Aryan Kumar and Aryan Kumar
c845a8dee2
[feat] Per-request community LoRA in the OpenAI video API ( #2026 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
2026-11-04 19:44:47 -08:00
Aryan Kumar and Aryan Kumar
3fe9c0c70d
[bugfix]: get the full CI suite green on main (pytest 9 flag teardown, Qwen golden env) ( #2059 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
2026-11-02 23:32:40 -08:00
Aryan Kumar and Aryan Kumar
ac81198d33
[new-model] Open-H3-Regenerate-2K: 768P H3 draft to 2560x1440, served on the OpenAI video routes ( #2035 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
2026-11-01 22:45:09 -08:00
Aryan Kumar and Aryan Kumar
d954fad9ec
[perf] MiniMax H3: faster FP4 attention padding, uint8 VAE pixel path and cuDNN FP4 GEMMs ( #2046 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
2026-11-01 22:23:10 -08:00
Aryan Kumar and Aryan Kumar
cb6a9a13e1
[bugfix] FA4 JIT cache vs tuning env, MXFP8 import and scale dtype, H3 VSA gate double quantization ( #2049 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
2026-11-01 19:11:25 -08:00
Aryan Kumar
7203b4976e
[bugfix]: compile only the LoRA delta, not the wrapped base layer
...
BaseLayerWithLoRA.forward was decorated with torch.compile, so wrapping a linear also compiled its base layer.
Compiling a packed NVFP4/FP8 linear on its own changes its numerics, so a wrapped layer at strength 0 could differ
from the unwrapped layer. forward now adds no compiled region: the base layer runs exactly as it would unwrapped
(eagerly, or traced with a caller's compiled block), and only the adapter delta is compiled.
2026-10-28 14:49:15 -07:00
Aryan Kumar
6bb444eef1
[bugfix]: run the base layer while a LoRA wrapper has no adapter
2026-10-28 14:49:15 -07:00
2099dd5533
[feat]: Community MiniMax-H3 LoRA loading with quantized side branch ( #1973 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
2026-10-19 14:29:23 -07:00
3f45f2da21
[bugfix] Repair DreamVerse and failing CI gates ( #1958 )
...
Co-authored-by: sy3535 <sy3535@maxwell.poly.edu >
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
2026-10-16 17:15:38 -07:00
e722023bd6
[perf] Wan kernel fusion: Triton-fused residual+LayerNorm+modulate inference path ( #1708 )
...
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com >
2026-10-07 04:08:07 -07:00
8444c0897a
[feat] FastH3 Ref2VA PDD inference with reference-video VSA sparsity ( #1907 )
...
Co-authored-by: Davids048 <jundasu@ucsd.edu >
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com >
2026-10-02 11:28:07 -07:00
William Lin
9028953625
[misc] yapf pass under CI's interpreter (3.12) + pin hook language_version ( #1702 )
2026-08-08 22:24:17 -07:00
Mac Lee and SolitaryThinker
31aa115611
[bugfix]: preserve FSDP hooks for RMSNorm qk norms ( #1513 )
...
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
2026-07-05 04:15:49 +00:00
2e35b0c6bd
[refactor]: linear/mlp FP4 path additions for Wan-2.1 (Attn-QAT 6/12) ( #1390 )
...
Co-authored-by: Peiyuan Zhang <a1286225768@gmail.com >
Co-authored-by: Matthew Noto <notomatthew31@gmail.com >
2026-06-08 14:51:57 -07:00
