README.md

metadata

license: mit
tags:
  - generated_from_trainer
datasets:
  - stereoset
metrics:
  - accuracy
model-index:
  - name: gpt2_stereoset_finetuned
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: stereoset
          type: stereoset
          config: intersentence
          split: validation
          args: intersentence
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.7087912087912088

gpt2_stereoset_finetuned

This model is a fine-tuned version of gpt2 on the stereoset dataset. It achieves the following results on the evaluation set:

• Loss: 0.6545
• Accuracy: 0.7088

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: 5e-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: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.21	5	1.1855	0.5259
No log	0.42	10	0.7056	0.5338
No log	0.62	15	0.7009	0.5400
No log	0.83	20	0.7230	0.5173
No log	1.04	25	0.6666	0.5989
No log	1.25	30	0.6812	0.5699
No log	1.46	35	0.6479	0.6272
No log	1.67	40	0.6323	0.6484
No log	1.88	45	0.6306	0.6515
No log	2.08	50	0.6474	0.6633
No log	2.29	55	0.6158	0.6641
No log	2.5	60	0.6059	0.6703
No log	2.71	65	0.6151	0.6695
No log	2.92	70	0.5860	0.6782
No log	3.12	75	0.5808	0.6907
No log	3.33	80	0.5953	0.6915
No log	3.54	85	0.5860	0.6994
No log	3.75	90	0.5918	0.6947
No log	3.96	95	0.5915	0.6797
No log	4.17	100	0.5779	0.7041
No log	4.38	105	0.5902	0.7151
No log	4.58	110	0.5740	0.7080
No log	4.79	115	0.5640	0.7088
No log	5.0	120	0.5786	0.6947
No log	5.21	125	0.5892	0.6978
No log	5.42	130	0.5722	0.7096
No log	5.62	135	0.5743	0.7064
No log	5.83	140	0.5873	0.7057
No log	6.04	145	0.5915	0.7033
No log	6.25	150	0.5978	0.7009
No log	6.46	155	0.6034	0.6931
No log	6.67	160	0.5908	0.7111
No log	6.88	165	0.5954	0.6947
No log	7.08	170	0.5882	0.7033
No log	7.29	175	0.5895	0.7151
No log	7.5	180	0.6077	0.7104
No log	7.71	185	0.6121	0.7151
No log	7.92	190	0.6086	0.7151
No log	8.12	195	0.6182	0.7127
No log	8.33	200	0.6412	0.7072
No log	8.54	205	0.6425	0.7049
No log	8.75	210	0.6369	0.7135
No log	8.96	215	0.6405	0.7111
No log	9.17	220	0.6431	0.7135
No log	9.38	225	0.6474	0.7127
No log	9.58	230	0.6595	0.7041
No log	9.79	235	0.6580	0.7041
No log	10.0	240	0.6545	0.7088

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1
• Datasets 2.9.0
• Tokenizers 0.13.2