Mistral-7B_task-2_180-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, the GaetanMichelet/chat-120_ft_task-2 and the GaetanMichelet/chat-180_ft_task-2 datasets. It achieves the following results on the evaluation set:

• Loss: 0.7895

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
1.1409	0.9412	8	1.1128
1.0026	2.0	17	0.9789
0.9165	2.9412	25	0.8753
0.7642	4.0	34	0.8236
0.7468	4.9412	42	0.8065
0.7471	6.0	51	0.7946
0.7355	6.9412	59	0.7895
0.6638	8.0	68	0.7944
0.5766	8.9412	76	0.8071
0.542	10.0	85	0.8387
0.4866	10.9412	93	0.8615
0.4049	12.0	102	0.8863
0.3762	12.9412	110	0.9319
0.3261	14.0	119	1.0103

Framework versions

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