Mistral-7B_task-3_180-samples_config-2_full_auto

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the GaetanMichelet/chat-60_ft_task-3_auto, the GaetanMichelet/chat-120_ft_task-3_auto and the GaetanMichelet/chat-180_ft_task-3_auto datasets. It achieves the following results on the evaluation set:

• Loss: 0.8526

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
1.2028	0.9412	8	1.1677
1.0682	2.0	17	1.0442
0.9635	2.9412	25	0.9281
0.8271	4.0	34	0.8806
0.8008	4.9412	42	0.8643
0.7975	6.0	51	0.8549
0.7857	6.9412	59	0.8526
0.6998	8.0	68	0.8610
0.5987	8.9412	76	0.8879
0.5553	10.0	85	0.9139
0.4959	10.9412	93	0.9456
0.4148	12.0	102	0.9965
0.3734	12.9412	110	1.0427
0.3239	14.0	119	1.0871

Framework versions

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