Llama-31-8B_task-2_180-samples_config-2_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: 1.0008

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: 16
• total_train_batch_size: 16
• 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
1.4945	0.9412	8	1.4964
1.3936	2.0	17	1.3700
1.2937	2.9412	25	1.2535
1.0479	4.0	34	1.1012
0.9934	4.9412	42	1.0544
0.9927	6.0	51	1.0315
0.9944	6.9412	59	1.0179
0.9293	8.0	68	1.0085
0.8712	8.9412	76	1.0033
0.8604	10.0	85	1.0008
0.8278	10.9412	93	1.0047
0.7609	12.0	102	1.0211
0.7599	12.9412	110	1.0358
0.7215	14.0	119	1.0646
0.6195	14.9412	127	1.0861
0.5401	16.0	136	1.1417
0.5303	16.9412	144	1.1947

Framework versions

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