Spaces:
Running
Running
fix yaml color
Browse files
README.md
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
---
|
| 2 |
title: Indic Language Identifier
|
| 3 |
emoji: 🔤
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: "4.24.0"
|
| 8 |
app_file: app.py
|
|
@@ -20,29 +20,4 @@ Bengali · English · Gujarati · Hindi · Kannada · Malayalam · Marathi · Od
|
|
| 20 |
|
| 21 |
```bash
|
| 22 |
pip install -r requirements.txt
|
| 23 |
-
python app.py
|
| 24 |
-
```
|
| 25 |
-
|
| 26 |
-
## Deploy on Hugging Face Spaces
|
| 27 |
-
|
| 28 |
-
1. Create a new Space at https://huggingface.co/spaces
|
| 29 |
-
2. Choose **Gradio** as the SDK
|
| 30 |
-
3. Upload these files:
|
| 31 |
-
- `app.py`
|
| 32 |
-
- `requirements.txt`
|
| 33 |
-
- `src/` folder (entire directory)
|
| 34 |
-
- `artifacts/models/` folder (if you have trained `.pkl` files)
|
| 35 |
-
|
| 36 |
-
4. HF Spaces will auto-install dependencies and launch `app.py`
|
| 37 |
-
|
| 38 |
-
## Model Priority
|
| 39 |
-
The app loads the best available model in this order:
|
| 40 |
-
1. `artifacts/models/ensemble.pkl` — full ensemble (CharNgram + FastText + Heuristic)
|
| 41 |
-
2. `artifacts/models/ngram_model.pkl` — CharNgram only
|
| 42 |
-
3. Unicode Heuristic fallback — no file needed, works out of the box
|
| 43 |
-
|
| 44 |
-
## Train Your Own Models
|
| 45 |
-
```bash
|
| 46 |
-
python train_pipeline.py --model all --max_per_lang 30000
|
| 47 |
-
```
|
| 48 |
-
Requires the AI4Bharat/Pralekha dataset (auto-downloaded from HuggingFace Hub).
|
|
|
|
| 1 |
---
|
| 2 |
title: Indic Language Identifier
|
| 3 |
emoji: 🔤
|
| 4 |
+
colorFrom: blue
|
| 5 |
+
colorTo: green
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: "4.24.0"
|
| 8 |
app_file: app.py
|
|
|
|
| 20 |
|
| 21 |
```bash
|
| 22 |
pip install -r requirements.txt
|
| 23 |
+
python app.py
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|