Instructions to use failed09/bashkir-fasttext with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- fastText
How to use failed09/bashkir-fasttext with fastText:
from huggingface_hub import hf_hub_download import fasttext model = fasttext.load_model(hf_hub_download("failed09/bashkir-fasttext", "model.bin")) - Notebooks
- Google Colab
- Kaggle
Card: use generic source wording
Browse files- README.md +5 -5
- SHA256SUMS +1 -1
README.md
CHANGED
|
@@ -22,8 +22,8 @@ tags:
|
|
| 22 |
|
| 23 |
## Overview
|
| 24 |
|
| 25 |
-
Compact subword word embeddings for Bashkir (`ba`), trained on a
|
| 26 |
-
|
| 27 |
quantized model (`bashkir_fasttext_300d_mini.bin`) intended for semantic search,
|
| 28 |
lexical tools and lightweight NLP experiments on CPU, mobile and edge devices.
|
| 29 |
|
|
@@ -31,7 +31,7 @@ lexical tools and lightweight NLP experiments on CPU, mobile and edge devices.
|
|
| 31 |
| --- | --- |
|
| 32 |
| Task | Feature extraction / word embeddings |
|
| 33 |
| Default artifact | `bashkir_fasttext_300d_mini.bin` |
|
| 34 |
-
| Source | A monolingual Bashkir-language
|
| 35 |
| Version / license | v0.1-beta / Apache-2.0 |
|
| 36 |
|
| 37 |
## Contents
|
|
@@ -75,8 +75,8 @@ monolingual Bashkir text → Unicode NFC tokenization → skip-gram 300d (char n
|
|
| 75 |
→ prune (top-50k words, top-100k subwords) → PQ FP16 → mini .bin
|
| 76 |
```
|
| 77 |
|
| 78 |
-
The model was trained from a
|
| 79 |
-
|
| 80 |
[`config.json`](config.json). Text preparation uses normalized Unicode tokens and
|
| 81 |
preserves Bashkir-specific letters. This release is a compact derivative of the
|
| 82 |
full 300-dimensional model: it retains subword information while reducing the
|
|
|
|
| 22 |
|
| 23 |
## Overview
|
| 24 |
|
| 25 |
+
Compact subword word embeddings for Bashkir (`ba`), trained on a monolingual
|
| 26 |
+
Bashkir-language dataset. The release ships a single
|
| 27 |
quantized model (`bashkir_fasttext_300d_mini.bin`) intended for semantic search,
|
| 28 |
lexical tools and lightweight NLP experiments on CPU, mobile and edge devices.
|
| 29 |
|
|
|
|
| 31 |
| --- | --- |
|
| 32 |
| Task | Feature extraction / word embeddings |
|
| 33 |
| Default artifact | `bashkir_fasttext_300d_mini.bin` |
|
| 34 |
+
| Source | A monolingual Bashkir-language dataset |
|
| 35 |
| Version / license | v0.1-beta / Apache-2.0 |
|
| 36 |
|
| 37 |
## Contents
|
|
|
|
| 75 |
→ prune (top-50k words, top-100k subwords) → PQ FP16 → mini .bin
|
| 76 |
```
|
| 77 |
|
| 78 |
+
The model was trained from a monolingual Bashkir-language dataset.
|
| 79 |
+
Corpus size, source reference and training parameters are recorded in
|
| 80 |
[`config.json`](config.json). Text preparation uses normalized Unicode tokens and
|
| 81 |
preserves Bashkir-specific letters. This release is a compact derivative of the
|
| 82 |
full 300-dimensional model: it retains subword information while reducing the
|
SHA256SUMS
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
23839280d4c0e4a3c7392795e4b8ec7487b219c5a7cc8c6063802f92ec120e4f META.json
|
| 2 |
-
|
| 3 |
51c65280ece23c79da7fc3f655da7eb9bbee3b8dd3684086d58e8ddabe0096e6 bashkir_fasttext_300d_mini.bin
|
| 4 |
7d94cf89587e637d4df24a34a4d9608412a9e77d7e99e50088e53783b781d37d config.json
|
|
|
|
| 1 |
23839280d4c0e4a3c7392795e4b8ec7487b219c5a7cc8c6063802f92ec120e4f META.json
|
| 2 |
+
434bbbcad59f1c6bf946c331c59bca3370a9e29ac1deb79e35225e0a900fe539 README.md
|
| 3 |
51c65280ece23c79da7fc3f655da7eb9bbee3b8dd3684086d58e8ddabe0096e6 bashkir_fasttext_300d_mini.bin
|
| 4 |
7d94cf89587e637d4df24a34a4d9608412a9e77d7e99e50088e53783b781d37d config.json
|