
+
"loss": 0.0508,
|
| 1682 |
+
"step": 2390
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.268431001890359,
|
| 1686 |
+
"grad_norm": 0.9719337224960327,
|
| 1687 |
+
"learning_rate": 2.950635751682872e-06,
|
| 1688 |
+
"loss": 0.0581,
|
| 1689 |
+
"step": 2400
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 2.2778827977315688,
|
| 1693 |
+
"grad_norm": 0.8925687670707703,
|
| 1694 |
+
"learning_rate": 2.9132385938668663e-06,
|
| 1695 |
+
"loss": 0.082,
|
| 1696 |
+
"step": 2410
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 2.287334593572779,
|
| 1700 |
+
"grad_norm": 4.842050075531006,
|
| 1701 |
+
"learning_rate": 2.87584143605086e-06,
|
| 1702 |
+
"loss": 0.0731,
|
| 1703 |
+
"step": 2420
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 2.2967863894139886,
|
| 1707 |
+
"grad_norm": 1.5382630825042725,
|
| 1708 |
+
"learning_rate": 2.8384442782348544e-06,
|
| 1709 |
+
"loss": 0.0595,
|
| 1710 |
+
"step": 2430
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 2.3062381852551983,
|
| 1714 |
+
"grad_norm": 0.8771998286247253,
|
| 1715 |
+
"learning_rate": 2.8010471204188483e-06,
|
| 1716 |
+
"loss": 0.0584,
|
| 1717 |
+
"step": 2440
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 2.3156899810964084,
|
| 1721 |
+
"grad_norm": 1.2943544387817383,
|
| 1722 |
+
"learning_rate": 2.7636499626028425e-06,
|
| 1723 |
+
"loss": 0.0674,
|
| 1724 |
+
"step": 2450
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 2.325141776937618,
|
| 1728 |
+
"grad_norm": 2.6974918842315674,
|
| 1729 |
+
"learning_rate": 2.7262528047868364e-06,
|
| 1730 |
+
"loss": 0.0606,
|
| 1731 |
+
"step": 2460
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 2.334593572778828,
|
| 1735 |
+
"grad_norm": 1.4208108186721802,
|
| 1736 |
+
"learning_rate": 2.6888556469708307e-06,
|
| 1737 |
+
"loss": 0.0731,
|
| 1738 |
+
"step": 2470
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 2.344045368620038,
|
| 1742 |
+
"grad_norm": 1.310441493988037,
|
| 1743 |
+
"learning_rate": 2.6514584891548245e-06,
|
| 1744 |
+
"loss": 0.1353,
|
| 1745 |
+
"step": 2480
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 2.3534971644612477,
|
| 1749 |
+
"grad_norm": 1.5753083229064941,
|
| 1750 |
+
"learning_rate": 2.6140613313388184e-06,
|
| 1751 |
+
"loss": 0.0648,
|
| 1752 |
+
"step": 2490
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 2.3629489603024574,
|
| 1756 |
+
"grad_norm": 0.9591426253318787,
|
| 1757 |
+
"learning_rate": 2.5766641735228127e-06,
|
| 1758 |
+
"loss": 0.042,
|
| 1759 |
+
"step": 2500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 2.3724007561436675,
|
| 1763 |
+
"grad_norm": 1.4900124073028564,
|
| 1764 |
+
"learning_rate": 2.5392670157068065e-06,
|
| 1765 |
+
"loss": 0.495,
|
| 1766 |
+
"step": 2510
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 2.381852551984877,
|
| 1770 |
+
"grad_norm": 1.3494096994400024,
|
| 1771 |
+
"learning_rate": 2.5018698578908008e-06,
|
| 1772 |
+
"loss": 0.0621,
|
| 1773 |
+
"step": 2520
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 2.391304347826087,
|
| 1777 |
+
"grad_norm": 3.060147762298584,
|
| 1778 |
+
"learning_rate": 2.4644727000747946e-06,
|
| 1779 |
+
"loss": 0.0506,
|
| 1780 |
+
"step": 2530
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 2.4007561436672966,
|
| 1784 |
+
"grad_norm": 4.2206292152404785,
|
| 1785 |
+
"learning_rate": 2.4270755422587885e-06,
|
| 1786 |
+
"loss": 0.0627,
|
| 1787 |
+
"step": 2540
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 2.4102079395085068,
|
| 1791 |
+
"grad_norm": 0.9725443720817566,
|
| 1792 |
+
"learning_rate": 2.3896783844427828e-06,
|
| 1793 |
+
"loss": 0.0461,
|
| 1794 |
+
"step": 2550
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.4196597353497165,
|
| 1798 |
+
"grad_norm": 3.230731964111328,
|
| 1799 |
+
"learning_rate": 2.3522812266267766e-06,
|
| 1800 |
+
"loss": 0.0399,
|
| 1801 |
+
"step": 2560
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 2.429111531190926,
|
| 1805 |
+
"grad_norm": 2.5825319290161133,
|
| 1806 |
+
"learning_rate": 2.3148840688107705e-06,
|
| 1807 |
+
"loss": 0.0595,
|
| 1808 |
+
"step": 2570
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 2.4385633270321363,
|
