Commit History

  1. Clarify upstream NEvo attribution

    mzx
    f010b2f
  2. Add image-only pipeline mode

    mzx
    81ab95d
  3. Add MPS-compatible component dtypes

    mzx
    2ac317e
  4. Condense Intended Use / Out-of-scope / Risks to a single paragraph

    Verified signature
    e66caaf
  5. License under CC BY-NC 4.0; add Intended Use / Out-of-scope / Risks section

    Verified signature
    8ea0da4

  1. Revert "fix params": restore original mutation/elite/video defaults

    Verified signature
    85ce233
  2. Sync from GitHub: save_video fps default 12 + README fps/batch docs + fix params (mutation/elite/video defaults)

    Verified signature
    f6486d6

  1. Expose encoder input resolution as score_size (default 224)

    Verified signature
    a29c99b
  2. README quickstart: also save out.best.image

    Verified signature
    da71068
  3. Double default max_evals (400/200) + README consistency + fix run_roi_samples

    Verified signature
    6b8af13
  4. Default seed to None (random each run); expose used seed in metadata

    Verified signature
    035b26d
  5. Default image + video generation to 512x512

    Verified signature
    c3881ea
  6. Progress bar: show only best-score-so-far (drop phase labels)

    Verified signature
    7f7dfce
  7. ~2.2x faster search (drop CPU PIL->tensor stall), live progress bar, README seed

    Verified signature
    715e655
  8. Per-stage eval budgets image_max_evals/video_max_evals (200/100)

    Verified signature
    38c35d7
  9. Fast default image/video kwargs + lighter default generations (10/5)

    Verified signature
    4227234
  10. Split into image_num_generations (50) / video_num_generations (25)

    Verified signature
    2035209
  11. Replace num_candidates/max_evals with num_generations (population_size + num_generations)

    Verified signature
    5525fdc
  12. Replace num_rounds with population_size (default 20), matching clean_search

    Verified signature
    23e4cfe
  13. Tune default batch sizes (video_batch_size 4->8) from LTX-13B/SDXL sweep

    Verified signature
    3e7e55c
  14. Align two-stage search with canonical: clean scoring, art_data stage-1, image/video batching

    Verified signature
    dbe598f
  15. Flag work-in-progress at top of model card

    Verified signature
    17bb85d
  16. default_score_frames = 24

    Verified signature
    708cf86
  17. Fix stage-1: replicate still image to N frames for the encoder

    Verified signature
    4448163
  18. Two-stage pipeline (image search then motion search) with per-stage budgets

    Verified signature
    05094e8
  19. Progress bar option for the genetic search

    Verified signature
    991697f
  20. Fix data path under HF symlink cache (drop __file__.resolve())

    Verified signature
    74f73c7
  21. Off-the-shelf custom_pipeline bootstrap + README

    Verified signature
    1c2e53c
  22. Quickstart on defaults + Faster run block; drop seed_prompts

    Verified signature
    cb12db7
  23. Double-column gallery, drop How-it-works, swappable placeholder models

    Verified signature
    68deec5
  24. Simplify How-it-works, drop gallery caption column, halve gallery videos

    Verified signature
    36c2be2
  25. Sync from GitHub: install instructions, updated gallery videos, deps

    Verified signature
    75c3413

  1. Quickstart one-line video save + export helper

    Verified signature
    a9a528a
  2. ROI-enhanced search space + docs; EBA caption

    Verified signature
    920eeb0
  3. Update EBA gallery clip

    Verified signature
    a6678dc
  4. Model card: EBA example, drop pct, fsaverage5 note

    Verified signature
    64044f7
  5. Add gallery assets

    Verified signature
    43c52ab
  6. Model card: gallery + website + citation

    Verified signature
    b99693e
  7. Remove stray .pytest_cache

    Verified signature
    a7015bb
  8. Publish self-contained NEVO stimulus-synthesis pipeline + model card

    Verified signature
    c5e4258
  9. initial commit

    Verified signature
    f360f62