Llama-31-8B_task-2_180-samples_config-1_full_auto

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-2_auto, the GaetanMichelet/chat-120_ft_task-2_auto and the GaetanMichelet/chat-180_ft_task-2_auto datasets. It achieves the following results on the evaluation set:

• Loss: 0.9985

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
1.4597	1.0	17	1.4484
1.2836	2.0	34	1.2670
1.0573	3.0	51	1.0740
0.9449	4.0	68	1.0341
0.9261	5.0	85	1.0129
0.8846	6.0	102	0.9985
0.8643	7.0	119	0.9999
0.7576	8.0	136	1.0171
0.6809	9.0	153	1.0483
0.6404	10.0	170	1.0951
0.5943	11.0	187	1.1375
0.4267	12.0	204	1.2362
0.446	13.0	221	1.2402

Framework versions

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