Fix blank output: re-run inference without streamer on TypeError so output files are written c165326 akhaliq HF Staff commited on Jun 23
Fix streaming: annotate run_ocr as Iterator[dict] so Gradio registers 1 output 642469d akhaliq HF Staff commited on Jun 23
Stream OCR output: TextIteratorStreamer backend + client.submit() frontend ff6816b akhaliq HF Staff commited on Jun 23
Scale for ZeroGPU: one 60s GPU slot per page, CPU PDF explode, clean frontend c55a3e8 akhaliq HF Staff commited on Jun 23
Fix ZeroGPU usage: model.cuda() at module level, add GPU duration limits (300s/600s) 6149236 akhaliq HF Staff commited on Jun 23
Install exact torch/transformers pins at runtime in app.py to bypass build-time conflict 4b46da9 akhaliq HF Staff commited on Jun 23
Add ZeroGPU support: @spaces.GPU on inference fns, hardware: zero-gpu in README 07a166f akhaliq HF Staff commited on Jun 23
