Mistral-7B_task-1_60-samples_config_1

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the GaetanMichelet/chat_60_ft_t1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8374

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: 0.0002
• train_batch_size: 1
• eval_batch_size: 2
• 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
1.1301	1.0	1	1.2585
0.1458	2.0	3	1.1321
0.6168	3.0	5	1.0142
0.5313	4.0	7	0.9243
0.4402	5.0	9	0.8867
0.3691	6.0	11	0.8374
0.3	7.0	13	0.8456
0.2353	8.0	14	0.8596
0.2353	9.0	15	0.8829
0.1921	10.0	17	0.9306
0.1529	11.0	19	0.9858
0.0986	12.0	21	1.0125
0.0737	13.0	23	1.0741

Framework versions

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