| 1812 |
+
"grad_norm": 1.2526172399520874,
|
| 1813 |
+
"learning_rate": 2.2774869109947643e-06,
|
| 1814 |
+
"loss": 0.0526,
|
| 1815 |
+
"step": 2580
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 2.448015122873346,
|
| 1819 |
+
"grad_norm": 2.662144184112549,
|
| 1820 |
+
"learning_rate": 2.2400897531787586e-06,
|
| 1821 |
+
"loss": 0.0611,
|
| 1822 |
+
"step": 2590
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 2.4574669187145557,
|
| 1826 |
+
"grad_norm": 1.0309128761291504,
|
| 1827 |
+
"learning_rate": 2.2026925953627525e-06,
|
| 1828 |
+
"loss": 0.0768,
|
| 1829 |
+
"step": 2600
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 2.4669187145557654,
|
| 1833 |
+
"grad_norm": 1.1861653327941895,
|
| 1834 |
+
"learning_rate": 2.1652954375467468e-06,
|
| 1835 |
+
"loss": 0.0611,
|
| 1836 |
+
"step": 2610
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 2.4763705103969755,
|
| 1840 |
+
"grad_norm": 2.0343453884124756,
|
| 1841 |
+
"learning_rate": 2.1278982797307406e-06,
|
| 1842 |
+
"loss": 0.0637,
|
| 1843 |
+
"step": 2620
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 2.4858223062381852,
|
| 1847 |
+
"grad_norm": 1.182011365890503,
|
| 1848 |
+
"learning_rate": 2.090501121914735e-06,
|
| 1849 |
+
"loss": 0.0451,
|
| 1850 |
+
"step": 2630
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 2.495274102079395,
|
| 1854 |
+
"grad_norm": 0.9003073573112488,
|
| 1855 |
+
"learning_rate": 2.0531039640987283e-06,
|
| 1856 |
+
"loss": 0.0526,
|
| 1857 |
+
"step": 2640
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 2.504725897920605,
|
| 1861 |
+
"grad_norm": 2.235811471939087,
|
| 1862 |
+
"learning_rate": 2.0157068062827226e-06,
|
| 1863 |
+
"loss": 0.0753,
|
| 1864 |
+
"step": 2650
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 2.5141776937618148,
|
| 1868 |
+
"grad_norm": 5.556035995483398,
|
| 1869 |
+
"learning_rate": 1.9783096484667165e-06,
|
| 1870 |
+
"loss": 0.1743,
|
| 1871 |
+
"step": 2660
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 2.5236294896030245,
|
| 1875 |
+
"grad_norm": 1.437063455581665,
|
| 1876 |
+
"learning_rate": 1.9409124906507107e-06,
|
| 1877 |
+
"loss": 0.0511,
|
| 1878 |
+
"step": 2670
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 2.5330812854442346,
|
| 1882 |
+
"grad_norm": 0.8690747022628784,
|
| 1883 |
+
"learning_rate": 1.9035153328347048e-06,
|
| 1884 |
+
"loss": 0.0634,
|
| 1885 |
+
"step": 2680
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 2.5425330812854443,
|
| 1889 |
+
"grad_norm": 1.155195713043213,
|
| 1890 |
+
"learning_rate": 1.8661181750186989e-06,
|
| 1891 |
+
"loss": 0.0545,
|
| 1892 |
+
"step": 2690
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 2.551984877126654,
|
| 1896 |
+
"grad_norm": 0.990226686000824,
|
| 1897 |
+
"learning_rate": 1.828721017202693e-06,
|
| 1898 |
+
"loss": 0.0704,
|
| 1899 |
+
"step": 2700
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 2.561436672967864,
|
| 1903 |
+
"grad_norm": 1.3275784254074097,
|
| 1904 |
+
"learning_rate": 1.7913238593866866e-06,
|
| 1905 |
+
"loss": 0.048,
|
| 1906 |
+
"step": 2710
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 2.570888468809074,
|
| 1910 |
+
"grad_norm": 0.962841808795929,
|
| 1911 |
+
"learning_rate": 1.7539267015706806e-06,
|
| 1912 |
+
"loss": 0.0469,
|
| 1913 |
+
"step": 2720
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 2.5803402646502835,
|
| 1917 |
+
"grad_norm": 3.382091522216797,
|
| 1918 |
+
"learning_rate": 1.7165295437546747e-06,
|
| 1919 |
+
"loss": 0.0715,
|
| 1920 |
+
"step": 2730
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 2.5897920604914932,
|
| 1924 |
+
"grad_norm": 1.1871252059936523,
|
| 1925 |
+
"learning_rate": 1.6791323859386688e-06,
|
| 1926 |
+
"loss": 0.0667,
|
| 1927 |
+
"step": 2740
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 2.599243856332703,
|
| 1931 |
+
"grad_norm": 2.9719226360321045,
|
| 1932 |
+
"learning_rate": 1.6417352281226628e-06,
|
| 1933 |
+
"loss": 0.0602,
|
| 1934 |
+
"step": 2750
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 2.608695652173913,
|
| 1938 |
+
"grad_norm": 1.1596277952194214,
|
| 1939 |
+
"learning_rate": 1.604338070306657e-06,
|
| 1940 |
+
"loss": 0.061,
