20230831144955
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: 0.6197
- Accuracy: 0.5
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 | 0.6175 | 0.5 |
0.6359 | 2.0 | 680 | 0.6236 | 0.5 |
0.635 | 3.0 | 1020 | 0.6211 | 0.5 |
0.635 | 4.0 | 1360 | 0.6253 | 0.5 |
0.6306 | 5.0 | 1700 | 0.6421 | 0.5 |
0.6268 | 6.0 | 2040 | 0.6297 | 0.5 |
0.6268 | 7.0 | 2380 | 0.6351 | 0.5 |
0.6314 | 8.0 | 2720 | 0.6053 | 0.5 |
0.6135 | 9.0 | 3060 | 0.6185 | 0.5 |
0.6135 | 10.0 | 3400 | 0.6316 | 0.5 |
0.6245 | 11.0 | 3740 | 0.6219 | 0.5 |
0.6198 | 12.0 | 4080 | 0.6203 | 0.5 |
0.6198 | 13.0 | 4420 | 0.6516 | 0.5 |
0.6151 | 14.0 | 4760 | 0.6231 | 0.5 |
0.6223 | 15.0 | 5100 | 0.6235 | 0.5 |
0.6223 | 16.0 | 5440 | 0.6204 | 0.5 |
0.6216 | 17.0 | 5780 | 0.6225 | 0.5 |
0.6168 | 18.0 | 6120 | 0.6176 | 0.5 |
0.6168 | 19.0 | 6460 | 0.6204 | 0.5 |
0.6179 | 20.0 | 6800 | 0.6179 | 0.5 |
0.6169 | 21.0 | 7140 | 0.6193 | 0.5 |
0.6169 | 22.0 | 7480 | 0.6414 | 0.5 |
0.6206 | 23.0 | 7820 | 0.6196 | 0.5 |
0.6181 | 24.0 | 8160 | 0.6248 | 0.5 |
0.6269 | 25.0 | 8500 | 0.6173 | 0.5 |
0.6269 | 26.0 | 8840 | 0.6234 | 0.5 |
0.6201 | 27.0 | 9180 | 0.6239 | 0.5 |
0.6162 | 28.0 | 9520 | 0.6182 | 0.5 |
0.6162 | 29.0 | 9860 | 0.6260 | 0.5 |
0.6166 | 30.0 | 10200 | 0.6190 | 0.5 |
0.6159 | 31.0 | 10540 | 0.6192 | 0.5 |
0.6159 | 32.0 | 10880 | 0.6261 | 0.5 |
0.6158 | 33.0 | 11220 | 0.6295 | 0.5 |
0.6166 | 34.0 | 11560 | 0.6238 | 0.5 |
0.6166 | 35.0 | 11900 | 0.6221 | 0.5 |
0.6163 | 36.0 | 12240 | 0.6198 | 0.5 |
0.6177 | 37.0 | 12580 | 0.6177 | 0.5 |
0.6177 | 38.0 | 12920 | 0.6202 | 0.5 |
0.6158 | 39.0 | 13260 | 0.6231 | 0.5 |
0.6147 | 40.0 | 13600 | 0.6209 | 0.5 |
0.6147 | 41.0 | 13940 | 0.6191 | 0.5 |
0.6173 | 42.0 | 14280 | 0.6195 | 0.5 |
0.6129 | 43.0 | 14620 | 0.6213 | 0.5 |
0.6129 | 44.0 | 14960 | 0.6245 | 0.5 |
0.6173 | 45.0 | 15300 | 0.6235 | 0.5 |
0.6128 | 46.0 | 15640 | 0.6184 | 0.5 |
0.6128 | 47.0 | 15980 | 0.6252 | 0.5 |
0.6174 | 48.0 | 16320 | 0.6216 | 0.5 |
0.6157 | 49.0 | 16660 | 0.6248 | 0.5 |
0.6151 | 50.0 | 17000 | 0.6191 | 0.5 |
0.6151 | 51.0 | 17340 | 0.6212 | 0.5 |
0.6132 | 52.0 | 17680 | 0.6197 | 0.5 |
0.6173 | 53.0 | 18020 | 0.6233 | 0.5 |
0.6173 | 54.0 | 18360 | 0.6223 | 0.5 |
0.6132 | 55.0 | 18700 | 0.6173 | 0.5 |
0.6129 | 56.0 | 19040 | 0.6218 | 0.5 |
0.6129 | 57.0 | 19380 | 0.6178 | 0.5 |
0.614 | 58.0 | 19720 | 0.6239 | 0.5 |
0.616 | 59.0 | 20060 | 0.6258 | 0.5 |
0.616 | 60.0 | 20400 | 0.6181 | 0.5 |
0.6136 | 61.0 | 20740 | 0.6195 | 0.5 |
0.6132 | 62.0 | 21080 | 0.6205 | 0.5 |
0.6132 | 63.0 | 21420 | 0.6177 | 0.5 |
0.6121 | 64.0 | 21760 | 0.6221 | 0.5 |
0.6164 | 65.0 | 22100 | 0.6190 | 0.5 |
0.6164 | 66.0 | 22440 | 0.6225 | 0.5 |
0.6073 | 67.0 | 22780 | 0.6205 | 0.5 |
0.615 | 68.0 | 23120 | 0.6189 | 0.5 |
0.615 | 69.0 | 23460 | 0.6188 | 0.5 |
0.6136 | 70.0 | 23800 | 0.6200 | 0.5 |
0.6127 | 71.0 | 24140 | 0.6197 | 0.5 |
0.6127 | 72.0 | 24480 | 0.6213 | 0.5 |
0.6111 | 73.0 | 24820 | 0.6197 | 0.5 |
0.6133 | 74.0 | 25160 | 0.6215 | 0.5 |
0.6113 | 75.0 | 25500 | 0.6197 | 0.5 |
0.6113 | 76.0 | 25840 | 0.6209 | 0.5 |
0.6124 | 77.0 | 26180 | 0.6192 | 0.5 |
0.6112 | 78.0 | 26520 | 0.6200 | 0.5 |
0.6112 | 79.0 | 26860 | 0.6198 | 0.5 |
0.612 | 80.0 | 27200 | 0.6197 | 0.5 |
Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3
- Downloads last month
- 4
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.