Commit History

2 days ago

  1. chore: root the deploy branch on the Space head Running

    e2b32f1
  2. fix: below 1024 px a control reached by Tab stops above the commit bar instead of under it: the page holds back the bar's height when it scrolls (round 4's N5, WCAG 2.4.11) (#20)⁠

    Unverified signature
    ea5b282
  3. fix: the model's own sound reaches the writer unclamped: both runners build LTX-2.5's vocoder with the pinned forward minus its final clamp to [-1, 1], so sound over full scale is turned down as a whole by the writer instead of clipped inside the model (#14)⁠

    Unverified signature
    80f9b11
  4. fix: an adapter that passes its check is staged again, and Remove, Remove all and its strength work again: the adapter list carries no change listener, so Gradio no longer hashes the staged adapters, which cannot be hashed; its own events collapse a failure card naming the adapters, when they changed the list (#20)⁠

    Unverified signature
    1d8deb6
  5. docs: every sentence stays true after the deploy: README and the release notes say this is deploy 2's code and point to issue #32 for what the Space shows, and no doc says the Space still runs deploy 1, is not deployed, or has not been redeployed since August (#32)⁠

    Unverified signature
    abc9150
  6. docs: the release notes' Deploy 2 section names its commits as integrated: UI-B's round-4 package (78036d2 ... 7cef5fb) by issue and these docs by issue, and drops the list of what was still to land (#32)

    Unverified signature
    b5a4494
  7. docs: UI_DESIGN quotes mode 11's sentence as the page shows it and as the brief gives it, "With a speech recording and no first-frame picture, some seeds come out as a black-and-white period film ...", instead of paraphrasing it (#28)

    Unverified signature
    1cda206
  8. docs: "Where things stand" says the README describes the code on its branch, which has changed since deploy 1, and that the Space still runs deploy 1, so those changes are not on it yet (#32)

    Unverified signature
    8f43118
  9. docs: the README names the size button as the page will after the round-4 review, "Snap to values that run", and its hint as naming the nearest values that run (#20)

    Unverified signature
    e7caf99
  10. docs: the release notes open a "Deploy 2 - 2026-11-17 (in preparation)" section: what changed since deploy 1, issue by issue with its commits (all 35 on fix/studio-2026-11), what lands next, what is not in it, and an "On the Space" heading for the result after the deploy (#32)

    Unverified signature
    85c4433
  11. docs: the README's local section says the progress bar names "loading the model" while the text encoder and transformer load and "decoding the frames" after the last step, with either decoder (#12)

    Unverified signature
    974496c
  12. docs: the README says scripts/local_generate.py runs the same MPS numerics self-test as the app and exits refused before loading anything on a PyTorch that fails it (#11)

    Unverified signature
    71e35bc
  13. docs: the README's watermark row says a bad pair stops an MPS start in the app's own words before the numerics self-test runs, so it is never reported as a PyTorch that computes wrongly (#10)

    Unverified signature
    ed8912d
  14. docs: the README says an API call's status Markdown now names each attached adapter and gives the run's time split into the progress bar's phases (#18)

    Unverified signature
    f9bde99
  15. docs: the model-picked length's sentence names where it applies, the modes that offer it, instead of "the other modes", two of which take the length from a clip (#16)

    Unverified signature
    58732a9
  16. docs: the README says what the note of a model-picked length reports, and that when the model's own figure lies outside the allowed range the note says so and what the range made of it (#16)

    Unverified signature
    c511137
  17. docs: the README says bitsandbytes' int8 cast warning reaches the Space's log once per process, not once per matrix product (#15)

    Unverified signature
    a0b76da
  18. docs: the README says sound the model makes over full scale is turned down to a -0.1 dBFS peak instead of clipped, in the video's track and the wav alike, with the diagnostics saying by how much, and that a user's recording is never changed (#14)

