Aryan Kumar
b25ec2ffbd
[misc]: merge #1989 FastH3 QAD recipe onto #2038 Pika H3 Flex
...
Resolve the FP4 attention kernel and backend conflicts with main's #1980 versions
(re-adding install_fp4_vsa_attention), keep main's FP32 FSDP grouping, and keep both
save_inference and save_tagged in the checkpointer.
2026-10-29 19:04:10 -07:00
376e445f0e
[feat]: FastH3 OmniRef inference on consumer devices (RTX 5090 / PRO 6000 / DGX Spark / 4090) ( #1980 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com >
2026-10-19 14:57:36 -07:00
Aryan Kumar
212f558aab
[feat]: add the shared FastH3 NVFP4 FP4-attention QAD recipe
...
Add the OmniRef (Ref2VA, VSA tile 128) and V2/V2.1 Consumer (T2VA,
VSA tile 64) profiles of one quantization-aware distillation recipe:
NVFP4 W4A4 q/k/v/out and FFN linears in all 50 blocks, the FP4
sparse-attention emulator, frozen VSA gates and norms, output plus
layer-reconstruction regression against a frozen bf16 teacher, and
best-by-score checkpoints with audio eligibility guards.
Training code: the PDD-QAD method, the two H3 model plugins, the
NVFP4 emulated linear, and the tile-64/128 Triton emulator with its
PyTorch numeric reference. The VSA attention impl gains a fine
attention override hook and lets tile 128 run grad-capable forwards
through it. The trainer gains opt-in tagged checkpoints and early
stop hooks for the QAD guards.
Also add two runnable configs, the training guide, and CPU/GPU tests.
2026-10-19 12:00:33 -07:00
cdfbd64b04
[perf] FP8: fuse the dynamic activation quantization into Triton kernels ( #1861 )
...
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com >
2026-10-07 03:20:04 -07:00
02a027c49f
[feat]: FastH3 on DGX Spark and Apple Silicon ( #1920 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
Co-authored-by: Aryan Kumar <aryank@Aryans-Mac-Studio.local >
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com >
2026-10-06 12:17:11 -07:00
8b51380466
[feat]: FastH3 on single RTX GPUs ( #1919 )
...
Co-authored-by: Aryan Kumar <aryan5v@users.noreply.github.com >
Co-authored-by: Aryan Kumar <aryank@Aryans-Mac-Studio.local >
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com >
2026-10-06 11:37:01 -07:00
Keith and SolitaryThinker
c4824c7764
[bugfix] FP8: gate the ROCm _scaled_mm path on CDNA4 (gfx950) instead of the capability tuple ( #1859 )
...
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
2026-09-16 19:58:55 -07:00
Kyle Hu
3a3ad3d209
[feat]: NVFP4 text encoder for MiniMax H3 ( #1837 )
2026-09-12 18:31:59 -07:00
Junda Su
4e52f47d1e
[feat] add MXFP8 support on H3 ( #1796 )
2026-09-05 17:08:58 -07:00
William Lin
af2934dd6b
[feat]: FA4-FP4 ATTN_QAT_INFER on sm_100/sm_103 + NVFP4 weight purge ( #1647 )
2026-07-27 15:46:52 -07:00
Lev Novitskiy and Claude Sonnet 5
8b23984c79
[feat] Add Kandinsky5 QAD training pipeline: data preprocessing, QAT finetune, QAT-aware DMD distillation ( #1601 )
...
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com >
2026-07-26 02:59:21 -07:00
William Lin
bf18371afe
[feat] Enable LTX-2 NVFP4 linear and attention QAT fine-tuning ( #1626 )
2026-07-26 02:19:34 -07:00
Raghav K and SolitaryThinker
0c90c8c24d
[bugfix] nvfp4: cast fp32 inputs to bf16 instead of asserting ( #1488 )
...
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
2026-07-10 21:39:17 +00:00
b2ade71467
[Bugfix] QAD 5090: Torch.compile and other optimizations (15/12) ( #1466 )
...
Co-authored-by: alexzms <3036648523@qq.com >
Co-authored-by: Kaiqin Kong <k1kong@ucsd.edu >
Co-authored-by: William Lin <SolitaryThinker@users.noreply.github.com >
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
2026-06-23 01:23:57 -07:00
Kevin Lin
82ed9fe58d
[feat] QAD 5090: FP8 linear layer inference ( #1465 )
2026-06-22 18:25:06 -07:00
dc66cd97ef
[feat] QAD 5090: FP8 QAT linear training (14/12) ( #1464 )
...
Co-authored-by: William Lin <SolitaryThinker@users.noreply.github.com >
Co-authored-by: Loay Rashid <42599591+loaydatrain@users.noreply.github.com >
Co-authored-by: Kaiqin Kong <k1kong@ucsd.edu >
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
2026-06-19 01:03:51 -07:00
6da206e196
[feat] QAD 5090: FP4 QAT linear STE for training (13/12) ( #1463 )
...
Co-authored-by: William Lin <SolitaryThinker@users.noreply.github.com >
Co-authored-by: Loay Rashid <42599591+loaydatrain@users.noreply.github.com >
Co-authored-by: Kaiqin Kong <k1kong@ucsd.edu >
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
2026-06-18 15:55:47 -07:00
8acd8e21f9
[feat]: Add NVFP4QAT quantization config (Attn-QAT 2/12) ( #1348 )
...
Co-authored-by: jzhang38 <42993249+jzhang38@users.noreply.github.com >
Co-authored-by: RandNMR73 <99706358+RandNMR73@users.noreply.github.com >
2026-05-14 16:53:48 -07:00
d45d82334f
[infra] Dreamverse 11/14: Add NVFP4 quantization support ( #1334 )
...
Co-authored-by: SolitaryThinker <wlsaidhi@gmail.com >
Co-authored-by: Matthew Noto <99706358+RandNMR73@users.noreply.github.com >
Co-authored-by: XOR-op <17672363+XOR-op@users.noreply.github.com >
Co-authored-by: Zhang Peiyuan <42993249+jzhang38@users.noreply.github.com >
2026-05-14 16:26:39 -07:00
alexzms
14261bc98c
[feat] pre-commit support 120 col num ( #1167 )
2026-03-11 20:19:30 -07:00
XOR-op
dd91542cd1
[feat] Support text encoder weight override and quantization ( #983 )
2026-01-03 15:33:33 -06:00
XOR-op
60ac6537df
[feat] Support absmax style quantization for FP8 ( #981 )
2026-01-02 16:00:18 -06:00
William Lin
f572319bd9
[Feature] Remove V1 folder ( #642 )
2025-07-24 22:43:12 -07:00
Aryan Kumar