|
| 1941 |
+
"step": 2760
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 2.618147448015123,
|
| 1945 |
+
"grad_norm": 1.5264619588851929,
|
| 1946 |
+
"learning_rate": 1.566940912490651e-06,
|
| 1947 |
+
"loss": 0.055,
|
| 1948 |
+
"step": 2770
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 2.6275992438563325,
|
| 1952 |
+
"grad_norm": 1.292921781539917,
|
| 1953 |
+
"learning_rate": 1.5295437546746446e-06,
|
| 1954 |
+
"loss": 0.0637,
|
| 1955 |
+
"step": 2780
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 2.6370510396975426,
|
| 1959 |
+
"grad_norm": 4.510688781738281,
|
| 1960 |
+
"learning_rate": 1.4921465968586387e-06,
|
| 1961 |
+
"loss": 0.0734,
|
| 1962 |
+
"step": 2790
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.6465028355387523,
|
| 1966 |
+
"grad_norm": 0.9487069249153137,
|
| 1967 |
+
"learning_rate": 1.4547494390426328e-06,
|
| 1968 |
+
"loss": 0.0467,
|
| 1969 |
+
"step": 2800
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 2.655954631379962,
|
| 1973 |
+
"grad_norm": 1.8110833168029785,
|
| 1974 |
+
"learning_rate": 1.4173522812266268e-06,
|
| 1975 |
+
"loss": 0.0479,
|
| 1976 |
+
"step": 2810
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 2.665406427221172,
|
| 1980 |
+
"grad_norm": 1.7220536470413208,
|
| 1981 |
+
"learning_rate": 1.3799551234106209e-06,
|
| 1982 |
+
"loss": 0.0665,
|
| 1983 |
+
"step": 2820
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 2.674858223062382,
|
| 1987 |
+
"grad_norm": 2.3197669982910156,
|
| 1988 |
+
"learning_rate": 1.342557965594615e-06,
|
| 1989 |
+
"loss": 0.0559,
|
| 1990 |
+
"step": 2830
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 2.6843100189035916,
|
| 1994 |
+
"grad_norm": 1.2168952226638794,
|
| 1995 |
+
"learning_rate": 1.305160807778609e-06,
|
| 1996 |
+
"loss": 0.0597,
|
| 1997 |
+
"step": 2840
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 2.6937618147448017,
|
| 2001 |
+
"grad_norm": 2.3563144207000732,
|
| 2002 |
+
"learning_rate": 1.2677636499626029e-06,
|
| 2003 |
+
"loss": 0.0698,
|
| 2004 |
+
"step": 2850
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 2.7032136105860114,
|
| 2008 |
+
"grad_norm": 1.7858084440231323,
|
| 2009 |
+
"learning_rate": 1.230366492146597e-06,
|
| 2010 |
+
"loss": 0.0493,
|
| 2011 |
+
"step": 2860
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 2.712665406427221,
|
| 2015 |
+
"grad_norm": 0.999500036239624,
|
| 2016 |
+
"learning_rate": 1.192969334330591e-06,
|
| 2017 |
+
"loss": 0.0555,
|
| 2018 |
+
"step": 2870
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 2.7221172022684312,
|
| 2022 |
+
"grad_norm": 2.3788414001464844,
|
| 2023 |
+
"learning_rate": 1.1555721765145849e-06,
|
| 2024 |
+
"loss": 0.0466,
|
| 2025 |
+
"step": 2880
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 2.731568998109641,
|
| 2029 |
+
"grad_norm": 1.2717732191085815,
|
| 2030 |
+
"learning_rate": 1.118175018698579e-06,
|
| 2031 |
+
"loss": 0.0573,
|
| 2032 |
+
"step": 2890
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 2.7410207939508506,
|
| 2036 |
+
"grad_norm": 1.2164254188537598,
|
| 2037 |
+
"learning_rate": 1.080777860882573e-06,
|
| 2038 |
+
"loss": 0.0613,
|
| 2039 |
+
"step": 2900
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 2.7504725897920603,
|
| 2043 |
+
"grad_norm": 1.483476996421814,
|
| 2044 |
+
"learning_rate": 1.043380703066567e-06,
|
| 2045 |
+
"loss": 0.0621,
|
| 2046 |
+
"step": 2910
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 2.7599243856332705,
|
| 2050 |
+
"grad_norm": 0.919227659702301,
|
| 2051 |
+
"learning_rate": 1.0059835452505611e-06,
|
| 2052 |
+
"loss": 0.0551,
|
| 2053 |
+
"step": 2920
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 2.76937618147448,
|
| 2057 |
+
"grad_norm": 1.3688384294509888,
|
| 2058 |
+
"learning_rate": 9.685863874345552e-07,
|
| 2059 |
+
"loss": 0.0457,
|
| 2060 |
+
"step": 2930
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 2.77882797731569,
|
| 2064 |
+
"grad_norm": 2.6967413425445557,
|
| 2065 |
+
"learning_rate": 9.311892296185491e-07,
|
| 2066 |
+
"loss": 0.0522,
|
| 2067 |
+
"step": 2940
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 2.7882797731568996,
|
| 2071 |
+
"grad_norm": 1.768357276916504,
