gpt2_stereoset_classifieronly
This model is a fine-tuned version of gpt2 on the stereoset dataset. It achieves the following results on the evaluation set:
- Loss: 0.5990
- Accuracy: 0.6923
- Tp: 0.3501
- Tn: 0.3422
- Fp: 0.1625
- Fn: 0.1452
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: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | Tp | Tn | Fp | Fn |
---|---|---|---|---|---|---|---|---|
0.8922 | 0.43 | 20 | 0.6913 | 0.5549 | 0.2402 | 0.3148 | 0.1900 | 0.2551 |
0.7884 | 0.85 | 40 | 0.6671 | 0.5934 | 0.2182 | 0.3752 | 0.1295 | 0.2771 |
0.6991 | 1.28 | 60 | 0.6561 | 0.6193 | 0.2206 | 0.3987 | 0.1060 | 0.2747 |
0.6819 | 1.7 | 80 | 0.6499 | 0.6311 | 0.2088 | 0.4223 | 0.0824 | 0.2865 |
0.6501 | 2.13 | 100 | 0.6379 | 0.6507 | 0.2991 | 0.3516 | 0.1531 | 0.1962 |
0.6566 | 2.55 | 120 | 0.6569 | 0.6185 | 0.1695 | 0.4490 | 0.0557 | 0.3257 |
0.6671 | 2.98 | 140 | 0.6313 | 0.6609 | 0.2943 | 0.3666 | 0.1381 | 0.2009 |
0.6551 | 3.4 | 160 | 0.6309 | 0.6484 | 0.3862 | 0.2622 | 0.2425 | 0.1091 |
0.633 | 3.83 | 180 | 0.6244 | 0.6656 | 0.3014 | 0.3642 | 0.1405 | 0.1939 |
0.6432 | 4.26 | 200 | 0.6320 | 0.6554 | 0.2402 | 0.4152 | 0.0895 | 0.2551 |
0.6326 | 4.68 | 220 | 0.6240 | 0.6601 | 0.2849 | 0.3752 | 0.1295 | 0.2104 |
0.6347 | 5.11 | 240 | 0.6259 | 0.6523 | 0.3689 | 0.2834 | 0.2214 | 0.1264 |
0.6204 | 5.53 | 260 | 0.6256 | 0.6499 | 0.3697 | 0.2802 | 0.2245 | 0.1256 |
0.6242 | 5.96 | 280 | 0.6172 | 0.6774 | 0.3210 | 0.3564 | 0.1484 | 0.1743 |
0.6189 | 6.38 | 300 | 0.6186 | 0.6546 | 0.3493 | 0.3053 | 0.1994 | 0.1460 |
0.625 | 6.81 | 320 | 0.6187 | 0.6727 | 0.2881 | 0.3846 | 0.1201 | 0.2072 |
0.5963 | 7.23 | 340 | 0.6173 | 0.6758 | 0.3571 | 0.3187 | 0.1860 | 0.1381 |
0.6214 | 7.66 | 360 | 0.6158 | 0.6695 | 0.3203 | 0.3493 | 0.1554 | 0.1750 |
0.6007 | 8.09 | 380 | 0.6123 | 0.6797 | 0.3611 | 0.3187 | 0.1860 | 0.1342 |
0.6454 | 8.51 | 400 | 0.6168 | 0.6570 | 0.3736 | 0.2834 | 0.2214 | 0.1217 |
0.6012 | 8.94 | 420 | 0.6115 | 0.6868 | 0.3320 | 0.3548 | 0.1499 | 0.1633 |
0.627 | 9.36 | 440 | 0.6485 | 0.6193 | 0.1656 | 0.4537 | 0.0510 | 0.3297 |
0.6213 | 9.79 | 460 | 0.6092 | 0.6829 | 0.3022 | 0.3807 | 0.1240 | 0.1931 |
0.6286 | 10.21 | 480 | 0.6109 | 0.6711 | 0.3603 | 0.3108 | 0.1939 | 0.1350 |
0.609 | 10.64 | 500 | 0.6134 | 0.6633 | 0.3611 | 0.3022 | 0.2025 | 0.1342 |
0.5958 | 11.06 | 520 | 0.6409 | 0.6248 | 0.4262 | 0.1986 | 0.3061 | 0.0691 |
0.6494 | 11.49 | 540 | 0.6332 | 0.6342 | 0.4192 | 0.2151 | 0.2896 | 0.0761 |
