Mistral-7B_task-2_120-samples_config-2_full

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the GaetanMichelet/chat-60_ft_task-2 and the GaetanMichelet/chat-120_ft_task-2 datasets. It achieves the following results on the evaluation set:

• Loss: 0.7894

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
1.2098	0.9091	5	1.1456
1.0811	2.0	11	1.0116
0.952	2.9091	16	0.9449
0.8442	4.0	22	0.8452
0.7828	4.9091	27	0.8171
0.7505	6.0	33	0.8018
0.7405	6.9091	38	0.7939
0.7084	8.0	44	0.7894
0.6615	8.9091	49	0.7894
0.6539	10.0	55	0.7934
0.5975	10.9091	60	0.8043
0.5255	12.0	66	0.8411
0.4808	12.9091	71	0.8796
0.4272	14.0	77	0.9054
0.3206	14.9091	82	0.9417
0.3136	16.0	88	0.9865

Framework versions

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