|
| 2072 |
+
"learning_rate": 8.93792071802543e-07,
|
| 2073 |
+
"loss": 0.0507,
|
| 2074 |
+
"step": 2950
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 2.7977315689981097,
|
| 2078 |
+
"grad_norm": 1.3043689727783203,
|
| 2079 |
+
"learning_rate": 8.563949139865371e-07,
|
| 2080 |
+
"loss": 0.0496,
|
| 2081 |
+
"step": 2960
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 2.8071833648393194,
|
| 2085 |
+
"grad_norm": 1.3980363607406616,
|
| 2086 |
+
"learning_rate": 8.189977561705311e-07,
|
| 2087 |
+
"loss": 0.06,
|
| 2088 |
+
"step": 2970
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 2.816635160680529,
|
| 2092 |
+
"grad_norm": 2.036473512649536,
|
| 2093 |
+
"learning_rate": 7.816005983545251e-07,
|
| 2094 |
+
"loss": 0.0519,
|
| 2095 |
+
"step": 2980
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 2.8260869565217392,
|
| 2099 |
+
"grad_norm": 1.1731027364730835,
|
| 2100 |
+
"learning_rate": 7.442034405385192e-07,
|
| 2101 |
+
"loss": 0.0537,
|
| 2102 |
+
"step": 2990
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 2.835538752362949,
|
| 2106 |
+
"grad_norm": 1.2910258769989014,
|
| 2107 |
+
"learning_rate": 7.068062827225131e-07,
|
| 2108 |
+
"loss": 0.0439,
|
| 2109 |
+
"step": 3000
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 2.8449905482041586,
|
| 2113 |
+
"grad_norm": 0.6368816494941711,
|
| 2114 |
+
"learning_rate": 6.694091249065071e-07,
|
| 2115 |
+
"loss": 0.0541,
|
| 2116 |
+
"step": 3010
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 2.854442344045369,
|
| 2120 |
+
"grad_norm": 1.0319303274154663,
|
| 2121 |
+
"learning_rate": 6.320119670905012e-07,
|
| 2122 |
+
"loss": 0.0337,
|
| 2123 |
+
"step": 3020
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 2.8638941398865785,
|
| 2127 |
+
"grad_norm": 1.931760311126709,
|
| 2128 |
+
"learning_rate": 5.946148092744952e-07,
|
| 2129 |
+
"loss": 0.0555,
|
| 2130 |
+
"step": 3030
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 2.873345935727788,
|
| 2134 |
+
"grad_norm": 1.5144755840301514,
|
| 2135 |
+
"learning_rate": 5.572176514584892e-07,
|
| 2136 |
+
"loss": 0.0555,
|
| 2137 |
+
"step": 3040
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 2.8827977315689983,
|
| 2141 |
+
"grad_norm": 9.888813972473145,
|
| 2142 |
+
"learning_rate": 5.198204936424832e-07,
|
| 2143 |
+
"loss": 0.0606,
|
| 2144 |
+
"step": 3050
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 2.892249527410208,
|
| 2148 |
+
"grad_norm": 10.548643112182617,
|
| 2149 |
+
"learning_rate": 4.824233358264772e-07,
|
| 2150 |
+
"loss": 0.064,
|
| 2151 |
+
"step": 3060
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 2.9017013232514177,
|
| 2155 |
+
"grad_norm": 1.567958950996399,
|
| 2156 |
+
"learning_rate": 4.4502617801047125e-07,
|
| 2157 |
+
"loss": 0.0644,
|
| 2158 |
+
"step": 3070
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 2.911153119092628,
|
| 2162 |
+
"grad_norm": 2.924102783203125,
|
| 2163 |
+
"learning_rate": 4.0762902019446526e-07,
|
| 2164 |
+
"loss": 0.0584,
|
| 2165 |
+
"step": 3080
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 2.9206049149338376,
|
| 2169 |
+
"grad_norm": 2.3769917488098145,
|
| 2170 |
+
"learning_rate": 3.702318623784593e-07,
|
| 2171 |
+
"loss": 0.054,
|
| 2172 |
+
"step": 3090
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 2.9300567107750473,
|
| 2176 |
+
"grad_norm": 2.570713996887207,
|
| 2177 |
+
"learning_rate": 3.328347045624533e-07,
|
| 2178 |
+
"loss": 0.0709,
|
| 2179 |
+
"step": 3100
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 2.939508506616257,
|
| 2183 |
+
"grad_norm": 1.1489148139953613,
|
| 2184 |
+
"learning_rate": 2.954375467464473e-07,
|
| 2185 |
+
"loss": 0.0605,
|
| 2186 |
+
"step": 3110
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 2.9489603024574667,
|
| 2190 |
+
"grad_norm": 1.264321208000183,
|
| 2191 |
+
"learning_rate": 2.580403889304413e-07,
|
| 2192 |
+
"loss": 0.0684,
|
| 2193 |
+
"step": 3120
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 2.958412098298677,
|
| 2197 |
+
"grad_norm": 1.6857093572616577,
|
| 2198 |
+
"learning_rate": 2.2064323111443533e-07,
|
| 2199 |
+
"loss": 0.0579,
|
| 2200 |
+
"step": 3130
