Commit History

  1. Fix: monkey-patch without staticmethod wrapper, spaces before torch, ssr_mode=False

    Verified signature
    e244295
  2. Fix route injection: try demo.app directly, fallback to monkey-patch

    Verified signature
    e3186fc
  3. Add model filter param to benchmark endpoint

    Verified signature
    a76c80d
  4. Fix: build llama.cpp with static linking, find binary after build

    Verified signature
    2ab0e84
  5. Build llama.cpp from source instead of pre-built binaries

    Verified signature
    37096e6
  6. Add explicit libllama.so verification logging

    Verified signature
    dd8b7b8
  7. Fix: detect stale cache, re-extract if libs missing

    Verified signature
    13fbb4e
  8. Fix: extract all llama.cpp libs, set LD_LIBRARY_PATH

    Verified signature
    47355da
  9. Add benchmark test endpoint to diagnose issues

    Verified signature
    1da23c9
  10. Add explicit exit code and model verification logging

    Verified signature
    1114348
  11. Fix: clear benchmark cache, add binary verification and debug logging

    Verified signature
    680baeb
  12. Add debug logging to benchmark parsing

    Verified signature
    6c771e9
  13. Rewrite benchmark: use llama.cpp binary directly instead of llama-cpp-python

    Verified signature
    2456ebd
  14. Fix: try pre-built wheel first, fall back to source build

    Verified signature
    a6debd5
  15. Remove llama-cpp-python from requirements (installed at runtime)

    Verified signature
    6e19d95
  16. Fix: install llama-cpp-python at runtime instead of build time

    Verified signature
    779eb8a
  17. Add llama-cpp-python for CPU inference

    Verified signature
    e135e67
  18. Add CPU LLM benchmark endpoint (Gemma E4B, Qwen3 8B, Qwen3.8 27B)

    Verified signature
    8e414b7
  19. Update API reference with news/images backends and extraction support

    Verified signature
    c638c2d
  20. Improve search: add news/images backends, page extraction, region support

    Verified signature
    3cbe392
  21. Add full API reference for chat agent integration

    Verified signature
    7dcbaae

  1. Remove unused searxng config

    Verified signature
    01cbdc2
  2. Switch to ddgs

    Verified signature
    10821eb
  3. Switch to ddgs package

    Verified signature
    d249282
  4. Add duckduckgo_search dep

    Verified signature
    de4169f
  5. Use duckduckgo_search package

    Verified signature
    176dfd1
  6. Remove categories param from search route

    Verified signature
    4485c4c
  7. Fix regex quoting for DuckDuckGo search

    Verified signature
    b047b4e
  8. Remove searxng deps

    Verified signature
    3c1c18c
  9. Use public SearXNG instances instead of bundling

    Verified signature
    b7ba30b
  10. Add SearXNG config

    Verified signature
    c35e741
  11. Add SearXNG deps

    Verified signature
    93b0b82
  12. Add SearXNG search resource

    Verified signature
    84090f0
  13. Read CPU/RAM from cgroup limits instead of host

    Verified signature
    f0a025b
  14. Fix specs to show host vs ZeroGPU allocation

    Verified signature
    eed5f09
  15. Disable watcher thread + explicit launch

    Verified signature
    b187817
  16. Fix import stuck in comment line

    Verified signature
    4029ada
  17. Fix __func__ error on HF runner

    Verified signature
    6b8100d
  18. Fix staticmethod patching via __func__

    Verified signature
    6d16737
  19. Patch App.create_app to inject routes into served app

    Verified signature
    618d7e8
  20. Add back @spaces.GPU touch required by ZeroGPU runtime

    Verified signature
    b456dcc
  21. Import spaces before torch to avoid CUDA init race

    Verified signature
    a980267
  22. Disable SSR mode to prevent Node.js crash

    Verified signature
    54c97ca
  23. Add explicit demo.launch() call

    Verified signature
    4586282
  24. Add API routes directly to Gradio's FastAPI app - no monkey-patch needed

    Verified signature
    bd1297a
  25. Clean Respite API: FastAPI app with Gradio mounted, no GPU touch

    Verified signature
    4d82070
  26. Fix README.md

    Verified signature
    b7bdd13
  27. Upload README.md with huggingface_hub

    Verified signature
    765d388
  28. Debug monkey-patch with error logging

    Verified signature
    38ffac8
  29. Use /respite/ prefix with working Gradio mount structure

    Verified signature
    d5b7b54