vaibhavrakshe161/medqa-hindi

Update evaluation results with latest metrics

Remove deleted result generator script

Fix confidence scoring; return 0.0 on error without hardcoded 0.75

fp16 configs changed

QLORA Loading fixed

QLORA Loading fixed

QLORA Loading fixed

Fix qlora CPU loading - remove quantization on CPU path

Fix Model loading issue

Fix Model loading issue

Fix transformers version - incompatible with torch 2.1.2

Unpin HF stack versions to resolve dependency conflicts

Fix huggingface-hub version conflict with transformers 4.57.6

Pin transformers to 4.57.6 to match training version

Remove duplicate ProxyHeadersMiddleware

Fix HTTPS proxy headers for HF Spaces

Fix RedirectResponse import + CPU fallback for no GPU

Fix frontend path resolution in Docker

Redirect root to frontend

start script changed to add username

Add HF Spaces metadata to README

Remove data files

README.md included with CI tag

Fix CI: add peft and accelerate to requirements-ci.txt

Modified requirements-ci.txt

Add conftest.py in test suite

Add test suite + CI pipeline (53 tests passing)

UI and path issues fixed

Initial commit