Llama-31-8B_task-3_120-samples_config-1

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-3 and the GaetanMichelet/chat-120_ft_task-3 datasets. It achieves the following results on the evaluation set:

• Loss: 0.4323

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: 0.0001
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss
2.1435	1.0	11	2.0795
0.7297	2.0	22	0.8106
0.3105	3.0	33	0.5279
0.4824	4.0	44	0.4706
0.3517	5.0	55	0.4750
0.3732	6.0	66	0.4323
0.2019	7.0	77	0.4858
0.1766	8.0	88	0.4912
0.1	9.0	99	0.5915
0.0326	10.0	110	0.6782
0.0057	11.0	121	0.6927
0.0107	12.0	132	0.8775
0.0022	13.0	143	0.8794

Framework versions

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