Mistral-7B-v0.1_cola_sparse_swiglu

This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5063
• Accuracy: {'accuracy': 0.852803738317757}
• Matthews Correlation: 0.6476

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 16
• seed: 2
• distributed_type: multi-GPU
• num_devices: 6
• gradient_accumulation_steps: 2
• total_train_batch_size: 96
• total_eval_batch_size: 96
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Matthews Correlation
0.5035	0.12	10	0.5489	{'accuracy': 0.8398849472674976}	0.6092
0.3931	0.25	20	0.4954	{'accuracy': 0.8456375838926175}	0.6393
0.3433	0.37	30	0.5437	{'accuracy': 0.8293384467881112}	0.6149
0.3062	0.5	40	0.5389	{'accuracy': 0.8389261744966443}	0.6401

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.1+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0