Mistral-7B_task-2_180-samples_config-2_auto

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 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: 0.4678

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
0.7094	0.9412	8	0.6848
0.5612	2.0	17	0.5560
0.5167	2.9412	25	0.4976
0.4072	4.0	34	0.4678
0.3543	4.9412	42	0.4694
0.2826	6.0	51	0.4985
0.1782	6.9412	59	0.5504
0.0966	8.0	68	0.6488
0.0553	8.9412	76	0.7935
0.0336	10.0	85	0.7940
0.0233	10.9412	93	0.8753

Framework versions

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