Files
voicebox/backend/routes
jamiepineandcapy-ai-staging[bot] b788dc383c fix(generation): make post-generation cleanup best-effort and cover /generate/stream
Wrap empty_device_cache in release_generation_memory(), which logs and
swallows cleanup failures so a poisoned CUDA context cannot replace a
finished generation's result, and call it from the streaming endpoint
too, which drives generate_chunked directly.
2026-10-04 00:01:18 +00:00
..