0.6012 | 11.91 | 560 | 0.6159 | 0.6593 | 0.3885 | 0.2708 | 0.2339 | 0.1068 |
0.606 | 12.34 | 580 | 0.6050 | 0.6947 | 0.3359 | 0.3587 | 0.1460 | 0.1593 |
0.5872 | 12.77 | 600 | 0.6135 | 0.6641 | 0.3878 | 0.2763 | 0.2284 | 0.1075 |
0.6026 | 13.19 | 620 | 0.6061 | 0.6962 | 0.3265 | 0.3697 | 0.1350 | 0.1688 |
0.6179 | 13.62 | 640 | 0.6118 | 0.6876 | 0.2826 | 0.4050 | 0.0997 | 0.2127 |
0.5744 | 14.04 | 660 | 0.6058 | 0.6923 | 0.3030 | 0.3893 | 0.1154 | 0.1923 |
0.6061 | 14.47 | 680 | 0.6072 | 0.6860 | 0.2849 | 0.4011 | 0.1036 | 0.2104 |
0.609 | 14.89 | 700 | 0.6025 | 0.7064 | 0.3367 | 0.3697 | 0.1350 | 0.1586 |
0.6019 | 15.32 | 720 | 0.6046 | 0.6876 | 0.3540 | 0.3336 | 0.1711 | 0.1413 |
0.6183 | 15.74 | 740 | 0.6087 | 0.6735 | 0.3791 | 0.2943 | 0.2104 | 0.1162 |
0.6173 | 16.17 | 760 | 0.6010 | 0.6954 | 0.3407 | 0.3548 | 0.1499 | 0.1546 |
0.5873 | 16.6 | 780 | 0.6078 | 0.6766 | 0.3815 | 0.2951 | 0.2096 | 0.1138 |
0.6095 | 17.02 | 800 | 0.6151 | 0.6625 | 0.3948 | 0.2677 | 0.2370 | 0.1005 |
0.5936 | 17.45 | 820 | 0.6026 | 0.6915 | 0.3469 | 0.3446 | 0.1601 | 0.1484 |
0.5821 | 17.87 | 840 | 0.6025 | 0.6931 | 0.3485 | 0.3446 | 0.1601 | 0.1468 |
0.6036 | 18.3 | 860 | 0.6032 | 0.7049 | 0.3391 | 0.3658 | 0.1389 | 0.1562 |
0.5872 | 18.72 | 880 | 0.6057 | 0.6813 | 0.3587 | 0.3226 | 0.1821 | 0.1366 |
0.6085 | 19.15 | 900 | 0.6045 | 0.6845 | 0.3571 | 0.3273 | 0.1774 | 0.1381 |
0.5972 | 19.57 | 920 | 0.6203 | 0.6562 | 0.4042 | 0.2520 | 0.2527 | 0.0911 |
0.5732 | 20.0 | 940 | 0.6095 | 0.6672 | 0.3807 | 0.2865 | 0.2182 | 0.1146 |
0.5718 | 20.43 | 960 | 0.6054 | 0.6868 | 0.2936 | 0.3932 | 0.1115 | 0.2017 |
0.5919 | 20.85 | 980 | 0.6031 | 0.6931 | 0.3501 | 0.3430 | 0.1617 | 0.1452 |
0.6175 | 21.28 | 1000 | 0.6088 | 0.6703 | 0.3823 | 0.2881 | 0.2166 | 0.1130 |
0.5793 | 21.7 | 1020 | 0.5986 | 0.6994 | 0.3430 | 0.3564 | 0.1484 | 0.1523 |
0.5943 | 22.13 | 1040 | 0.6064 | 0.6852 | 0.2826 | 0.4027 | 0.1020 | 0.2127 |
0.5716 | 22.55 | 1060 | 0.5996 | 0.6947 | 0.3485 | 0.3462 | 0.1586 | 0.1468 |
0.6115 | 22.98 | 1080 | 0.6111 | 0.6727 | 0.3893 | 0.2834 | 0.2214 | 0.1060 |
0.5984 | 23.4 | 1100 | 0.6058 | 0.6837 | 0.3807 | 0.3030 | 0.2017 | 0.1146 |
0.5882 | 23.83 | 1120 | 0.5993 | 0.6962 | 0.3352 | 0.3611 | 0.1436 | 0.1601 |
0.5924 | 24.26 | 1140 | 0.6128 | 0.6680 | 0.3909 | 0.2771 | 0.2276 | 0.1044 |
0.5984 | 24.68 | 1160 | 0.6017 | 0.6970 | 0.3242 | 0.3728 | 0.1319 | 0.1711 |
