Mistral-7B_task-3_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-3 and the GaetanMichelet/chat-120_ft_task-3 datasets. It achieves the following results on the evaluation set:

Loss: 0.8600

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
1.2758	0.9091	5	1.2130
1.1641	2.0	11	1.0929
1.0378	2.9091	16	1.0190
0.9062	4.0	22	0.9140
0.8437	4.9091	27	0.8845
0.8116	6.0	33	0.8704
0.7944	6.9091	38	0.8623
0.7673	8.0	44	0.8600
0.7099	8.9091	49	0.8655
0.7036	10.0	55	0.8723
0.6352	10.9091	60	0.8909
0.5565	12.0	66	0.9372
0.4968	12.9091	71	0.9536
0.4323	14.0	77	0.9887
0.3236	14.9091	82	1.0386