    Unverified signature
    3498ce9
  19. docs: the README describes the page after the round-4 review where a reader needs it (the line beside Generate names what returns, the monitor keeps a run and offers another mode's sample and the way back, the player answers keyboard and touch, the Done card's length words, the sound download, an adapter's next step, Restore's note), and UI_DESIGN adds the players' access (#20)

    Unverified signature
    4d60faf
  20. docs: mode 11's black-and-white sentence names the cause the evidence shows, the seed, not the recording: "With a speech recording and no first-frame picture, some seeds come out as a black-and-white period film", with the same two remedies as the page (#28)

    Unverified signature
    8d0b074
  21. docs: the diffusion decoder is built once at start-up, beside the pipeline, and packed with it on the Space, with each run building its own only if that build failed: the README's decoder paragraph and two PLAN passages say so (#17)

    Unverified signature
    d8bbbdb
  22. docs: the handoff's and ZEROGPU_DISK's status boxes no longer say the Space has run nothing since August: it runs deploy 1, whose Space record is SPACE_MODE_MATRIX §10.1, and the handoff points at the release notes (#29)

    Unverified signature
    70ddbf2
  23. docs: the weights come from the image: the README, PLAN and the handoff say preload_from_hub fetches the 35 model files (71.92 GB, no temporal upsampler) and the prompt enhancer's 8 (2.49 GB) before the app starts, and start-up downloads only what is missing, if the disk can hold it (#23)⁠

    Unverified signature
    38e36e9
  24. docs: the reservation follows size and length: the README's table is the estimator's (the default run 60 s, needing 120 s of quota), the measured envelope is a stated limit, and SPACE_MODE_MATRIX §10.3 records the six size-and-length runs and the answer to the operator's question (#21)⁠

    Unverified signature
    ca1f2a8
  25. docs: wording in deploy 1's record: the README's adapter-on-the-Space sentence says Validate once, the matrix's 10.1 intro reads as one paragraph, and the release notes' browser runs open with a sentence (#25)

    Unverified signature
    c8c5965
  26. docs: the README's second browser session names the memory check's figure against its 90 GB limit, so the 93.0 GB reads as the check's sum and not as the other jobs' load (#29)

    Unverified signature
    0eb8e93
  27. docs: the decoder choices are named by what a user sees: Auto, "Steadier (diffusion decoder)" and "Sharper (standard decoder)" in the README and UI_DESIGN, with what the Space's A/B showed and what the diffusion decode costs (#27)⁠

    Unverified signature
    360edcc
  28. docs: the README says audio to video without a first-frame picture can turn an old-sounding recording into a black-and-white period film, and how to get colour: "in colour" in the prompt, or a first-frame picture (#28)⁠

    Unverified signature
    d527b6d
  29. docs: the README's known issues name Gradio 6.24.0's un-awaited get_current_user warning on page loads, say it is Gradio's own and fixed in 6.25.0, and why the Studio stays on 6.24.0 for now (#24)⁠

    Unverified signature
    7dbfae6
  30. docs: the small truths: the Space is public and runs deploy 1, the README says what the page shows about cost on a Mac in place of the comment that asked for it, the browser session on the fixed build is written, and a speaking face gets the page's own words (#29)⁠

    Unverified signature
    da5f47d
  31. docs: deploy 1's Space run is in the repo: SPACE_MODE_MATRIX §10.1 call by call, PR #1's corrected texts as the release notes' Deploy 1 section, and the README's Space sentences carry what the Space measured (#25)⁠

    Unverified signature
    0646770
  32. docs: UI_DESIGN gives the figure line and the Upscale line as the reservation now sizes them: the default run's figure reads "704 × 480, 5.0 s, needs 120 s quota", and the Upscale line names its seconds at the default size and length and points to the cost panel (#21)⁠

    Unverified signature
    7cef5fb
  33. fix: mode 11's first-frame line names the cause the round-4 verdict found: with a speech recording and no first-frame picture, some seeds come out as a black-and-white period film; add "in colour" to the prompt, or give a first-frame picture (#28)⁠

