Downloads · 30 days
5
18% of all-time downloads
DistilBERT-POWO_Epiphyte_Scratch
ViktorDo/DistilBERT-POWO_Epiphyte_Scratch
DistilBERT-POWO_Epiphyte_Scratch is a text classification model from ViktorDo. Use it when you need a label for a piece of text. It is set up for transformers.
should probably proofread and complete it, then remove this comment. --
All-time downloads
28
Public
Repo size
3.7 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin267 MB · 100%
At a glance
- Task
- Text Classification
- Library
- transformers
- Model type
- distilbert
- Access
- Public
- Created
- Jan 24, 2023
- Updated
- Jan 24, 2023
- SHA
- 820bcd70
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. -->
DistilBERT-POWO_Epiphyte_Scratch
This model is a fine-tuned version of ViktorDo/DistilBERT-POWO_Scratch on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1020
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: 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: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.0725 | 1.0 | 2063 | 0.0894 |
| 0.0588 | 2.0 | 4126 | 0.0986 |
| 0.0533 | 3.0 | 6189 | 0.1020 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.1+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2
- Task
- Text Classification
- Library
- transformers
- Type
- distilbert
- Created
- Jan 24, 2023
- Updated
- Jan 24, 2023