README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
metrics:
  - accuracy
model-index:
  - name: bert-large-uncased_winobias_classifieronly
    results: []

bert-large-uncased_winobias_classifieronly

This model is a fine-tuned version of bert-large-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6934
• Accuracy: 0.5095

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: 128
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.38	5	0.8249	0.4987
No log	0.77	10	0.8058	0.4987
No log	1.15	15	0.7886	0.4987
No log	1.54	20	0.7731	0.4987
No log	1.92	25	0.7593	0.4981
No log	2.31	30	0.7470	0.4987
No log	2.69	35	0.7363	0.4994
No log	3.08	40	0.7276	0.4994
No log	3.46	45	0.7205	0.4994
No log	3.85	50	0.7142	0.5
No log	4.23	55	0.7088	0.5
No log	4.62	60	0.7044	0.4994
No log	5.0	65	0.7009	0.5
No log	5.38	70	0.6983	0.5013
No log	5.77	75	0.6966	0.5006
No log	6.15	80	0.6955	0.4994
No log	6.54	85	0.6948	0.5032
No log	6.92	90	0.6943	0.5158
No log	7.31	95	0.6940	0.5114
No log	7.69	100	0.6938	0.5120
No log	8.08	105	0.6936	0.5088
No log	8.46	110	0.6936	0.5101
No log	8.85	115	0.6935	0.5101
No log	9.23	120	0.6935	0.5107
No log	9.62	125	0.6934	0.5120
No log	10.0	130	0.6934	0.5107
No log	10.38	135	0.6934	0.5126
No log	10.77	140	0.6934	0.5133
No log	11.15	145	0.6934	0.5139
No log	11.54	150	0.6934	0.5114
No log	11.92	155	0.6934	0.5088
No log	12.31	160	0.6934	0.5076
No log	12.69	165	0.6934	0.5063
No log	13.08	170	0.6934	0.5101
No log	13.46	175	0.6934	0.5107
No log	13.85	180	0.6934	0.5101
No log	14.23	185	0.6934	0.5095
No log	14.62	190	0.6934	0.5095
No log	15.0	195	0.6934	0.5095

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1
• Datasets 2.10.1
• Tokenizers 0.13.2