Mistral-7B-v0_1_cola_sp0_ar0

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: 0.3526
• Accuracy: 0.8773

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: 48
• eval_batch_size: 80
• seed: 1
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 2
• total_train_batch_size: 192
• total_eval_batch_size: 160
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 750

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8621	1.23	50	0.8680	0.6261
0.6235	2.47	100	0.5236	0.7651
0.2987	3.7	150	0.4271	0.8121
0.3941	4.94	200	0.3921	0.8265
0.3579	6.17	250	0.3769	0.8293
0.2617	7.41	300	0.3744	0.8360
0.3043	8.64	350	0.3863	0.8360
0.2354	9.88	400	0.3478	0.8581
0.1854	11.11	450	0.3686	0.8437
0.1896	12.35	500	0.3892	0.8543
0.1561	13.58	550	0.4256	0.8504
0.1448	14.81	600	0.4322	0.8437
0.0355	16.05	650	0.5269	0.8351
0.0256	17.28	700	0.5536	0.8476
0.0549	18.52	750	0.6612	0.8485

Framework versions

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