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 2.9678638941398865,
|
| 2204 |
+
"grad_norm": 2.269068717956543,
|
| 2205 |
+
"learning_rate": 1.8324607329842932e-07,
|
| 2206 |
+
"loss": 0.0606,
|
| 2207 |
+
"step": 3140
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 2.977315689981096,
|
| 2211 |
+
"grad_norm": 1.0288152694702148,
|
| 2212 |
+
"learning_rate": 1.4584891548242336e-07,
|
| 2213 |
+
"loss": 0.0577,
|
| 2214 |
+
"step": 3150
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 2.9867674858223063,
|
| 2218 |
+
"grad_norm": 0.940092146396637,
|
| 2219 |
+
"learning_rate": 1.0845175766641736e-07,
|
| 2220 |
+
"loss": 0.0412,
|
| 2221 |
+
"step": 3160
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 2.996219281663516,
|
| 2225 |
+
"grad_norm": 1.5147079229354858,
|
| 2226 |
+
"learning_rate": 7.105459985041138e-08,
|
| 2227 |
+
"loss": 0.0586,
|
| 2228 |
+
"step": 3170
|
| 2229 |
+
}
|
| 2230 |
+
],
|
| 2231 |
+
"logging_steps": 10,
|
| 2232 |
+
"max_steps": 3174,
|
| 2233 |
+
"num_input_tokens_seen": 0,
|
| 2234 |
+
"num_train_epochs": 3,
|
| 2235 |
+
"save_steps": 5000,
|
| 2236 |
+
"stateful_callbacks": {
|
| 2237 |
+
"TrainerControl": {
|
| 2238 |
+
"args": {
|
| 2239 |
+
"should_epoch_stop": false,
|
| 2240 |
+
"should_evaluate": false,
|
| 2241 |
+
"should_log": false,
|
| 2242 |
+
"should_save": true,
|
| 2243 |
+
"should_training_stop": true
|
| 2244 |
+
},
|
| 2245 |
+
"attributes": {}
|
| 2246 |
+
}
|
| 2247 |
+
},
|
| 2248 |
+
"total_flos": 7.245637388284723e+16,
|
| 2249 |
+
"train_batch_size": 1,
|
| 2250 |
+
"trial_name": null,
|
| 2251 |
+
"trial_params": null
|
| 2252 |
+
}
|
checkpoint-3174/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e28477db53b123d5e2a2a22d53d9a1733756369389c6c1a915f0f71e98972c0a
|
| 3 |
+
size 5240
|
checkpoint-504/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "appvoid/arco",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"head_dim": 96,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1536,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 4096,
|
| 15 |
+
"max_position_embeddings": 8192,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 16,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 8,
|
| 21 |
+
"pad_token_id": 0,
|
| 22 |
+
"pretraining_tp": 1,
|
| 23 |
+
"rms_norm_eps": 1e-05,
|
| 24 |
+
"rope_scaling": null,
|
| 25 |
+
"rope_theta": 100000,
|
| 26 |
+
"sliding_window": null,
|
| 27 |
+
"tie_word_embeddings": false,
|
| 28 |
+
"torch_dtype": "float32",
|
| 29 |
+
"transformers_version": "4.46.1",
|
| 30 |
+
"use_cache": true,
|
| 31 |
+
"vocab_size": 32000
|
| 32 |
+
}
|
checkpoint-504/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.46.1"
|
| 7 |
+
}
|
checkpoint-504/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c3519532810c36348d7f5af956b992ac058b8be0f88f754a2f928d8c09de4259
|
| 3 |
+
size 2054379856
|
checkpoint-504/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be09bd3e38c7561efb72d3681a1a1888a712b9307d1002a9ad336ae27a764e4b
|
| 3 |
+
size 4108850298
|
checkpoint-504/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:386fcc8cc1089aade9450d86fb239ea3483f455fd2d78d8378645feecfec9d69
|
| 3 |
+
size 14244
|
checkpoint-504/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:021d0cb088f99ee86c98722ec5d1bb58d63ca4c9df1501dfadbe903ee6c8fbcb
|
| 3 |
+
size 1064
|
checkpoint-504/trainer_state.json
ADDED
|
@@ -0,0 +1,383 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 5000,
|
| 6 |
+
"global_step": 504,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01984126984126984,
|
| 13 |
+
"grad_norm": 2912.028564453125,
|
| 14 |
+
"learning_rate": 2e-08,
|
| 15 |
+
"loss": 14.5926,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03968253968253968,
|
| 20 |
+
"grad_norm": 6037.033203125,
|
| 21 |
+
"learning_rate": 2.0000000000000002e-07,
|
| 22 |
+
"loss": 14.6525,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.05952380952380952,
|
| 27 |
+
"grad_norm": 7095.4931640625,
|
| 28 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 29 |
+
"loss": 13.8958,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07936507936507936,
|
| 34 |
+
"grad_norm": 3047.11572265625,
|
| 35 |
+
"learning_rate": 6.000000000000001e-07,
|
| 36 |
+
