Skip to content

Actions: hao-ai-lab/FastVideo

Actions

Welcome First-Time Contributors

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
community-welcome.yml
630 workflow runs
630 workflow runs

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
[kernel] sm_100a CUDA backward for VSA block-sparse attention (blk64) Welcome First-Time Contributors #630: Pull request #1819 opened by lpc0220
22s
22s
fix(cosmos): make AdaLayerNorm autocast device-agnostic Welcome First-Time Contributors #629: Pull request #1818 opened by li-lizhe
18s
18s
fix(metrics): set SceneMetric device_map for any non-CPU device Welcome First-Time Contributors #628: Pull request #1817 opened by li-lizhe
30s
30s
[Bug] Cosmos AdaLayerNorm autocast hardcodes device_type="cuda", breaking on non-CUDA accelerators Welcome First-Time Contributors #627: Issue #1816 opened by li-lizhe
20s
20s
[Bug] VBench SceneMetric loads Qwen2.5-Omni on CPU on any non-CUDA accelerator (device_map hardcodes cuda) Welcome First-Time Contributors #626: Issue #1815 opened by li-lizhe
23s
23s
[bugfix]: require CUDA 12.9 for SM103a VSA builds Welcome First-Time Contributors #625: Pull request #1814 opened by SolitaryThinker
20s
20s
[perf] avoid full-sequence materialization in VSA coarse/sparse combine Welcome First-Time Contributors #624: Pull request #1813 opened by boxwrench
39s
39s
[feat] Add native SM103a VSA support Welcome First-Time Contributors #623: Pull request #1812 opened by lishunyang12
42s
42s
[docs]: add one-Spark FastH3 cookbook runtime with a device-count row Welcome First-Time Contributors #622: Pull request #1811 opened by aryan5v
26s
26s
[docs]: add cookbook link to README Welcome First-Time Contributors #621: Pull request #1810 opened by SolitaryThinker
24s
24s
[docs] Announce local FastH3 support Welcome First-Time Contributors #620: Pull request #1809 opened by aryan5v
30s
30s
[bugfix] Make HYWorld memory retrieval deterministic Welcome First-Time Contributors #619: Pull request #1808 opened by zjn20030811
55s
55s
[bugfix] Gate the fused Ulysses all-to-all kernel on one host, and stop re-voting per call Welcome First-Time Contributors #618: Pull request #1807 opened by shaoxiongduan
27s
27s
Adam/h3 rvm posttraining Welcome First-Time Contributors #617: Pull request #1806 opened by Abecid
21s
21s
[feat]: collapse cookbook recipe pages into an accordion layout, add … Welcome First-Time Contributors #616: Pull request #1805 opened by Ishxn20
20s
20s
[docs]: add FastH3 Preview v1 news links Welcome First-Time Contributors #615: Pull request #1804 opened by SolitaryThinker
22s
22s
[feat]: run FastH3 across two DGX Sparks with Ray sequence parallel Welcome First-Time Contributors #614: Pull request #1803 opened by aryan5v
26s
26s
Adds fastvideo serve support for FastMetal (Wan on Mac/MLX) — all three sizes: 1.3B, 14B, and 5B. Welcome First-Time Contributors #613: Pull request #1802 opened by Ishxn20
18s
18s
[feat] Add fastvideo serve configs for Wan CUDA models Welcome First-Time Contributors #612: Pull request #1801 opened by Ishxn20
35s
35s
Add H3 support into Dreamverse Welcome First-Time Contributors #611: Pull request #1800 opened by Davids048
33s
33s
[feat] Add opt-in FlashInfer attention backend for Wan Welcome First-Time Contributors #610: Pull request #1799 opened by klhhhhh
19s
19s
[feat] Add an H3 server cookbook and prompt playground Welcome First-Time Contributors #609: Pull request #1798 opened by aryan5v
19s
19s
[bugfix] Write generated documentation as UTF-8 Welcome First-Time Contributors #608: Pull request #1797 opened by shahradzomorrodi
39s
39s
[feat] add MXFP8 support on H3 Welcome First-Time Contributors #607: Pull request #1796 opened by Davids048
30s
30s
[feat]: add opt-in CUDA TAEH3 preview decode for FastH3 Welcome First-Time Contributors #606: Pull request #1795 opened by aryan5v
25s
25s