Downloads · 30 days
61
28% of all-time downloads
deberta-mlm-test
domenicrosati/deberta-mlm-test
deberta-mlm-test is a fill-mask model from domenicrosati. Use it when you need the model to fill a missing word. It is set up for transformers. The card lists the license as mit.
should probably proofread and complete it, then remove this comment. --
All-time downloads
216
Public
Repo size
8.2 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin284 MB · 96%
At a glance
- Task
- Fill-Mask
- Library
- transformers
- License
- mit
- Model type
- deberta-v2
- Access
- Public
- Created
- Jun 28, 2022
- Updated
- Jun 29, 2022
- SHA
- 94d7b781
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. -->
deberta-mlm-test
This model is a fine-tuned version of microsoft/deberta-v3-xsmall on the None dataset. It achieves the following results on the evaluation set:
- Loss: 3.2792
- Accuracy: 0.4766
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
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 4.4466 | 1.0 | 2067 | 4.1217 | 0.3847 |
| 3.9191 | 2.0 | 4134 | 3.6562 | 0.4298 |
| 3.6397 | 3.0 | 6201 | 3.4417 | 0.4550 |
| 3.522 | 4.0 | 8268 | 3.3239 | 0.4692 |
| 3.4504 | 5.0 | 10335 | 3.2792 | 0.4766 |
Framework versions
- Transformers 4.20.1
- Pytorch 1.11.0a0+17540c5
- Datasets 2.3.2
- Tokenizers 0.12.1
- Task
- Fill-Mask
- Library
- transformers
- Type
- deberta-v2
- License
- mit
- Created
- Jun 28, 2022
- Updated
- Jun 29, 2022