metadata

license: mit
base_model: gpt2-xl
tags:
  - generated_from_trainer
datasets:
  - tyzhu/lmind_nq_train600_eval300_v1_recite_qa
metrics:
  - accuracy
model-index:
  - name: lmind_nq_train600_eval300_v1_recite_qa_gpt2-xl_1e-4
    results:
      - task:
          name: Causal Language Modeling
          type: text-generation
        dataset:
          name: tyzhu/lmind_nq_train600_eval300_v1_recite_qa
          type: tyzhu/lmind_nq_train600_eval300_v1_recite_qa
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8390196078431372

lmind_nq_train600_eval300_v1_recite_qa_gpt2-xl_1e-4

This model is a fine-tuned version of gpt2-xl on the tyzhu/lmind_nq_train600_eval300_v1_recite_qa dataset. It achieves the following results on the evaluation set:

Loss: 0.3982
Accuracy: 0.8390

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: 0.0001
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.0096	1.0	93	1.3802	0.6793
0.7666	2.0	186	0.6686	0.7849
0.31	3.0	279	0.4719	0.8184
0.1608	4.0	372	0.4038	0.8311
0.1101	5.0	465	0.3742	0.8372
0.0839	6.0	558	0.3734	0.8393
0.0743	7.0	651	0.3625	0.8404
0.0756	8.0	744	0.3654	0.8399
0.0694	9.0	837	0.3742	0.8400
0.0669	10.0	930	0.3712	0.8403
0.0692	11.0	1023	0.3812	0.8397
0.0717	12.0	1116	0.3797	0.8395
0.0762	13.0	1209	0.3892	0.8393
0.0823	14.0	1302	0.3993	0.8384
0.0789	15.0	1395	0.3946	0.8389
0.0737	16.0	1488	0.3927	0.8393
0.0739	17.0	1581	0.3977	0.8381
0.0741	18.0	1674	0.4060	0.8379
0.0741	19.0	1767	0.4047	0.8389
0.0715	20.0	1860	0.3982	0.8390

Framework versions

Transformers 4.34.0
Pytorch 2.1.0+cu121
Datasets 2.14.5
Tokenizers 0.14.1