20230830102630
This model is a fine-tuned version of bert-large-cased on the super_glue dataset. It achieves the following results on the evaluation set:
- Loss: 0.6751
- Accuracy: 0.4984
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: 0.0005
- train_batch_size: 16
- eval_batch_size: 8
- seed: 11
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 80.0
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy |
---|---|---|---|---|
No log | 1.0 | 340 | 0.6962 | 0.5 |
0.7003 | 2.0 | 680 | 0.6657 | 0.5345 |
0.6959 | 3.0 | 1020 | 0.6703 | 0.5 |
0.6959 | 4.0 | 1360 | 0.6845 | 0.5 |
0.6978 | 5.0 | 1700 | 0.6767 | 0.5 |
0.6854 | 6.0 | 2040 | 0.6876 | 0.5 |
0.6854 | 7.0 | 2380 | 0.6705 | 0.5 |
0.6851 | 8.0 | 2720 | 0.6806 | 0.5 |
0.6845 | 9.0 | 3060 | 0.6881 | 0.5 |
0.6845 | 10.0 | 3400 | 0.6737 | 0.5 |
0.6835 | 11.0 | 3740 | 0.6734 | 0.5 |
0.6821 | 12.0 | 4080 | 0.7058 | 0.5 |
0.6821 | 13.0 | 4420 | 0.7057 | 0.5 |
0.682 | 14.0 | 4760 | 0.7057 | 0.5 |
0.6827 | 15.0 | 5100 | 0.6771 | 0.5 |
0.6827 | 16.0 | 5440 | 0.6848 | 0.5 |
0.6803 | 17.0 | 5780 | 0.7044 | 0.5 |
0.6821 | 18.0 | 6120 | 0.6720 | 0.4984 |
0.6821 | 19.0 | 6460 | 0.6716 | 0.5 |
0.6784 | 20.0 | 6800 | 0.6855 | 0.5 |
0.6821 | 21.0 | 7140 | 0.6705 | 0.5 |
0.6821 | 22.0 | 7480 | 0.6753 | 0.5 |
0.6888 | 23.0 | 7820 | 0.6745 | 0.4953 |
0.6821 | 24.0 | 8160 | 0.6716 | 0.5 |
0.682 | 25.0 | 8500 | 0.6702 | 0.5 |
0.682 | 26.0 | 8840 | 0.6791 | 0.5 |
0.6829 | 27.0 | 9180 | 0.6771 | 0.5 |
0.6807 | 28.0 | 9520 | 0.6719 | 0.5 |
0.6807 | 29.0 | 9860 | 0.6739 | 0.5 |
0.6783 | 30.0 | 10200 | 0.6716 | 0.5 |
0.6789 | 31.0 | 10540 | 0.6706 | 0.5 |
0.6789 | 32.0 | 10880 | 0.7163 | 0.5 |
0.6798 | 33.0 | 11220 | 0.6703 | 0.5 |
0.6785 | 34.0 | 11560 | 0.6822 | 0.5 |
0.6785 | 35.0 | 11900 | 0.6715 | 0.5 |
0.6783 | 36.0 | 12240 | 0.6720 | 0.5 |
0.6781 | 37.0 | 12580 | 0.6733 | 0.5 |
0.6781 | 38.0 | 12920 | 0.6707 | 0.5 |
0.6798 | 39.0 | 13260 | 0.6950 | 0.5 |
0.6755 | 40.0 | 13600 | 0.6705 | 0.5 |
0.6755 | 41.0 | 13940 | 0.6715 | 0.5 |
0.6776 | 42.0 | 14280 | 0.6704 | 0.5 |
0.6772 | 43.0 | 14620 | 0.6789 | 0.5 |
0.6772 | 44.0 | 14960 | 0.6707 | 0.5 |
0.6755 | 45.0 | 15300 | 0.6925 | 0.5 |
0.6748 | 46.0 | 15640 | 0.6727 | 0.5 |
0.6748 | 47.0 | 15980 | 0.6801 | 0.5 |
0.6754 | 48.0 | 16320 | 0.6714 | 0.5 |
0.6762 | 49.0 | 16660 | 0.6882 | 0.5 |
0.6753 | 50.0 | 17000 | 0.6710 | 0.5 |
0.6753 | 51.0 | 17340 | 0.6707 | 0.5 |
0.6734 | 52.0 | 17680 | 0.6726 | 0.5063 |
0.678 | 53.0 | 18020 | 0.6727 | 0.5 |
0.678 | 54.0 | 18360 | 0.6751 | 0.5 |
0.6719 | 55.0 | 18700 | 0.6712 | 0.5 |
0.6726 | 56.0 | 19040 | 0.6721 | 0.5 |
0.6726 | 57.0 | 19380 | 0.6715 | 0.5 |
0.6732 | 58.0 | 19720 | 0.6717 | 0.5016 |
0.6736 | 59.0 | 20060 | 0.6819 | 0.5 |
0.6736 | 60.0 | 20400 | 0.6728 | 0.5141 |
0.6732 | 61.0 | 20740 | 0.6716 | 0.5016 |
0.6727 | 62.0 | 21080 | 0.6747 | 0.5 |
0.6727 | 63.0 | 21420 | 0.6715 | 0.4984 |
0.6726 | 64.0 | 21760 | 0.6737 | 0.5 |
0.6721 | 65.0 | 22100 | 0.6724 | 0.5 |
0.6721 | 66.0 | 22440 | 0.6744 | 0.5 |
0.6711 | 67.0 | 22780 | 0.6720 | 0.5 |
0.6725 | 68.0 | 23120 | 0.6722 | 0.4984 |
0.6725 | 69.0 | 23460 | 0.6722 | 0.4984 |
0.6713 | 70.0 | 23800 | 0.6722 | 0.4984 |
0.6708 | 71.0 | 24140 | 0.6743 | 0.5 |
0.6708 | 72.0 | 24480 | 0.6794 | 0.5 |
0.6703 | 73.0 | 24820 | 0.6756 | 0.5 |
0.6702 | 74.0 | 25160 | 0.6760 | 0.5 |
0.6688 | 75.0 | 25500 | 0.6741 | 0.4984 |
0.6688 | 76.0 | 25840 | 0.6753 | 0.5 |
0.67 | 77.0 | 26180 | 0.6730 | 0.4984 |
0.6688 | 78.0 | 26520 | 0.6751 | 0.4984 |
0.6688 | 79.0 | 26860 | 0.6750 | 0.4984 |
0.6685 | 80.0 | 27200 | 0.6751 | 0.4984 |
Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3
- Downloads last month
- 3
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.