Downloads · 30 days
21
2% of all-time downloads
bert-base-multilingual-cased-urgency
Narshion/bert-base-multilingual-cased-urgency
bert-base-multilingual-cased-urgency is a fill-mask model from Narshion. 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
1.1K
Public
Repo size
1.4 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin712 MB · 100%
At a glance
- Task
- Fill-Mask
- Library
- transformers
- Model type
- bert
- Access
- Public
- Created
- Mar 2, 2022
- Updated
- Sep 15, 2021
- SHA
- 3ec2e67d
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. -->
bert-base-multilingual-cased-urgency
This model is a fine-tuned version of bert-base-multilingual-cased on the mWACH NEO dataset. It achieves the following results on the evaluation set:
- Loss: 2.2797
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: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 4.1408 | 1.0 | 5659 | 3.6705 |
| 2.8777 | 2.0 | 11318 | 2.5536 |
| 2.561 | 3.0 | 16977 | 2.2740 |
Framework versions
- Transformers 4.10.0
- Pytorch 1.9.0+cu102
- Datasets 1.11.0
- Tokenizers 0.10.3
- Task
- Fill-Mask
- Library
- transformers
- Type
- bert
- Created
- Mar 2, 2022
- Updated
- Sep 15, 2021