See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: true
base_model: unsloth/mistral-7b-instruct-v0.2
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - d952e0a5c02eb0c2_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/d952e0a5c02eb0c2_train_data.json
  type:
    field_input: rejected
    field_instruction: prompt
    field_output: chosen
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
early_stopping_threshold: 0.001
eval_max_new_tokens: 128
eval_steps: 40
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: false
hub_model_id: mrferr3t/7aed58b8-860d-4784-aebf-9bf246a01654
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0003
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 100
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
micro_batch_size: 32
mlflow_experiment_name: /tmp/d952e0a5c02eb0c2_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
s2_attention: null
sample_packing: false
save_steps: 40
saves_per_epoch: 0
sequence_len: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.02
wandb_entity: null
wandb_mode: online
wandb_name: 17d283c1-9977-4f78-b3fc-bc3f50b2f052
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 17d283c1-9977-4f78-b3fc-bc3f50b2f052
warmup_ratio: 0.05
weight_decay: 0.0
xformers_attention: null

7aed58b8-860d-4784-aebf-9bf246a01654

This model is a fine-tuned version of unsloth/mistral-7b-instruct-v0.2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3922

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0003
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_bnb_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 155
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0020	1	5.4757
No log	0.0806	40	1.9050
No log	0.1611	80	1.6329
4.8943	0.2417	120	1.6521
4.8943	0.3223	160	1.4903
2.9767	0.4028	200	1.3754
2.9767	0.4834	240	1.3377
2.9767	0.5639	280	1.2167
2.613	0.6445	320	1.1355
2.613	0.7251	360	0.9053
2.007	0.8056	400	0.7135
2.007	0.8862	440	0.7040
2.007	0.9668	480	0.6939
1.4776	1.0473	520	0.7442
1.4776	1.1279	560	0.6491
0.8086	1.2085	600	0.6461
0.8086	1.2890	640	0.5510
0.8086	1.3696	680	0.5785
0.8686	1.4502	720	0.5694
0.8686	1.5307	760	0.4838
0.778	1.6113	800	0.4864
0.778	1.6918	840	0.3634
0.778	1.7724	880	0.3510
0.7255	1.8530	920	0.3188
0.7255	1.9335	960	0.3315
0.746	2.0141	1000	0.3846
0.746	2.0947	1040	0.3922

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1