0.5781 | 25.11 | 1180 | 0.6018 | 0.7002 | 0.3352 | 0.3650 | 0.1397 | 0.1601 |
0.5937 | 25.53 | 1200 | 0.6051 | 0.6845 | 0.3619 | 0.3226 | 0.1821 | 0.1334 |
0.5678 | 25.96 | 1220 | 0.5998 | 0.7002 | 0.3297 | 0.3705 | 0.1342 | 0.1656 |
0.5776 | 26.38 | 1240 | 0.6202 | 0.6523 | 0.3972 | 0.2551 | 0.2496 | 0.0981 |
0.5891 | 26.81 | 1260 | 0.6080 | 0.6821 | 0.3791 | 0.3030 | 0.2017 | 0.1162 |
0.5915 | 27.23 | 1280 | 0.6026 | 0.6947 | 0.2998 | 0.3948 | 0.1099 | 0.1954 |
0.5972 | 27.66 | 1300 | 0.5994 | 0.6931 | 0.3556 | 0.3375 | 0.1672 | 0.1397 |
0.5721 | 28.09 | 1320 | 0.6038 | 0.6829 | 0.3736 | 0.3093 | 0.1954 | 0.1217 |
0.5813 | 28.51 | 1340 | 0.5981 | 0.6954 | 0.3367 | 0.3587 | 0.1460 | 0.1586 |
0.5914 | 28.94 | 1360 | 0.5982 | 0.6986 | 0.3367 | 0.3619 | 0.1429 | 0.1586 |
0.5848 | 29.36 | 1380 | 0.5977 | 0.7002 | 0.3399 | 0.3603 | 0.1444 | 0.1554 |
0.5772 | 29.79 | 1400 | 0.6024 | 0.6876 | 0.3673 | 0.3203 | 0.1845 | 0.1279 |
0.581 | 30.21 | 1420 | 0.6004 | 0.6939 | 0.3611 | 0.3328 | 0.1719 | 0.1342 |
0.5881 | 30.64 | 1440 | 0.5969 | 0.7002 | 0.3462 | 0.3540 | 0.1507 | 0.1491 |
0.601 | 31.06 | 1460 | 0.5970 | 0.6994 | 0.3328 | 0.3666 | 0.1381 | 0.1625 |
0.5759 | 31.49 | 1480 | 0.5971 | 0.6986 | 0.3375 | 0.3611 | 0.1436 | 0.1578 |
0.5738 | 31.91 | 1500 | 0.5969 | 0.7002 | 0.3454 | 0.3548 | 0.1499 | 0.1499 |
0.5576 | 32.34 | 1520 | 0.5983 | 0.6931 | 0.3493 | 0.3438 | 0.1609 | 0.1460 |
0.58 | 32.77 | 1540 | 0.5976 | 0.7009 | 0.3359 | 0.3650 | 0.1397 | 0.1593 |
0.5798 | 33.19 | 1560 | 0.5980 | 0.7017 | 0.3469 | 0.3548 | 0.1499 | 0.1484 |
0.5802 | 33.62 | 1580 | 0.5988 | 0.6954 | 0.3477 | 0.3477 | 0.1570 | 0.1476 |
0.587 | 34.04 | 1600 | 0.5997 | 0.6931 | 0.3532 | 0.3399 | 0.1648 | 0.1421 |
0.5499 | 34.47 | 1620 | 0.6081 | 0.6797 | 0.3830 | 0.2967 | 0.2080 | 0.1122 |
0.5878 | 34.89 | 1640 | 0.5989 | 0.6970 | 0.3438 | 0.3532 | 0.1515 | 0.1515 |
0.5855 | 35.32 | 1660 | 0.6073 | 0.6829 | 0.3815 | 0.3014 | 0.2033 | 0.1138 |
0.5836 | 35.74 | 1680 | 0.5977 | 0.7002 | 0.3359 | 0.3642 | 0.1405 | 0.1593 |
0.5576 | 36.17 | 1700 | 0.5984 | 0.6986 | 0.3399 | 0.3587 | 0.1460 | 0.1554 |
0.5929 | 36.6 | 1720 | 0.6035 | 0.6907 | 0.3697 | 0.3210 | 0.1837 | 0.1256 |
0.5672 | 37.02 | 1740 | 0.6023 | 0.6923 | 0.3705 | 0.3218 | 0.1829 | 0.1248 |
0.5774 | 37.45 | 1760 | 0.5986 | 0.6947 | 0.3509 | 0.3438 | 0.1609 | 0.1444 |
0.5785 | 37.87 | 1780 | 0.5990 | 0.6962 | 0.3195 | 0.3768 | 0.1279 | 0.1758 |
0.5885 | 38.3 | 1800 | 0.5979 | 0.6994 | 0.3375 | 0.3619 | 0.1429 | 0.1578 |
