Update model card for Polish
Browse files
README.md
CHANGED
|
@@ -1,24 +1,62 @@
|
|
| 1 |
-
---
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pipeline_tag: fill-mask
|
| 3 |
+
language: pol
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- trimmed
|
| 7 |
+
library_name: transformers
|
| 8 |
+
base_model: google/mt5-base
|
| 9 |
+
base_model_relation: quantized
|
| 10 |
+
datasets:
|
| 11 |
+
- lbourdois/fineweb-2-trimming
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# mt5-base-pol-32768
|
| 15 |
+
This model is a **61.64% smaller** version of [google/mt5-base](https://huggingface.co/google/mt5-base) optimized for **Polish** language via vocabulary size reduction using the [trimming](https://huggingface.co/blog/lbourdois/introduction-to-trimming) method.
|
| 16 |
+
This trimmed model should perform similarly to the original model with only 16,384 tokens and a much smaller memory footprint. However, it may not perform well for other languages as tokens not commonly used in the selected languages were removed from the vocabulary.
|
| 17 |
+
|
| 18 |
+
## Model Statistics
|
| 19 |
+
| Metric | Original | Trimmed | Reduction |
|
| 20 |
+
|--------|----------|---------|-----------|
|
| 21 |
+
| **Vocabulary size** | 250,112 tokens | 16,384 tokens | **93.45%** |
|
| 22 |
+
| **Model size** | 300,176,768 params | 223,395,072 params | **61.64%** |
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
## Mining Dataset Statistics
|
| 27 |
+
- **Number of texts used for mining**: 200,000 texts
|
| 28 |
+
- **Dataset**: [lbourdois/fineweb-2-trimming](https://huggingface.co/datasets/lbourdois/fineweb-2-trimming)
|
| 29 |
+
|
| 30 |
+
## Usage
|
| 31 |
+
```python
|
| 32 |
+
from transformers import AutoModel, AutoTokenizer
|
| 33 |
+
|
| 34 |
+
model_name = "alphaedge-ai/mt5-base-pol-16384"
|
| 35 |
+
model = AutoModel.from_pretrained(model_name)
|
| 36 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 37 |
+
```
|
| 38 |
+
|
| 39 |
+
## Citations
|
| 40 |
+
|
| 41 |
+
#### mT5
|
| 42 |
+
```
|
| 43 |
+
@misc{xue2021mt5massivelymultilingualpretrained,
|
| 44 |
+
title={mT5: A massively multilingual pre-trained text-to-text transformer},
|
| 45 |
+
author={Linting Xue and Noah Constant and Adam Roberts and Mihir Kale and Rami Al-Rfou and Aditya Siddhant and Aditya Barua and Colin Raffel},
|
| 46 |
+
year={2021},
|
| 47 |
+
eprint={2010.11934},
|
| 48 |
+
archivePrefix={arXiv},
|
| 49 |
+
primaryClass={cs.CL},
|
| 50 |
+
url={https://arxiv.org/abs/2010.11934},
|
| 51 |
+
}
|
| 52 |
+
```
|
| 53 |
+
|
| 54 |
+
#### Trimming blog post
|
| 55 |
+
```
|
| 56 |
+
@misc{hf_blogpost_trimming,
|
| 57 |
+
title={Introduction to Trimming},
|
| 58 |
+
author={Loïck BOURDOIS and Tom AARSEN and Bram VANROY and Christopher AKIKI and Woojun JUNG and Manuel ROMERO and Prithiv SAKTHI},
|
| 59 |
+
year={2026},
|
| 60 |
+
url={https://huggingface.co/blog/lbourdois/introduction-to-trimming},
|
| 61 |
+
}
|
| 62 |
+
```
|