Ary4r/software-defect-prediction-api

fix: pin starlette<0.39.0 and set server_name=0.0.0.0 in demo.launch()

fix: update fastapi requirement to >=0.115.0 to fix gradio Pydantic v2 compatibility error

fix: update pydantic requirement for python 3.14 compatibility

fix: change scikit-learn version requirement to >=1.7.0 for Python 3.10 compatibility

fix: remove share=True and update scikit-learn version requirement

Update app.py and sync with Hugging Face Space

Update app.py demo.launch(share=True)

Final push: Enable share=True in demo.launch()

Support both /docs and /docs/ trailing slash URL formats

Decorate predict_defect_gradio with @spaces.GPU for HF ZeroGPU compatibility

Remove spaces GPU decorator to fix ZeroGPU PyTorch emulation error during prediction

Add /docs and /openapi.json routes to Gradio demo.app

Add dummy @spaces.GPU function to resolve ZeroGPU runtime check

Set server_name='0.0.0.0' and remove share=True for clean HF Space server launch

Update app.py demo.launch(share=True)

Add share=True to demo.launch()

Upgrade Gradio sdk_version to 4.44.1 in README.md

Move import gradio as gr to line 4 at top of app.py

Launch blocking Gradio 4.44.0 server on 0.0.0.0:7860 to keep HF Space active

Remove gradio from requirements.txt to resolve version conflict with README.md

Update dependencies and push to Hugging Face Space

Pin fastapi==0.104.1 and pydantic<2.6.0 to fix FieldInfo AttributeError

Pin starlette<0.28.0 to fix TemplateResponse unhashable dict error

Use automatic demo.launch() HF Space launcher

Fix NameError by placing import gradio as gr at top of app.py

Pass app object directly to uvicorn.run(app) to prevent re-import shutdown

Run uvicorn.run('app:app', host='0.0.0.0', port=7860) to host Gradio 4.44.0 LTS server

Launch blocking Gradio 4.44.0 LTS server on port 7860 to prevent Exit code 0

Standardize deployment to Gradio 4.44.0 LTS + FastAPI on HF Space

Force CPU mode CUDA_VISIBLE_DEVICES='' and import spaces before PyTorch

Set GRADIO_SSR=0 env var to disable Node SSR proxy before importing gradio

Disable Node.js SSR proxy in demo.launch(ssr=False) for direct Python FastAPI server

Move import gradio as gr to top of app.py before PyTorch initialization

Merge FastAPI router into demo.app and launch blocking server on port 7860

Export top-level app object for HF Space supervisor without uvicorn.run

Run uvicorn server directly on mounted app to serve port 7860

Explicitly set server_name 0.0.0.0 and server_port 7860 in demo.launch()

Execute demo.launch() unconditionally to keep HF Space module import alive

Use demo.launch() to manage Gradio server lifecycle on HF Space

Start uvicorn server on port 7860 to keep app process alive

Remove redundant uvicorn.run to prevent port 7860 binding error

Fix import conflict between app.py file and app/ folder

Remove pinned gradio and huggingface_hub to allow HF Space Gradio 6.25.0

Pin huggingface_hub<0.22.0 to resolve HfFolder import issue

Update Gradio version to 4.44.0 to fix huggingface_hub HfFolder import error

Add README.md with HF Space metadata config

Deploy software defect prediction API to Hugging Face Spaces