LTX2.5-Studio / scripts

Commit History

2 days ago

  1. fix: the decoder choices are named by what they do: "Auto", "Steadier (diffusion decoder)" and "Sharper (standard decoder)", in the choice, its line, the Upscale lock line, the Done card and the backend's refusals; the values sent stay auto, diffusion and conv (#27)⁠

    Unverified signature
    1c8a58d
  2. fix: the GPU reservation follows each run's size and length, from the Space's measurements with a 1.3 margin, so a default run needs 120 s of quota to start instead of 360, and on the Space a run past the measured envelope is refused before any GPU time (#21)⁠

    Unverified signature
    ac68de5
  3. fix: no entry can sample on an MPS PyTorch that fails the numerics self-test: build_local_runner applies the app's gate on mps, so scripts/local_generate.py --device mps exits refused (2) before loading anything, and the remedy names the README's floor, PyTorch 2.14 (#11)⁠

    Unverified signature
    444bc2b

3 days ago

  1. fix: the Space harness judges what a mode promises over the files a call returns, so a video-with-sound run passes with its mp4 carrying both tracks and the wav beside it judged as the sound on its own⁠

    Unverified signature
    a40a311
  2. fix: the local script leaves the change level to the mode and writes the run record into its JSON, and the record names the device that ran⁠

    Unverified signature
    419e0a5
  3. fix: the Space harness measures an audio track's onsets and band levels, so noise at a normal level no longer reads as sound⁠

    Unverified signature
    fee4645
  4. fix: the Space harness gates a chained input when its file exists, runs the app's gate as the Space does, and can carry the lead's matrix over several days⁠

    Unverified signature
    6ae3c5a
  5. chore: the hinode scripts go with the host: nothing in the repo runs, imports or globs them⁠

    Unverified signature
    9f61032
  6. fix: an explicit diffusion decode, or stage 2, that the decoder cannot take is refused before dispatch⁠

    Unverified signature
    18a60f4
  7. fix: the Space harness records the app's new picture check as it returns it, and sends the real decoder field by name⁠

    Unverified signature
    a63770b
  8. feat: the decoder is a per-run choice, and auto decodes with the model's diffusion decoder wherever it can run⁠

    Unverified signature
    174bb90
  9. feat: the Space harness runs a plan of calls by field name inside a hard daily GPU budget, and keeps the evidence of every call⁠

    Unverified signature
    d043b52
  10. fix: stage 2 runs for every mode that offers it, and a video too small for the decoder is refused before stage 1⁠

    Unverified signature
    d2490e1
  11. fix: refuse an MPS watermark pair that makes every empty_cache() fail, before anything loads⁠

    Unverified signature
    adad9b6
  12. fix: the local runner gives model-file pages back β€” the committed gauge no longer counts weights twice⁠

    Unverified signature
    75d0c53
  13. test: the real-pipeline tier β€” every mode through the app's own generate() on a tiny random checkpoint, on the CPU and on MPS⁠

    Unverified signature
    45cef14
  14. feat: scripts/local_generate.py β€” one local generation through the app's own generate(), with its memory story⁠

    Unverified signature
    af26f09

4 days ago

  1. test: `scripts/gate.sh` β€” the gate says which environment its verdict belongs to⁠

    Unverified signature
    c51fc56
  2. chore: drop the two unused imports that were the whole of `ruff check .`'s output⁠

    Unverified signature
    b438b1f

  1. feat: prompt enhancement ships β€” a real toggle, in the web process, at zero GPU cost⁠

    Unverified signature
    2f36d8a

  1. fix: motion is reported, never refused β€” the temporal gate demoted after it destroyed a real render⁠

    Unverified signature
    8c042d4
  2. feat: measure whether there is a picture in the frames β€” structure, not distinctness⁠

    Unverified signature
    e54477c

  1. docs: 11 of 11 on the Space, and what that still does not claim⁠

    Unverified signature
    17b589c
  2. docs: the fork-boundary fix is a pair, and half of it is a regression⁠

    Unverified signature
    f927c4e
  3. fix: the quota ceiling only ever saw its own invocation⁠

    Unverified signature
    62d612f
  4. scripts: a Space matrix runner that bounds its spend and refuses to cross it⁠

    Unverified signature
    f7e7379

  1. scripts: fail the sync in 20s with a diagnosis, not 75s with an ssh error⁠

    Unverified signature
    8c40da2
  2. scripts: an unreachable box means staleness is UNKNOWN, not fine⁠

    Unverified signature
    487b597

  1. scripts: refuse to run against a stale box, and name the delta⁠

    Unverified signature
    419f635
  2. scripts: noise_scale=1.0 as a mechanism check, labelled as not a pass⁠

    Unverified signature
    4804da1
  3. scripts: change the fixture, keep the criterion strict, stop the message misleading⁠

    Unverified signature
    1a033a0
  4. scripts: gate H17 on placement and length, not on harmonic retention⁠

    Unverified signature
    d413797
  5. scripts: A/B the transpose in one process, to prove the cause not just the effect⁠

    Unverified signature
    636f96c
  6. scripts: an opt-in device move, to answer the frequency question while blocked⁠

    Unverified signature
    2e85b86
  7. scripts: measure band energy per tone, and require all of them⁠

    Unverified signature
    62975e1
  8. scripts: read the H17 fixture through load_waveform, as the app does⁠

    Unverified signature
    dfacddd
  9. scripts: test the shape load_waveform actually returns⁠

    Unverified signature
    50e60f9
  10. scripts: batch the waveform, or the latents come back the wrong rank⁠

    Unverified signature
    81a3e24
  11. scripts: prove the mel cast is sufficient, not just necessary⁠

    Unverified signature
    482bb84
  12. scripts: H17, validate the constructed mel by reconstruction⁠

    Unverified signature
    db817a4
  13. scripts: the docstring still said "inaudible" after I retired the word⁠

    Unverified signature
    3ce9bcf
  14. docs+scripts: "very quiet" rather than "near-silent", and the H9 clip's provenance⁠

    Unverified signature
    8d73795
  15. scripts: never overwrite a run log, and record each run's inputs⁠

    Unverified signature
    1079f3c
  16. scripts: sweep the video condition strength, and report level with every run⁠

    Unverified signature
    2bdb935
  17. scripts: measure the audio tensor before encode, to localise two findings⁠

    Unverified signature
    15b76b3
  18. scripts: measure audio level, because a silent track passes every stream check⁠

    Unverified signature
    9eaae1e
  19. scripts: judge stereo difference relative to level, not absolutely⁠

    Unverified signature
    8724a1d
  20. scripts: make the artifacts judgeable, and measure the same-seed pairs⁠

    Unverified signature
    863e0e3
  21. scripts: two stage-2 runs in one process, to settle the NATTEN question⁠

    Unverified signature
    ed718e6
  22. scripts: add H14 stage 2 and the two same-seed comparison pairs⁠

    Unverified signature
    f584611