sql-agent / src

Commit History

  1. Fix SVG download: use single quotes inside style attr, no duplicate style

    Verified signature
    964b628
  2. Cap to top 30 categories, hide gr.File X button, single-row buttons, no duplicate title

    Verified signature
    7d5a424
  3. Result narration (Qwen-powered) + UI polish for upload area + dark-mode fixes

    Verified signature
    5ae5227
  4. Result narration (Qwen-powered) + UI polish for upload area + dark-mode fixes

    Verified signature
    7af78a0
  5. SQL self-correction: 3 retries with error context + GROUP BY rule in prompt

    Verified signature
    f5b7f67
  6. SQL self-correction: 3 retries with error context + GROUP BY rule in prompt

    Verified signature
    e6681c9
  7. SVG: standalone download (XML prolog, white bg, fixed dims) + Plotly first

    Verified signature
    0ac4ef2
  8. SVG: standalone download (XML prolog, white bg, fixed dims) + Plotly first

    Verified signature
    89360e0
  9. Better SQL prompt: column-name mapping rules + few-shot + distinct value hints

    Verified signature
    a7fa6d3
  10. Better SQL prompt: column-name mapping rules + few-shot + distinct value hints

    Verified signature
    b0a300f

  1. Pin kaleido==0.2.1 + native SVG fallback when kaleido fails

    Verified signature
    064c0d2
  2. UX: CSV/SVG download links, schema preview on load, cleaner SVG theme

    Verified signature
    a172c0c
  3. Load LoRA via PeftModel on top of standard base models (fixes r=16 vs r=8 mismatch)

    Verified signature
    1bbdff9
  4. Load LoRA via PeftModel on top of standard base models (fixes r=16 vs r=8 mismatch)

    Verified signature
    eb30a86
  5. Load LoRA via PeftModel on top of standard base models (fixes r=16 vs r=8 mismatch)

    Verified signature
    730b25d
  6. ZeroGPU best practice: load models at module level (cuda), inference only inside @spaces.GPU

    Verified signature
    420b1db
  7. ZeroGPU best practice: load models at module level (cuda), inference only inside @spaces.GPU

    Verified signature
    61aee8d
  8. ZeroGPU best practice: load models at module level (cuda), inference only inside @spaces.GPU

    Verified signature
    a57eca6
  9. ZeroGPU best practice: load models at module level (cuda), inference only inside @spaces.GPU

    Verified signature
    c2ac226
  10. Sequential model load/unload: keep peak VRAM at ~14GB

    Verified signature
    292fdf4
  11. Sequential model load/unload: keep peak VRAM at ~14GB

    Verified signature
    fab3ba1
  12. Initial deploy: Apple/Claude design, DuckDB, 3 trained LoRAs

    Verified signature
    a067ada