File size: 1,358 Bytes
ee26d29 f1e431e ee26d29 c47238c ee26d29 c46c77f ee26d29 c46c77f ee26d29 9f09438 759f456 9f09438 759f456 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 | ---
title: Pathshala AI
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 4.44.1
app_file: app.py
python_version: 3.11
pinned: false
---
# Pathshala AI
Pathshala AI is a bilingual AI tutor demo for rural primary students in Nepal.
This Hugging Face Space supports:
- Uploading a text-based PDF textbook directly in the Space
- Asking questions in English, Nepali, or romanized Nepali
- Retrieving relevant textbook portions from the uploaded PDF
- Showing a simple English answer and Nepali explanation
- Generating Nepali quiz questions
- Basic quiz grading
For the full web-app workflow, deploy the FastAPI backend separately and add a Space variable named `BACKEND_URL`.
Without `BACKEND_URL`, the Space can still run the same style of workflow locally. Add these Space secrets/variables to match the web app more closely:
- `LLM_BASE_URL`, `LLM_API_KEY`, `LLM_MODEL` for the AMD/vLLM tutor
- `TRANSLATION_PROVIDER=gemini`, `GEMINI_API_KEY`, `GEMINI_MODEL` for Nepali adaptation and romanized question normalization
- `TRANSLATION_PROVIDER=openai`, `OPENAI_API_KEY`, `OPENAI_MODEL` if you want to use OpenAI for Nepali adaptation instead
- `OCR_PROVIDER=gemini`, `OCR_MAX_PAGES=5` for scanned or custom-font PDFs
Use `LLM_MODEL=Qwen/Qwen2.5-7B-Instruct` to match the project default unless your vLLM endpoint exposes a different model name.
|