|
--- |
|
tags: |
|
- mt5 |
|
- adapter-transformers |
|
- DefinitionGeneration |
|
- definition-modeling |
|
datasets: |
|
- jfeil/GermanDefinitionGeneration-Distillation |
|
license: gpl-3.0 |
|
language: |
|
- de |
|
base_model: |
|
- google/mt5-base |
|
--- |
|
|
|
# Adapter `jfeil/GermanDefinitionGeneration-Distillation` for google/mt5-base |
|
|
|
An [adapter](https://adapterhub.ml) for the `google/mt5-base` model that was trained on the [jfeil/GermanDefinitionGeneration-Distillation](https://huggingface.co/datasets/jfeil/GermanDefinitionGeneration-Distillation/) dataset. |
|
|
|
This adapter was created for usage with the **[Adapters](https://github.com/Adapter-Hub/adapters)** library. |
|
|
|
## Usage |
|
|
|
First, install `adapters`: |
|
|
|
``` |
|
pip install -U adapters |
|
``` |
|
|
|
Now, the adapter can be loaded and activated like this: |
|
|
|
```python |
|
from adapters import AutoAdapterModel |
|
|
|
model = AutoAdapterModel.from_pretrained("google/mt5-base") |
|
adapter_name = model.load_adapter("jfeil/GermanDefinitionGeneration-Distillation", source="hf", set_active=True) |
|
``` |
|
|
|
## Architecture & Training |
|
|
|
TODO |
|
|
|
## Optimal Inferencing Parameters |
|
|
|
* num_beams 5 |
|
* repetition_penalty 1.5 |
|
* Prompt: `<EXAMPLE_SENTENCE> Was ist die Definition von <EXAMPLE_WORD>?` |
|
|
|
## Evaluation results |
|
|
|
Tested on Wiktionary test set: |
|
|
|
| | MeteorScore | ROUGE-L | BERTscore-F1 | MoverScore | Sentence Embedding Score | |
|
|--------------|-------------|---------|--------------|------------|--------------------------| |
|
| Wiktionary | 0.175 | 0.187 | 0.751 | 0.568 | 0.478 | |
|
| Distillation | 0.103 | 0.109 | 0.675 | 0.536 | 0.369 | |
|
|
|
## Citation |
|
|
|
TODO |
