Mistral-7B-v0.1_cola_sparse_swiglu_scratch

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: 0.3823
• Accuracy: {'accuracy': 0.8621495327102804}
• Matthews Correlation: 0.6666

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
1.0186	0.31	25	0.8870	{'accuracy': 0.7152444870565676}	0.2813
0.7573	0.62	50	0.6207	{'accuracy': 0.788111217641419}	0.4870
0.5186	0.93	75	0.5819	{'accuracy': 0.7986577181208053}	0.4984
0.4259	1.24	100	0.5096	{'accuracy': 0.8149568552253116}	0.5459
0.5015	1.55	125	0.4887	{'accuracy': 0.8302972195589645}	0.6063
0.4622	1.86	150	0.4573	{'accuracy': 0.8437200383509108}	0.6273
0.3411	2.17	175	0.4755	{'accuracy': 0.835091083413231}	0.5958
0.3637	2.48	200	0.4302	{'accuracy': 0.8341323106423778}	0.5941
0.3376	2.8	225	0.3974	{'accuracy': 0.8446788111217641}	0.6325
0.2879	3.11	250	0.5189	{'accuracy': 0.8130393096836049}	0.6287
0.2691	3.42	275	0.4033	{'accuracy': 0.8331735378715245}	0.6148

Framework versions

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