"loss": 13.0648,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0992063492063492,
|
| 41 |
+
"grad_norm": 20132.697265625,
|
| 42 |
+
"learning_rate": 8.000000000000001e-07,
|
| 43 |
+
"loss": 11.7415,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.11904761904761904,
|
| 48 |
+
"grad_norm": 65280.9921875,
|
| 49 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 50 |
+
"loss": 10.186,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.1388888888888889,
|
| 55 |
+
"grad_norm": 18013.744140625,
|
| 56 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 57 |
+
"loss": 8.9537,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.15873015873015872,
|
| 62 |
+
"grad_norm": 12039.76171875,
|
| 63 |
+
"learning_rate": 1.4000000000000001e-06,
|
| 64 |
+
"loss": 8.8499,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.17857142857142858,
|
| 69 |
+
"grad_norm": 4152.63037109375,
|
| 70 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 71 |
+
"loss": 8.1076,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.1984126984126984,
|
| 76 |
+
"grad_norm": 112633.828125,
|
| 77 |
+
"learning_rate": 1.8000000000000001e-06,
|
| 78 |
+
"loss": 7.3218,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.21825396825396826,
|
| 83 |
+
"grad_norm": 12219.0498046875,
|
| 84 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 85 |
+
"loss": 6.8102,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.23809523809523808,
|
| 90 |
+
"grad_norm": 5337.48681640625,
|
| 91 |
+
"learning_rate": 2.2e-06,
|
| 92 |
+
"loss": 6.4298,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.25793650793650796,
|
| 97 |
+
"grad_norm": 66286.9609375,
|
| 98 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 99 |
+
"loss": 6.1412,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2777777777777778,
|
| 104 |
+
"grad_norm": 22970.162109375,
|
| 105 |
+
"learning_rate": 2.6e-06,
|
| 106 |
+
"loss": 6.1932,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.2976190476190476,
|
| 111 |
+
"grad_norm": 1229.7572021484375,
|
| 112 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 113 |
+
"loss": 5.807,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.31746031746031744,
|
| 118 |
+
"grad_norm": 4040.58544921875,
|
| 119 |
+
"learning_rate": 3e-06,
|
| 120 |
+
"loss": 5.3959,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.3373015873015873,
|
| 125 |
+
"grad_norm": 79031.90625,
|
| 126 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 127 |
+
"loss": 5.1443,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.35714285714285715,
|
| 132 |
+
"grad_norm": 2377.986328125,
|
| 133 |
+
"learning_rate": 3.4000000000000005e-06,
|
| 134 |
+
"loss": 5.1,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.376984126984127,
|
| 139 |
+
"grad_norm": 6816.78173828125,
|
| 140 |
+
"learning_rate": 3.6000000000000003e-06,
|
| 141 |
+
"loss": 5.0767,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3968253968253968,
|
| 146 |
+
"grad_norm": 329.9869384765625,
|
| 147 |
+
"learning_rate": 3.8000000000000005e-06,
|
| 148 |
+
"loss": 4.7978,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4166666666666667,
|
| 153 |
+
"grad_norm": 551.7224731445312,
|
| 154 |
+
"learning_rate": 4.000000000000001e-06,
|
| 155 |
+
"loss": 4.3107,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.4365079365079365,
|
| 160 |
+
"grad_norm": 306.0798645019531,
|
| 161 |
+
"learning_rate": 4.2000000000000004e-06,
|
| 162 |
+
"loss": 4.126,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.45634920634920634,
|
| 167 |
+
"grad_norm": 259.38702392578125,
|
| 168 |
+
"learning_rate": 4.4e-06,
|
| 169 |
+
"loss": 3.3982,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.47619047619047616,
|
| 174 |
+
"grad_norm": 152.5589141845703,
|
| 175 |
+
"learning_rate": 4.600000000000001e-06,
|
| 176 |
+
"loss": 2.9569,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.49603174603174605,
|
| 181 |
+
"grad_norm": 188.98681640625,
|
| 182 |
+
"learning_rate": 4.800000000000001e-06,
|
| 183 |
+
"loss": 2.3802,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5158730158730159,
|
| 188 |
+
"grad_norm": 214.31179809570312,
|
| 189 |
+
"learning_rate": 5e-06,
|
| 190 |
+