    Unverified signature
    9887158
  34. test: the access test pins the sample jump's desktop rules as they stand after M4: hidden beside the monitor at 1024 px and wider, shown there with mr-over-run while the monitor holds a run of another mode, and with no run the jump carries no mr-over-run (#20)

    Unverified signature
    e4fc218
  35. docs: UI_DESIGN's failure-card paragraph, which this branch extended, is rewrapped at the file's width (#20)

    Unverified signature
    67478d3
  36. docs: UI_DESIGN's lines this branch added wrap at the file's width (#20)

    Unverified signature
    e48ebc3
  37. style: this branch's new code and tests are laid out as ruff format lays out the files they sit in; no behaviour changes (#20)⁠

    Unverified signature
    f16d35d
  38. fix: Restore settings says what it did, the seed included: a quiet line in Recent runs reads "Settings from 08:06 restored; the seed is random again, as it was for that run. Your current audio stays as it is." until the next run (#20)⁠

    Unverified signature
    6150e0e
  39. docs: UI_DESIGN says a LoRA rejection gives one next step for its kind and "Nothing was spent." only on ZeroGPU, and gives the idle monitor's line on each platform; it reflows the strength sentence minor 1 added (#20)

    Unverified signature
    e6f95d8
  40. fix: a LoRA rejection gives one next step for its kind ("Use an adapter made for LTX-2.5.", "Download the file again, then upload it.", ...), and off ZeroGPU neither the card nor the idle monitor speaks of spending (#20)⁠

    Unverified signature
    1583c93
  41. fix: a failure card no longer outlives Show, a Restore or its own remedy within one mode: each turns it into its one line, "Last run: <mode>, <time>. <headline>." (#20)⁠

    Unverified signature
    7896071
  42. fix: a cost lamp's state is not in colour alone: an unlit lamp is hidden from assistive tech and the lamps' row is a named group, so what is read is the costs that apply (#20)⁠

    Unverified signature
    44edefb
  43. fix: the run summary's time for a run under half a second reads "under 1 s" (on ZeroGPU "under 1 s of GPU time, under 2 s of your quota"), never "Took 0 s" (#20)⁠

    Unverified signature
    74d0925
  44. fix: each sample's caption says it was made earlier by this app from its example, with that example's inputs and seed, and no longer promises the same result from the same prompt and seed (#20)⁠

    Unverified signature
    b46dfc2
  45. fix: the sample captions name the example each was made from, state the samples' real lengths ("4.7 s, the longest that fits the 5.0 s recording") and say "no upscale" in place of "single pass" (#20)⁠

    Unverified signature
    9ab0812
  46. fix: the Done card says "chosen by the model" only for a length strictly inside the range the user allowed, and at either end which end it is; "How quota works" gives the timeout card's step; no "legal" on the page (#20)⁠

    Unverified signature
    a8bf7a4
  47. fix: "Use this prompt" turns Enhance off in the same step and says so ("Copied to your prompt; Enhance is off"), so the next run sends the enhanced prompt as it is; in a mode that takes no prompt it parks both instead of writing a hidden value (#20)⁠

    Unverified signature
    8183336
  48. fix: off CUDA a parked Upscale promises no size: while this machine cannot run it for the request, its label is the bare "Upscale ×2" and the cost panel shows no Upscale lamp; at a size it can run, both come back (#20)⁠

    Unverified signature
    44c216d
  49. fix: arrow keys on a LoRA strength slider reach the run: the slider's input is wired as well as its release, last value wins, and the value it sends is held to the slider's own range (#20)⁠

    Unverified signature
    d468ab6
  50. fix: the line beside Generate names what the run returns: the returned size, with "upscaled" when Upscale ×2 is on ("1408 × 960 upscaled, 5.0 s, needs … quota"), and "Sound" in a sound-only mode ("Sound, 5.0 s, …"), keeping the length and the quota or device words (#20)⁠

    Unverified signature
    05737f6