bert-base-uncased-medmcqa-distill-of-bert-base-uncased-gpqa

This model is a fine-tuned version of google-bert/bert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 10.1269
• Accuracy: 0.5657

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: 0.0005
• train_batch_size: 16
• eval_batch_size: 16
• seed: 321
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	63	10.5322	0.2929
No log	2.0	126	10.2377	0.3283
No log	3.0	189	11.0437	0.4141
No log	4.0	252	8.4998	0.3535
No log	5.0	315	9.0941	0.3737
No log	6.0	378	9.5258	0.3889
No log	7.0	441	9.1010	0.4444
1.9584	8.0	504	8.6866	0.3434
1.9584	9.0	567	9.2581	0.3939
1.9584	10.0	630	9.2968	0.3687
1.9584	11.0	693	10.2154	0.3838
1.9584	12.0	756	10.1269	0.5657
1.9584	13.0	819	9.4519	0.3889
1.9584	14.0	882	9.7303	0.3737
1.9584	15.0	945	9.8894	0.3586
2.7625	16.0	1008	9.8269	0.3283
2.7625	17.0	1071	9.7131	0.3333
2.7625	18.0	1134	9.7690	0.3485
2.7625	19.0	1197	9.7724	0.3333
2.7625	20.0	1260	9.8628	0.4848

Framework versions

• Transformers 4.34.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.14.0