0.5449 | 38.72 | 1820 | 0.6030 | 0.6923 | 0.3713 | 0.3210 | 0.1837 | 0.1240 |
0.5857 | 39.15 | 1840 | 0.5990 | 0.7009 | 0.3328 | 0.3681 | 0.1366 | 0.1625 |
0.5839 | 39.57 | 1860 | 0.6003 | 0.6907 | 0.3548 | 0.3359 | 0.1688 | 0.1405 |
0.5806 | 40.0 | 1880 | 0.5976 | 0.6962 | 0.3414 | 0.3548 | 0.1499 | 0.1538 |
0.5692 | 40.43 | 1900 | 0.5976 | 0.7025 | 0.3399 | 0.3626 | 0.1421 | 0.1554 |
0.593 | 40.85 | 1920 | 0.5984 | 0.6947 | 0.3430 | 0.3516 | 0.1531 | 0.1523 |
0.5736 | 41.28 | 1940 | 0.5992 | 0.6931 | 0.3556 | 0.3375 | 0.1672 | 0.1397 |
0.5653 | 41.7 | 1960 | 0.5978 | 0.6970 | 0.3438 | 0.3532 | 0.1515 | 0.1515 |
0.5631 | 42.13 | 1980 | 0.6006 | 0.6947 | 0.3603 | 0.3344 | 0.1703 | 0.1350 |
0.5794 | 42.55 | 2000 | 0.5983 | 0.6994 | 0.3336 | 0.3658 | 0.1389 | 0.1617 |
0.5876 | 42.98 | 2020 | 0.5984 | 0.6939 | 0.3422 | 0.3516 | 0.1531 | 0.1531 |
0.5726 | 43.4 | 2040 | 0.6005 | 0.6962 | 0.3634 | 0.3328 | 0.1719 | 0.1319 |
0.566 | 43.83 | 2060 | 0.5982 | 0.6970 | 0.3242 | 0.3728 | 0.1319 | 0.1711 |
0.5603 | 44.26 | 2080 | 0.5994 | 0.6947 | 0.3579 | 0.3367 | 0.1680 | 0.1374 |
0.5697 | 44.68 | 2100 | 0.6037 | 0.6892 | 0.3728 | 0.3163 | 0.1884 | 0.1224 |
0.5624 | 45.11 | 2120 | 0.5981 | 0.7002 | 0.3297 | 0.3705 | 0.1342 | 0.1656 |
0.5648 | 45.53 | 2140 | 0.5979 | 0.6962 | 0.3422 | 0.3540 | 0.1507 | 0.1531 |
0.578 | 45.96 | 2160 | 0.6024 | 0.6907 | 0.3713 | 0.3195 | 0.1852 | 0.1240 |
0.5593 | 46.38 | 2180 | 0.5977 | 0.7002 | 0.3391 | 0.3611 | 0.1436 | 0.1562 |
0.5755 | 46.81 | 2200 | 0.5979 | 0.6978 | 0.3336 | 0.3642 | 0.1405 | 0.1617 |
0.59 | 47.23 | 2220 | 0.6046 | 0.6868 | 0.3736 | 0.3132 | 0.1915 | 0.1217 |
0.5648 | 47.66 | 2240 | 0.5997 | 0.6931 | 0.3564 | 0.3367 | 0.1680 | 0.1389 |
0.5812 | 48.09 | 2260 | 0.5979 | 0.6954 | 0.3336 | 0.3619 | 0.1429 | 0.1617 |
0.5796 | 48.51 | 2280 | 0.5979 | 0.6962 | 0.3336 | 0.3626 | 0.1421 | 0.1617 |
0.5701 | 48.94 | 2300 | 0.5981 | 0.6947 | 0.3454 | 0.3493 | 0.1554 | 0.1499 |
0.5807 | 49.36 | 2320 | 0.5988 | 0.6931 | 0.3501 | 0.3430 | 0.1617 | 0.1452 |
0.5836 | 49.79 | 2340 | 0.5990 | 0.6923 | 0.3501 | 0.3422 | 0.1625 | 0.1452 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1
- Datasets 2.10.1
- Tokenizers 0.13.2
- Downloads last month
- 33
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.
Dataset used to train henryscheible/gpt2_stereoset_classifieronly
Evaluation results
- Accuracy on stereosetvalidation set self-reported0.692