TestingRef2va / app.py

Commit History

  1. Mirror the conditioner's 16-canvas table (fast 4:3, 3:4 and 21:9 added)

    multimodalart HF Staff
    Verified signature
    2225cd2
  2. Point at the public multimodalart/qwen3vl-conditioner, called without a token

    multimodalart HF Staff
    Verified signature
    1f02c70
  3. Fast canvas stays the default; the examples carry full-resolution canvases

    multimodalart HF Staff
    Verified signature
    1aa28b5
  4. Full 16:9 as the default canvas; match the conditioner's renamed fast label

    multimodalart HF Staff
    Verified signature
    6237c6a
  5. Dynamic GPU duration from the packed sequence, cached examples, torchaudio note

    multimodalart HF Staff
    Verified signature
    f03920f
  6. torchaudio for off-rate reference soundtracks; title, fillable width, slot height

    multimodalart HF Staff
    Verified signature
    4bdd7d9
  7. Images/audio/video tab order, expandable image slots, native attention for the VAEs

    multimodalart HF Staff
    Verified signature
    79e1481
  8. Mirror the generator's title, theme and advanced-options layout

    multimodalart HF Staff
    Verified signature
    cf77190
  9. One tab per reference modality, in reading order

    multimodalart HF Staff
    Verified signature
    508462c
  10. MiniMax-H3 ref2va, the denoising half of the split deployment

    multimodalart HF Staff
    Verified signature
    9e3b8ca