Mistral-7B_task-2_180-samples_config-2_full_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: 0.7670

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
1.0907	0.9412	8	1.0689
0.9681	2.0	17	0.9509
0.8795	2.9412	25	0.8435
0.7349	4.0	34	0.7955
0.721	4.9412	42	0.7799
0.7181	6.0	51	0.7700
0.7032	6.9412	59	0.7670
0.6292	8.0	68	0.7738
0.5437	8.9412	76	0.7970
0.5035	10.0	85	0.8213
0.4565	10.9412	93	0.8529
0.3822	12.0	102	0.8878
0.3563	12.9412	110	0.9018
0.3062	14.0	119	0.9819