RandNMR73
|
9c3d7e33fd
|
clean
|
2026-04-08 06:16:57 +00:00 |
|
RandNMR73
|
061c205457
|
vsa with qat training kernel
|
2026-02-14 11:23:34 +00:00 |
|
RandNMR73
|
68002a0ae7
|
update default num heads
|
2026-01-29 01:07:02 +00:00 |
|
RandNMR73
|
c1a6d7f21f
|
update benchmark scripts
|
2026-01-29 00:56:58 +00:00 |
|
RandNMR73
|
011b9e707d
|
add combined benchmarks
|
2026-01-29 00:48:24 +00:00 |
|
RandNMR73
|
af03817af0
|
add fa2 benchmarking
|
2026-01-29 00:38:09 +00:00 |
|
RandNMR73
|
a8661636d0
|
fp4 1.3B inference
|
2026-01-29 00:19:08 +00:00 |
|
RandNMR73
|
96609c05d0
|
sage3 inference 1.3B baseline
|
2026-01-28 14:57:40 +00:00 |
|
RandNMR73
|
9361c72c31
|
remove incorrect two level qkv and revert
|
2026-01-28 11:49:22 +00:00 |
|
RandNMR73
|
3ee13575c1
|
use the product of the two sfs
|
2026-01-28 11:14:58 +00:00 |
|
RandNMR73
|
58cb2f48f8
|
test 2 level quant q, k, v
|
2026-01-28 10:40:34 +00:00 |
|
RandNMR73
|
098ee30e37
|
smooth k ablation:
|
2026-01-28 10:14:18 +00:00 |
|
RandNMR73
|
0543dbd689
|
two level P ablation
|
2026-01-28 08:38:15 +00:00 |
|
RandNMR73
|
3aa5b1fc0e
|
update
|
2026-01-28 02:36:39 +00:00 |
|
RandNMR73
|
fc104818de
|
modified sage3 with cfg=3 inference
|
2026-01-28 02:17:56 +00:00 |
|
RandNMR73
|
33a2d512b7
|
update sage3 to use per_block_mean=True
|
2026-01-27 13:07:40 +00:00 |
|
RandNMR73
|
643c8e9cd1
|
fix 14B validation videos
|
2026-01-27 12:57:08 +00:00 |
|
RandNMR73
|
6f532a183b
|
update benchmarking script attention flop counts
|
2026-01-27 04:45:30 +00:00 |
|
RandNMR73
|
75d6cf1a31
|
update benchmark scripts
|
2026-01-27 04:38:11 +00:00 |
|
RandNMR73
|
72b61ab1fe
|
add sage3 fwd + bf16 bwd script
|
2026-01-27 04:08:56 +00:00 |
|
RandNMR73
|
0c128c8536
|
ignore fastvideo_kernel
|
2026-01-27 04:07:19 +00:00 |
|
RandNMR73
|
dd2096abab
|
fix sage3 api bug
|
2026-01-27 04:02:37 +00:00 |
|
RandNMR73
|
e57ffd2b3f
|
update batch inference script
|
2026-01-27 02:44:14 +00:00 |
|
RandNMR73
|
1f39e5c790
|
ignore fastvideo_kernel
|
2026-01-27 02:19:16 +00:00 |
|
RandNMR73
|
afd6769a2f
|
update pyrpoject.toml again
|
2026-01-27 00:54:15 +00:00 |
|
RandNMR73
|
2e7b3f787d
|
update pyproject.toml
|
2026-01-27 00:51:30 +00:00 |
|
RandNMR73
|
d127e513e9
|
updates
|
2026-01-27 00:07:01 +00:00 |
|
Matthew Noto
|
c8d3a059db
|
update .gitmodules
|
2026-01-26 09:03:24 +00:00 |
|
RandNMR73
|
c00fac99b9
|
fix random seed in training pipeline bug
|
2026-01-25 22:10:16 +00:00 |
|
RandNMR73
|
aa49fb9dd8
|
add a bunch of scripts + modify sage3 to support turning off two level quant P
|
