Files

T

Junda ChenandClaude Opus 5.5 be8a13b749 [perf] FastH3 two-Spark: gather packed FP4 inputs before head projections

Opt-in h3_vsa_fp4.sp_input_gather (default false; requires sp_projection_overlap) gathers the
already-packed FP4 inputs and rotary tables, then computes each rank's assigned Q/K/V heads locally;
the gate projection overlaps the smaller transfer. Supported on the measured GB10/cuDNN shape only;
other shapes keep the existing exchange. Enabled in the Spark-pair example.
T2AV on 2 Sparks: 19.35 -> 18.76 s/clip (merge gate check, same SP reduction order rules).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

2026-10-24 02:16:48 +00:00

api

[docs]: modernize documentation build (#1503 )

2026-06-28 02:02:09 -07:00

assets

[feat]: FastH3 Omni (Ref2VA) on Apple Silicon MLX (#2000 )

2026-10-21 12:01:54 -07:00

attention

[feat]: FastH3 Omni (Ref2VA) on Apple Silicon MLX (#2000 )

2026-10-21 12:01:54 -07:00

contributing

[ci] Run single-GPU CI lanes on a DGX Spark runner; weekly full suite on main (#2007 )

2026-10-22 13:44:55 -07:00

cookbook

[feat]: FastH3 Omni (Ref2VA) on Apple Silicon MLX (#2000 )

2026-10-21 12:01:54 -07:00

design

[perf] FastH3 on DGX Spark: exact fused kernels, SP exchange overlap, Omni prefix attention

2026-10-23 22:53:50 +00:00

distillation

[docs] Migrate documentation and cookbook to Blume (#1969 )

2026-10-16 16:09:04 -07:00

getting_started

[perf] FastH3 two-Spark: gather packed FP4 inputs before head projections

2026-10-24 02:16:48 +00:00

inference

[feat]: FastH3 Omni (Ref2VA) on Apple Silicon MLX (#2000 )

2026-10-21 12:01:54 -07:00

[feat] Cookbook serving, hardware catalog, and landing page (#1999 )

2026-10-20 09:10:29 -07:00

tests

[feat] Cookbook serving, hardware catalog, and landing page (#1999 )

2026-10-20 09:10:29 -07:00

training

[feat] Add H3 PDD/DMPDD training and packed Ref2VA preprocessing (#1984 )

2026-10-20 11:15:57 -07:00

utilities

[feat] Add streaming and GPU-accelerated LoRA extraction (#1784 )

2026-10-07 04:48:35 -07:00

check_cookbook_assets.py

[docs] Migrate documentation and cookbook to Blume (#1969 )

2026-10-16 16:09:04 -07:00

contact.md

[docs] Add William Lin to the contact page (#1977 )

2026-10-17 20:52:55 -07:00

eslint.config.mjs

[feat] Add serving cookbook recipes and configuration API (#1941 )

2026-10-09 15:49:08 -07:00

generate_examples.py

[docs] Migrate documentation and cookbook to Blume (#1969 )

2026-10-16 16:09:04 -07:00

index.md

[docs] Migrate documentation and cookbook to Blume (#1969 )

2026-10-16 16:09:04 -07:00

node_modules

[feat] Cookbook serving, hardware catalog, and landing page (#1999 )

2026-10-20 09:10:29 -07:00

package-lock.json

[feat] Add serving cookbook recipes and configuration API (#1941 )

2026-10-09 15:49:08 -07:00

package.json

[feat] Add capacity metadata to existing cookbook model manifests (#1983 )

2026-10-20 08:24:19 +00:00

prettier.config.mjs

[feat] Add serving cookbook recipes and configuration API (#1941 )

2026-10-09 15:49:08 -07:00

README.md

[feat] Add capacity metadata to existing cookbook model manifests (#1983 )

2026-10-20 08:24:19 +00:00

README.md

FastVideo documentation

Canonical guides live in this directory. The Blume application in docs-site builds and publishes them at haoailab.com/FastVideo.

See Documenting FastVideo for local setup, source ownership, page structure, validation, and deployment. The existing GitHub Pages workflow publishes Blume from main with the same domain and base path. MkDocs is no longer part of the build.

Cookbook sources

Recipe metadata, shared JavaScript, and tests remain under docs/. From the repository root:

npm ci --prefix docs
npm run build:catalog --prefix docs
npm run check:cookbook --prefix docs
npm run test:cookbook --prefix docs
python -m unittest discover -s docs/tests -p 'test_*.py'

The combined browser bundle, notices, index and catalogs are generated together under assets/cookbook-config/. Do not hand-edit or commit those outputs. The site preparation builds and copies them into Blume's public assets. See the recipe authoring guide and configuration API contract before changing the schema. The capacity estimation guide covers reusable hardware/component catalogs and optional recorded inference runs in existing model manifests.