Commit History

Add Firebase auth verification, /api/me endpoint, and API_DOC_FIREBASE.txt
f2895a9

LogicGoInfotechSpaces commited on

Fix: use @api.post for /api/image-edit endpoint registration
72a96c5

LogicGoInfotechSpaces commited on

Add /api/image-edit endpoint for image upload and result API
be32143

LogicGoInfotechSpaces commited on

Remove gradio, fastapi, uvicorn—streamlit-only requirements for Space
60462d4

LogicGoInfotechSpaces commited on

Add Streamlit UI for Smile Changer (no API endpoints, Gradio logic reused, ready for Streamlit deployment)
88f684a

LogicGoInfotechSpaces commited on

Add unauthenticated GET /health root health check endpoint for integrations
d884e14

LogicGoInfotechSpaces commited on

API: serve FastAPI at root; mount UI at /app; add root index and /health alias for integration without proxy
a499383

LogicGoInfotechSpaces commited on

API: add startup self-check that hits /api/health and logs result
d42e603

LogicGoInfotechSpaces commited on

API: add /api/health (public) and /api/ping (auth) for diagnostics
e3f5da2

LogicGoInfotechSpaces commited on

Fix runtime: mount FastAPI + use uvicorn for __main__; add uvicorn dep
9473940

LogicGoInfotechSpaces commited on

API: add FastAPI endpoints, bearer auth, align default true; docs updated
57bfe5c

LogicGoInfotechSpaces commited on

Comprehensive model and library optimization - all components verified and error-handled
0154081

LogicGoInfotechSpaces commited on

Add debugging for face alignment to diagnose smile editing issues
5fd5d22

LogicGoInfotechSpaces commited on

Fix beard/mustache editing - clarify that negative values add facial hair
e89dadb

LogicGoInfotechSpaces commited on

Make facer and fpie dependencies optional for face alignment
600df15

LogicGoInfotechSpaces commited on

Fix hardcoded CUDA call in StyleCLIP fused_act for curly hair editing
14cc518

LogicGoInfotechSpaces commited on

Fix Gradio slider update and CUDA device issues for all attributes
6c0e8ac

LogicGoInfotechSpaces commited on

Fix file path issue - download to current directory instead of nested pretrained_models
5050e58

LogicGoInfotechSpaces commited on

Improve debugging with better error handling and file size info
07eb20a

LogicGoInfotechSpaces commited on

Add debugging to list all downloaded files and identify naming issues
138df91

LogicGoInfotechSpaces commited on

Add delay and verification after model download to handle timing issues
68a881b

LogicGoInfotechSpaces commited on

Fix CUDA operations to check device: only use fused ops when input is on CUDA
80d11e4

LogicGoInfotechSpaces commited on

Add error handling for missing checkpoint files in load_weights methods
bfac43d

LogicGoInfotechSpaces commited on

Add device detection and CPU fallback directly in FSEFull and FSEInverter constructors
372a1fe

LogicGoInfotechSpaces commited on

Fix device handling in FSEFull and FSEInverter: convert device string to torch.device object
87d55be

LogicGoInfotechSpaces commited on

Add error handling for missing StyleGAN discriminator weights in both FSEFull and FSEInverter classes
51cbd8f

LogicGoInfotechSpaces commited on

Fix DeltaEditor device handling: add device parameter, error handling for missing files, replace .cuda() with .to(device)
96c63c3

LogicGoInfotechSpaces commited on

Fix CLIP model device handling: pass device to StyleCLIPGlobalDirection and use .to(device) instead of .cuda()
ad8ad02

LogicGoInfotechSpaces commited on

Avoid CUDA init on Spaces: map tensors to device, no .cuda() in LatentEditor
cfa4803

LogicGoInfotechSpaces commited on

Pass HF_TOKEN to snapshot_download to access private model repo
5d50fba

LogicGoInfotechSpaces commited on

Import spaces before torch-related modules and set CUDA env early (ZeroGPU fix)
8195245

LogicGoInfotechSpaces commited on

Add spaces.GPU no-op handler and dependency to satisfy Spaces startup check
1641332

LogicGoInfotechSpaces commited on

Make metrics import optional across inference to avoid piq/pytorch_fid on Space
0340b9d

LogicGoInfotechSpaces commited on

Make metrics import optional to drop pytorch_fid requirement for UI
c7e8984

LogicGoInfotechSpaces commited on

CPU fallback: avoid CUDA JIT by disabling fused ops when build unavailable; force CPU
704a5f0

LogicGoInfotechSpaces commited on