gpt2-large-NaturalQuestions_1000-ep20

This model is a fine-tuned version of gpt2-large on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 1.6582

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: 2e-05
  • train_batch_size: 12
  • eval_batch_size: 24
  • seed: 1799
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss
1.2555 0.6 50 1.0453
0.9434 1.19 100 1.0152
0.5866 1.79 150 1.0386
0.3851 2.38 200 1.1380
0.2864 2.98 250 1.1513
0.1613 3.57 300 1.2101
0.1386 4.17 350 1.2794
0.0886 4.76 400 1.2708
0.0716 5.36 450 1.3450
0.0611 5.95 500 1.3262
0.0353 6.55 550 1.4632
0.0361 7.14 600 1.4107
0.0275 7.74 650 1.4471
0.026 8.33 700 1.4811
0.0177 8.93 750 1.4831
0.015 9.52 800 1.4696
0.0146 10.12 850 1.5135
0.0126 10.71 900 1.5145
0.0118 11.31 950 1.5373
0.0117 11.9 1000 1.5292
0.0141 12.5 1050 1.5793
0.0074 13.1 1100 1.5453
0.0122 13.69 1150 1.5645
0.0092 14.29 1200 1.5721
0.0045 14.88 1250 1.5957
0.0054 15.48 1300 1.6253
0.0048 16.07 1350 1.6197
0.0035 16.67 1400 1.6325
0.0054 17.26 1450 1.6444
0.0041 17.86 1500 1.6465
0.0034 18.45 1550 1.6622
0.0036 19.05 1600 1.6589
0.0035 19.64 1650 1.6578

Framework versions

  • Transformers 4.29.2
  • Pytorch 1.10.0+cu111
  • Datasets 2.5.1
  • Tokenizers 0.13.3
Downloads last month
12
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.