Downloads · 30 days
107
14% of all-time downloads
DNADebertaSentencepiece30k_continuation
Vlasta/DNADebertaSentencepiece30k_continuation
DNADebertaSentencepiece30k_continuation is a fill-mask model from Vlasta. Use it when you need the model to fill a missing word. It is set up for transformers.
should probably proofread and complete it, then remove this comment. --
All-time downloads
746
Public
Repo size
9.9 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin266 MB · 100%
At a glance
- Task
- Fill-Mask
- Library
- transformers
- Model type
- deberta
- Access
- Public
- Created
- Jul 31, 2022
- Updated
- Aug 4, 2022
- SHA
- 2523a0c8
Try a prompt
Model card
From the Hugging Face model README
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
DNADebertaSentencepiece30k_continuation
This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:
- Loss: 6.0813
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 6.4099 | 0.41 | 5000 | 6.3686 |
| 6.4014 | 0.81 | 10000 | 6.3544 |
| 6.3816 | 1.22 | 15000 | 6.3338 |
| 6.3652 | 1.62 | 20000 | 6.3161 |
| 6.3477 | 2.03 | 25000 | 6.2981 |
| 6.3305 | 2.44 | 30000 | 6.2851 |
| 6.3173 | 2.84 | 35000 | 6.2725 |
| 6.306 | 3.25 | 40000 | 6.2559 |
| 6.2903 | 3.66 | 45000 | 6.2447 |
| 6.2806 | 4.06 | 50000 | 6.2342 |
| 6.2654 | 4.47 | 55000 | 6.2213 |
| 6.2592 | 4.87 | 60000 | 6.2101 |
| 6.2481 | 5.28 | 65000 | 6.2023 |
| 6.2394 | 5.69 | 70000 | 6.1929 |
| 6.2295 | 6.09 | 75000 | 6.1833 |
| 6.219 | 6.5 | 80000 | 6.1800 |
| 6.2143 | 6.91 | 85000 | 6.1698 |
| 6.2031 | 7.31 | 90000 | 6.1629 |
| 6.2036 | 7.72 | 95000 | 6.1523 |
| 6.1923 | 8.12 | 100000 | 6.1522 |
| 6.1868 | 8.53 | 105000 | 6.1426 |
| 6.1827 | 8.94 | 110000 | 6.1356 |
| 6.1767 | 9.34 | 115000 | 6.1322 |
| 6.1717 | 9.75 | 120000 | 6.1255 |
| 6.1649 | 10.16 | 125000 | 6.1221 |
| 6.1591 | 10.56 | 130000 | 6.1176 |
| 6.1562 | 10.97 | 135000 | 6.1111 |
| 6.15 | 11.37 | 140000 | 6.1063 |
| 6.1488 | 11.78 | 145000 | 6.1046 |
| 6.1449 | 12.19 | 150000 | 6.1023 |
| 6.1397 | 12.59 | 155000 | 6.0961 |
| 6.135 | 13.0 | 160000 | 6.0938 |
| 6.1315 | 13.41 | 165000 | 6.0891 |
| 6.1302 | 13.81 | 170000 | 6.0853 |
| 6.1295 | 14.22 | 175000 | 6.0838 |
| 6.1276 | 14.62 | 180000 | 6.0834 |
Framework versions
- Transformers 4.19.2
- Pytorch 1.11.0
- Datasets 2.2.2
- Tokenizers 0.12.1
- Task
- Fill-Mask
- Library
- transformers
- Type
- deberta
- Created
- Jul 31, 2022
- Updated
- Aug 4, 2022