Drop a textbook PDF here
or click to select a PDF from your device
No PDF selected
Runtime: detecting…
OCR: ready on demand
Model: Llama 3.2 1B q4f16
Transformers.js v3.8.1
Tesseract.js v6.0.1
Device Compatibility Check
Test this browser/device before starting AI generation.
Not tested yet.
The check tests browser capabilities, WebGPU, and a real local LLM inference. Model downloads try Hugging Face first and a CORS-compatible mirror fallback if required. The optional AI runtime test performs a real short inference with the same local Llama model used by this application; the first run may download the model and take several minutes on slower devices.
Local source pipeline:
Each page is first inspected with PDF.js. Pages with insufficient selectable text are rendered locally and passed to Tesseract.js OCR. The resulting page text is merged into one source document before either LLM pass. No PDF text or OCR result is sent to an application backend.
Preparing…0%
Initializing…
Generated Questions
Upload a PDF and generate a question bank.
Quality Gate
Quality Gate report will appear here.