README.md

metadata

base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
datasets:
  - GaetanMichelet/chat-60_ft_task-2
  - GaetanMichelet/chat-120_ft_task-2
library_name: peft
license: llama3.1
tags:
  - alignment-handbook
  - trl
  - sft
  - generated_from_trainer
model-index:
  - name: Llama-31-8B_task-2_120-samples_config-3
    results: []

Llama-31-8B_task-2_120-samples_config-3

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-2 and the GaetanMichelet/chat-120_ft_task-2 datasets. It achieves the following results on the evaluation set:

• Loss: 0.7139

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: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 150

Training results

Training Loss	Epoch	Step	Validation Loss
1.0459	1.0	11	1.1227
1.0223	2.0	22	1.1149
1.0795	3.0	33	1.1018
0.9982	4.0	44	1.0787
0.9702	5.0	55	1.0444
0.9509	6.0	66	0.9990
0.9573	7.0	77	0.9500
0.8624	8.0	88	0.9071
0.8804	9.0	99	0.8747
0.8515	10.0	110	0.8457
0.7864	11.0	121	0.8208
0.8648	12.0	132	0.8018
0.736	13.0	143	0.7867
0.7882	14.0	154	0.7728
0.7452	15.0	165	0.7604
0.6818	16.0	176	0.7485
0.7119	17.0	187	0.7387
0.7107	18.0	198	0.7307
0.6405	19.0	209	0.7238
0.6075	20.0	220	0.7188
0.6323	21.0	231	0.7152
0.557	22.0	242	0.7139
0.5692	23.0	253	0.7158
0.558	24.0	264	0.7198
0.5153	25.0	275	0.7296
0.4964	26.0	286	0.7367
0.4713	27.0	297	0.7403
0.4144	28.0	308	0.7620
0.4184	29.0	319	0.7954

Framework versions

• PEFT 0.12.0
• Transformers 4.44.0
• Pytorch 2.1.2+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1