2026-01-25 04:24:34 +00:00 |
|
RandNMR73
|
a7098ad378
|
rebase
|
2026-01-22 13:02:32 +00:00 |
|
RandNMR73
|
900cc8d707
|
checkpoint (qat attn in progress)
|
2026-01-22 12:32:14 +00:00 |
|
Matthew Noto
|
ce95729140
|
fix DeepGEMM path
|
2026-01-22 12:15:29 +00:00 |
|
Matthew Noto
|
5121718004
|
add inference repo
|
2026-01-22 12:15:26 +00:00 |
|
XOR-op
|
aec9c4d313
|
fix: SP for hunyuanvideo 1.5 (#1026)
|
2026-01-22 12:09:56 +00:00 |
|
 Shao DuanandWill Lin
|
51bef40779
|
Added LTX-2 Distilled T2V Generation (#1016)
Co-authored-by: Will Lin <wlsaidhi@gmail.com>
|
2026-01-22 12:09:56 +00:00 |
|
 alexzmsandWilliam Lin
|
e2979d6d56
|
[kernel] [bugfix] [ci] bump v0.2.4. Fix STA output handling, TurboDiffusion CUDA norm dtypes for fastvideo-kernel unit tests. (#1020)
Co-authored-by: William Lin <SolitaryThinker@users.noreply.github.com>
|
2026-01-22 12:09:55 +00:00 |
|
William Lin
|
9ba9a3f2bd
|
[kernel] Fix fastvideo-kernel release workflow (#1019)
|
2026-01-22 12:09:55 +00:00 |
|
XOR-op
|
bee1a5af98
|
[feat] Hooks API and layerwise offloading for all DiTs (#1006)
|
2026-01-22 12:09:55 +00:00 |
|
William Lin
|
5ec0e0947c
|
[chore] release fastvideo-kernel 0.2.3 (#1018)
|
2026-01-22 12:09:55 +00:00 |
|
alexzms
|
0224ea9a14
|
[Bug Fix] Add autograd wrapper for block-sparse attention in fastvideo-kernel + fix CMake extension linking (#1015)
|
2026-01-22 12:09:55 +00:00 |
|
William Lin
|
f78843ca3f
|
[CI] Fix OOM issues in ssim tests (#1011)
|
2026-01-22 12:09:50 +00:00 |
|
 alexzmsandWill Lin
|
8c9940713a
|
[CI] SSIM tests optimization: load all model weights from Modal persistent Volume (#958)
Co-authored-by: Will Lin <wlsaidhi@gmail.com>
|
2026-01-22 12:09:05 +00:00 |
|
KyleShao
|
ae439b1290
|
[feat] Introduce Cosmos 2.5 Text2World pipeline (#974)
|
2026-01-22 12:09:04 +00:00 |
|
William Lin
|
0d6c9a5033
|
[misc] [bugfix] unpin 'av' in pyproject (#1009)
|
2026-01-22 12:09:04 +00:00 |
|
XOR-op
|
d8b9d7c6ab
|
[feat!] Disable FSDP inference by default (#1001)
|
2026-01-22 12:09:04 +00:00 |
|
Loay Rashid
|
989f1d5462
|
[CI] Fixed Turbodiffusion I2V CI (#1002)
|
2026-01-22 12:09:04 +00:00 |
|
![gemini-code-assist[bot]](/assets/img/avatar_default.png) William Linandgemini-code-assist[bot]
|
5753d9273f
|
[ci] temporarily disable turbodiffusion ssim test (#1000)
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
|
2026-01-22 12:09:03 +00:00 |
|
Will Lin
|
6b9626c0f1
|
Revert "dit"
This reverts commit a6a9c9ca07.
|
2026-01-22 12:09:03 +00:00 |
|
Will Lin
|
c6dc7e8181
|
dit
|
2026-01-22 12:09:03 +00:00 |
|