"loss": 2.0057,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5357142857142857,
|
| 195 |
+
"grad_norm": 184.34437561035156,
|
| 196 |
+
"learning_rate": 5.2e-06,
|
| 197 |
+
"loss": 1.6501,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.5555555555555556,
|
| 202 |
+
"grad_norm": 119.70015716552734,
|
| 203 |
+
"learning_rate": 5.400000000000001e-06,
|
| 204 |
+
"loss": 1.4393,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5753968253968254,
|
| 209 |
+
"grad_norm": 172.4916534423828,
|
| 210 |
+
"learning_rate": 5.600000000000001e-06,
|
| 211 |
+
"loss": 1.2536,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5952380952380952,
|
| 216 |
+
"grad_norm": 152.96202087402344,
|
| 217 |
+
"learning_rate": 5.8e-06,
|
| 218 |
+
"loss": 1.1011,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6150793650793651,
|
| 223 |
+
"grad_norm": 229.5851593017578,
|
| 224 |
+
"learning_rate": 6e-06,
|
| 225 |
+
"loss": 0.9963,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6349206349206349,
|
| 230 |
+
"grad_norm": 192.8756103515625,
|
| 231 |
+
"learning_rate": 6.200000000000001e-06,
|
| 232 |
+
"loss": 0.893,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.6547619047619048,
|
| 237 |
+
"grad_norm": 285.22601318359375,
|
| 238 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 239 |
+
"loss": 0.7744,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.6746031746031746,
|
| 244 |
+
"grad_norm": 121.977783203125,
|
| 245 |
+
"learning_rate": 6.600000000000001e-06,
|
| 246 |
+
"loss": 0.6989,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.6944444444444444,
|
| 251 |
+
"grad_norm": 135.0439453125,
|
| 252 |
+
"learning_rate": 6.800000000000001e-06,
|
| 253 |
+
"loss": 0.559,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7142857142857143,
|
| 258 |
+
"grad_norm": 157.01791381835938,
|
| 259 |
+
"learning_rate": 7e-06,
|
| 260 |
+
"loss": 0.5721,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7341269841269841,
|
| 265 |
+
"grad_norm": 139.25894165039062,
|
| 266 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 267 |
+
"loss": 0.5385,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.753968253968254,
|
| 272 |
+
"grad_norm": 94.0282974243164,
|
| 273 |
+
"learning_rate": 7.4e-06,
|
| 274 |
+
"loss": 0.5169,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.7738095238095238,
|
| 279 |
+
"grad_norm": 64.70858764648438,
|
| 280 |
+
"learning_rate": 7.600000000000001e-06,
|
| 281 |
+
"loss": 0.4738,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.7936507936507936,
|
| 286 |
+
"grad_norm": 68.39224243164062,
|
| 287 |
+
"learning_rate": 7.800000000000002e-06,
|
| 288 |
+
"loss": 0.4533,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.8134920634920635,
|
| 293 |
+
"grad_norm": 55.44282150268555,
|
| 294 |
+
"learning_rate": 8.000000000000001e-06,
|
| 295 |
+
"loss": 0.4296,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.8333333333333334,
|
| 300 |
+
"grad_norm": 66.44039154052734,
|
| 301 |
+
"learning_rate": 8.2e-06,
|
| 302 |
+
"loss": 0.3933,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8531746031746031,
|
| 307 |
+
"grad_norm": 64.27642822265625,
|
| 308 |
+
"learning_rate": 8.400000000000001e-06,
|
| 309 |
+
"loss": 0.4348,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.873015873015873,
|
| 314 |
+
"grad_norm": 1226.5478515625,
|
| 315 |
+
"learning_rate": 8.6e-06,
|
| 316 |
+
"loss": 0.6043,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.8928571428571429,
|
| 321 |
+
"grad_norm": 94.26041412353516,
|
| 322 |
+
"learning_rate": 8.8e-06,
|
| 323 |
+
"loss": 0.3775,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.9126984126984127,
|
| 328 |
+
"grad_norm": 63.64945983886719,
|
| 329 |
+
"learning_rate": 9e-06,
|
| 330 |
+
"loss": 0.3306,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.9325396825396826,
|
| 335 |
+
"grad_norm": 41.9366340637207,
|
| 336 |
+
"learning_rate": 9.200000000000002e-06,
|
| 337 |
+
"loss": 0.3727,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.9523809523809523,
|
| 342 |
+
"grad_norm": 36.641475677490234,
|
| 343 |
+
"learning_rate": 9.4e-06,
|
| 344 |
+
"loss": 0.3561,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.9722222222222222,
|
| 349 |
+
