Mistral-7B_task-3_180-samples_config-1_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: 0.8536

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
1.1289	1.0	17	1.1091
0.9469	2.0	34	0.9381
0.8431	3.0	51	0.8753
0.777	4.0	68	0.8566
0.7472	5.0	85	0.8536
0.6439	6.0	102	0.8732
0.6021	7.0	119	0.8925
0.4804	8.0	136	0.9462
0.387	9.0	153	0.9822
0.3473	10.0	170	1.0300
0.2927	11.0	187	1.0960
0.1918	12.0	204	1.1832

Framework versions

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