You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Bound streaming upstream stalls with configurable stream progress watchdogs.
Added CLLAMA_STREAM_FIRST_BYTE_TIMEOUT_MS (default 300000) for streaming header/first-body waits.
Added CLLAMA_STREAM_IDLE_TIMEOUT_MS (default 120000) for inter-chunk idle waits.
Preserve safe fallback semantics: retry the next declared candidate only before downstream bytes are committed; emit an SSE error and do not retry after partial stream output.
Image
ghcr.io/mostlydev/cllama:v0.7.3 published for linux/amd64 and linux/arm64.