"grad_norm": 30.62155532836914,
|
| 350 |
+
"learning_rate": 9.600000000000001e-06,
|
| 351 |
+
"loss": 0.3152,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.9920634920634921,
|
| 356 |
+
"grad_norm": 36.66413497924805,
|
| 357 |
+
"learning_rate": 9.800000000000001e-06,
|
| 358 |
+
"loss": 0.3266,
|
| 359 |
+
"step": 500
|
| 360 |
+
}
|
| 361 |
+
],
|
| 362 |
+
"logging_steps": 10,
|
| 363 |
+
"max_steps": 504,
|
| 364 |
+
"num_input_tokens_seen": 0,
|
| 365 |
+
"num_train_epochs": 1,
|
| 366 |
+
"save_steps": 5000,
|
| 367 |
+
"stateful_callbacks": {
|
| 368 |
+
"TrainerControl": {
|
| 369 |
+
"args": {
|
| 370 |
+
"should_epoch_stop": false,
|
| 371 |
+
"should_evaluate": false,
|
| 372 |
+
"should_log": false,
|
| 373 |
+
"should_save": true,
|
| 374 |
+
"should_training_stop": true
|
| 375 |
+
},
|
| 376 |
+
"attributes": {}
|
| 377 |
+
}
|
| 378 |
+
},
|
| 379 |
+
"total_flos": 2.2449113063424e+16,
|
| 380 |
+
"train_batch_size": 2,
|
| 381 |
+
"trial_name": null,
|
| 382 |
+
"trial_params": null
|
| 383 |
+
}
|
checkpoint-504/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f7f5cc37fb273660e0da0ca04c5eaac1a9426581b1d68b8b725281618e94ab4
|
| 3 |
+
size 5240
|
config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "h2oai/h2o-danube3-500m-chat",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"head_dim": 96,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1536,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 4096,
|
| 15 |
+
"max_position_embeddings": 8192,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 16,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 8,
|
| 21 |
+
"pad_token_id": 0,
|
| 22 |
+
"pretraining_tp": 1,
|
| 23 |
+
"rms_norm_eps": 1e-05,
|
| 24 |
+
"rope_scaling": null,
|
| 25 |
+
"rope_theta": 100000,
|
| 26 |
+
"sliding_window": null,
|
| 27 |
+
"tie_word_embeddings": false,
|
| 28 |
+
"torch_dtype": "float32",
|
| 29 |
+
"transformers_version": "4.46.1",
|
| 30 |
+
"use_cache": true,
|
| 31 |
+
"vocab_size": 32000
|
| 32 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.46.1"
|
| 7 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e34da29d939d3c71535fdf34b3344bfa5bf1faadb8407db8583d575ec04d8328
|
| 3 |
+
size 2054379856
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"cls_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"eos_token": {
|
| 17 |
+
"content": "</s>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"pad_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
},
|
| 30 |
+
"sep_token": {
|
| 31 |
+
"content": "</s>",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false
|
| 36 |
+
},
|
| 37 |
+
"unk_token": {
|
| 38 |
+
"content": "<unk>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false
|
| 43 |
+
}
|
| 44 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
|
| 3 |
+
size 493443
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": false,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": true
|
| 29 |
+
}
|
| 30 |
+
},
|
| 31 |
+
"additional_special_tokens": [],
|
| 32 |
+
"bos_token": "<s>",
|
| 33 |
+
"chat_template": "{% for message in messages %}{% if message['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% if ((message['role'] == 'user') != (loop.index0 % 2 == 0)) or ((message['role'] == 'assistant') != (loop.index0 % 2 == 1)) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '<|prompt|>' + message['content'].strip() + eos_token }}{% elif message['role'] == 'assistant' %}{{ '<|answer|>' + message['content'].strip() + eos_token }}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '<|answer|>' }}{% endif %}",
|
| 34 |
+
"clean_up_tokenization_spaces": false,
|
| 35 |
+
"cls_token": "</s>",
|
| 36 |
+
"eos_token": "</s>",
|
| 37 |
+
"legacy": true,
|
| 38 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 39 |
+
"pad_token": "<unk>",
|
| 40 |
+
"sep_token": "</s>",
|
| 41 |
+
"sp_model_kwargs": {},
|
| 42 |
+
"spaces_between_special_tokens": false,
|
| 43 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 44 |
+
"unk_token": "<unk>",
|
| 45 |
+
"use_default_system_prompt": false
|
| 46 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e28477db53b123d5e2a2a22d53d9a1733756369389c6c1a915f0f71e98972c0a
|
| 3 |
+
size 5240
|