medmcqa-tiny-bert

This model is a fine-tuned version of Intel/dynamic_tinybert on MedMCQA dataset. It achieves the following results on the evaluation set:

• Train Loss: 0.4325
• Validation Loss: 3.1445
• Train Accuracy: 0.293
• Epoch: 9

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:

• optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': True, 'is_legacy_optimizer': False, 'learning_rate': {'module': 'keras.optimizers.schedules', 'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 5e-05, 'decay_steps': 5000, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, 'registered_name': None}, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
• training_precision: float32

Training results

Train Loss	Validation Loss	Train Accuracy	Epoch
1.3858	1.3862	0.329	0
1.3878	1.3850	0.321	1
1.3784	1.3869	0.318	2
1.3172	1.3945	0.33	3
1.1564	1.5962	0.307	4
0.9487	1.6876	0.295	5
0.7610	2.1023	0.29	6
0.6154	2.5488	0.289	7
0.5057	2.8837	0.292	8
0.4325	3.1445	0.293	9

Framework versions

• Transformers 4.37.2
• TensorFlow 2.15.0
• Datasets 2.17.1
• Tokenizers 0.15.2