Spaces:
Running on Zero
Running on Zero
Commit History
3 days ago
-
fix: the Space's weights come from the image: preload_from_hub is live for exactly the 35 files the int8 Space loads (71.92 GB; the temporal upsampler, which nothing loads, is no longer fetched), every file carries its size in model_files, and start-up downloads only what the image lacks, saying how much (#23)
Unverified signatureThe Space is public now, so the gated repo can be preloaded; while it was private the preload step ran without the Space's secrets and got 401, and every cold start downloaded 72.19 GB itself (17 min 24 s on 2026-11-16, 14 min 55 s of it the download). model_files is the single source of the set and its sizes: each component lists its files with their byte sizes from the Hub's own listing at the pinned commits, size_gb is derived from them, and file_size / total_size_bytes answer per file and per set. temporal_latent_upsampler moves to the declared-but-not-fetched group: StageRunner wires only the spatial latent_upsampler and model_index.json does not list it, so it cost 0.26 GB on every start for nothing. The set goes from 37 files / 72,185,863,109 B to 35 files / 71,923,919,050 B. README front matter: preload_from_hub is live, generated by model_files.preload_from_hub_entries() (two entries, one per pinned commit: the decoder is pinned later), and its comment says what it does, why…41dcf05
4 days ago
-
fix: the diffusion decoder loads the vendor's corrected weights, pinned per component, and loads exactly
Unverified signatureThe decoder file at the app's revision 871165de is not the vendor's final decoder: Lightricks replaced it in aba1b702 ("Decoder weights match the source bitwise"); 405 of 407 tensors differ, 10.9 % relative L2, config unchanged. Every stage-2 run has been decoded with superseded weights, and the decode policy now makes this decoder the default where it can run. - model_files pins diffusion_decoder alone to aba1b702 through ComponentSpec.revision; every other component stays at 871165de (the newer model_index.json declares components the app neither fetches nor nulls). - download_plan() groups the allow-list by revision; ensure_weights fetches each file at its own revision; preload_from_hub_entries() gives one entry per revision (preload_from_hub_entry() is the main-revision one). - One resolver for a component's directory: model_files.component_dir (path arithmetic, the HF cache keeps each revision in its own snapshot), wrapped by preflight.component_dir; LTX25_MODEL_SNAPS…66424db -
chore: the wheel ships the LoRA shape manifest, and pyproject's comments stop describing a repo that no longer exists
Unverified signature`ltx25_studio/lora/*.json` is package data now. A wheel built from HEAD's pyproject contained no JSON at all, so an installed package would have lost stage 6.5's reference shapes without a word. A spec builds a wheel offline from a copy of the tree with the environment's own setuptools and checks that every JSON beside the LoRA code is in it. Another checks that the checkout path the Space reads resolves. `testpaths` is set, because tests/ does exist. The markers are described as labels, because nothing deselects them.1c46a6a
-
fix: the pipeline takes PipelineQuantizationConfig, not BitsAndBytesConfig
Unverified signatureThe Space deploy failed on a type, not on disk and not on RAM: ValueError: `quantization_config` must be an instance of `PipelineQuantizationConfig`. A bare BitsAndBytesConfig is what a COMPONENT's from_pretrained takes; DiffusionPipeline.from_pretrained rejects it. Read at the pinned SHA rather than guessed: quant_backend="bitsandbytes_8bit" is a key in both backend mappings, quantizers/auto.py:45 and :57. This is a rung past the one the last test was built for. reached -> correct argument -> ARGUMENT OF THE RIGHT TYPE FOR THE CALLEE. The kwargs test proved the config arrived and could not tell the two classes apart, because both arrive equally well. Bound B is now an ARGUMENT -- components_to_quantize=["transformer"] -- rather than an intent a source scan had to assert. That is strictly better: widening is a change to a value a test reads, not a change to what a call happens to produce. So test_only_the_transformer_is_quantised is retired, with its reason kept where it lived…b55f9af
-
fix: no advertised lever may be attached to nothing
Unverified signatureAudited all six env vars the orchestrator was told were no-push escape hatches. FOUR were inert, not one: LTX25_TRANSFORMER_PRECISION inert -> DELETED LTX25_TEXT_ENCODER_PRECISION inert -> DELETED LTX25_GPU_DURATION_SECONDS inert -> WIRED LTX25_LORA_MAX_GB inert -> WIRED LTX25_ZEROGPU_SIZE worked LTX25_OUTPUT_RETENTION_GB worked LTX25_OUTPUT_DIR worked The precision pair is worse than the eight defects before it. Those failed silently; this one spoke — expected_weights_gb applied the precision factor while the loader ignored it, so the startup report stated 53.20 GB for a load that was still 72.19 GB, wrong in the direction that misdirects the next decision. Deleted rather than half-wired: wiring needs bitsandbytes or torchao, neither pinned, neither verifiable without a GPU, both using tensor subclasses whose interaction with ZeroGPU's packer is unknown. estimate_duration read the module constant instead of the config, so …9829d39 -
fix: download weights at runtime; a private Space cannot preload
Unverified signatureThe build succeeded and the preload init step failed with 401 on a gated file. 401 is unauthenticated, not unauthorised: the preload runs as a separate platform-managed step that does not receive Space secrets, so the request arrives with no credentials however valid the token is. Verified both ways — the public 2.3 Space preloads a gated=manual repo fine, this private one 401s on gated=auto, and the same URL fetches 200 by hand. preflight.ensure_weights() fetches the 37 files inside the app, where the token exists, from the same download_allow_list() so the single-source property is unchanged. It logs every file, because a 72 GB cold start that goes quiet for forty minutes is indistinguishable from one that has stopped, and it fails naming the file, its position, the underlying error class and its URL. This is the better design even if that diagnosis is wrong, which is why it beat making the Space public: it removes the ownership bug class outright. Preloaded files belong to the bui…f355817 -
fix: short_description within the 60-char limit HF enforces
Unverified signatureHF rejected the push at YAML metadata verification — the field was 64 characters against a 60 limit. Now 46. The guard is the point. test_deploy_drift.py had 14 tests asserting the frontmatter is internally consistent — sdk_version matches the gradio pin, the preload entry matches model_files, the revision is carried — and not one asserting a constraint the platform actually enforces. All 14 passed while a field was four characters too long to accept. Same shape as everything else on this build: the cheap half thorough, the expensive half absent. Adds the limits we have SEEN enforced or that HF documents: the 60 character description, the valid SDK set, the card colour palette, and the 10-entry preload cap. Deliberately not guesses — inventing a limit would fail a valid deploy for a rule nobody imposes, which is the same error in the other direction. Mutation-tested: restoring the real 64-character value turns the new test red.ae4a9cb -
test: deploy drift, ConfigError, zerogpu boundary, conditioning index trap, audio config
Closes the gaps the lead asked about. - Deploy drift: the gradio pin must EQUAL README sdk_version and be exact, the preload_from_hub entry must equal model_files.preload_from_hub_entry() including the revision SHA (without it the preload and the app resolve different cache keys and the Space re-downloads 72 GB), and every requirement needs a floor and a ceiling. - ConfigError: a misconfiguration is not a generation failure, so it must not live in that family — the UI tells an operator and a user different things. - zerogpu: a CUDA tensor anywhere in the returned structure is refused, since the failure mode is a silent hang; size= is built from the installed spaces signature; a duration= callable must match before schedule time. - conditioning: LTX2VideoCondition.index is a LATENT index. Pixel 9 is latent 2, and an out-of-range index raises instead of being silently skipped. - audio: every mel parameter comes from audio_vae.config, and audio.py may not contain a samp…2655c68