-
4a56274d10
[feat] v2: Wan2.1 i2v port (Wan2.1-Fun-1.3B-InP) — CLIP encoder + first-frame conditioning, GPU-verified
Will Lin
2026-06-18 11:52:00 +00:00
-
9ede9af123
[feat] v2 Wan loop+adapter: optional i2v conditioning hooks (cond concat + CLIP context); T2V unchanged
Will Lin
2026-06-18 11:42:35 +00:00
-
3d405f6dfa
[feat] v2 torch backend: CLIP image-encoder adapter + image_encoder component kind (i2v groundwork)
Will Lin
2026-06-18 11:35:18 +00:00
-
d891771ba3
[revert] v2: drop FastWan/VSA registry entries — generic Wan loader can't map their gated-attn params
Will Lin
2026-06-18 11:31:31 +00:00
-
67dea39052
[feat] v2: alias FastVideo/Wan2.1-VSA-T2V-14B-720P to the Wan-14B card (bucket B)
Will Lin
2026-06-18 11:19:17 +00:00
-
5f1d2ef7d2
[feat] v2: port Wan2.1-T2V-14B (bucket B) + document the all-models backlog
Will Lin
2026-06-18 11:10:05 +00:00
-
440b99523e
[refactor] v2 torch backend: TorchComponent base + one generic builder + v2.* facade (Phase 1b/1c)
Will Lin
2026-06-18 11:01:45 +00:00
-
1cb2b4e84c
[feat] v2: per-model sampling defaults on ModelCard (Phase 1a)
Will Lin
2026-06-18 10:47:28 +00:00
-
51898f48e9
[refactor] v2: rename models/ (recipe layer) -> recipes/; move toy backend -> platform/backends/toy.py
Will Lin
2026-06-18 10:35:41 +00:00
-
4e331eb7ff
[refactor] v2: use absolute imports (v2.*) everywhere instead of relative
SolitaryThinker
2026-06-18 09:28:52 +00:00
-
c6d2976fc2
[feat] v2 VideoGenerator: A/V convenience path (generate_video -> T2VS -> mp4 + 24kHz wav)
SolitaryThinker
2026-06-18 08:59:43 +00:00
-
6096b00aeb
[feat] v2 LTX-2.3 T2VS GPU-verified: audio VAE/vocoder wiring + dual-connector audio fix
SolitaryThinker
2026-06-18 08:44:07 +00:00
-
1d23399d81
[feat] v2 LTX-2.3 T2VS: single-stage joint audio+video card/loop/program (CPU-verified)
SolitaryThinker
2026-06-18 08:36:59 +00:00
-
fefcd415ff
[feat] v2 LTX-2 adapters: A/V foundation (joint DiT forward + dual text connector + audio decode)
SolitaryThinker
2026-06-18 08:23:01 +00:00
-
7ac2ff0d1c
[refactor] v2: shared model registry (HF-id primary + arch fallback) for all entrypoints
SolitaryThinker
2026-06-18 07:57:55 +00:00
-
fa9c58b419
[fix] v2: name LTX-2 cards by architecture (2-stage vs single-stage) + Wan2.1 is T2V-only
SolitaryThinker
2026-06-18 07:46:09 +00:00
-
0662b42510
[docs] v2: LTX-2 base/2.3 GPU-verified on rebuilt x86 stack + remaining-port mechanisms
SolitaryThinker
2026-06-18 07:13:28 +00:00
-
ae6d8085de
[docs] v2: LTX-2.3 example + roadmap (A14B offload working, base/2.3 ported, env status)
SolitaryThinker
2026-06-18 06:27:16 +00:00
-
d0648ba8d9
[feat] v2: Wan2.2-A14B MoE CPU offload (fits 1 GPU) + LTX-2 base/2.3 single-stage port
SolitaryThinker
2026-06-18 06:24:15 +00:00
-
e10828346f
[feat] v2: architecture-driven dispatch + real LTX-2 upsampler + Wan2.2-A14B MoE card
SolitaryThinker
2026-06-18 05:41:42 +00:00
-
655f362cf4
[feat] v2 port: Wan2.2-TI2V-5B (T2V) — 4th GPU-verified model
SolitaryThinker
2026-06-18 04:55:29 +00:00
-
3541e81d66
[feat] v2 VideoGenerator: convenience API (from_pretrained/generate_video) + porting roadmap
SolitaryThinker
2026-06-18 04:33:56 +00:00
-
f51497ee6d
[feat] v2 VideoGenerator: typed fastvideo.api entrypoint over the v2 engine
SolitaryThinker
2026-06-18 03:56:17 +00:00
-
d8af2e60d2
[feat] v2 ltx2: two-stage distilled GPU bring-up (LTX2Transformer3DModel 18.88B)
SolitaryThinker
2026-06-18 02:51:39 +00:00
-
f79919ba8a
[feat] v2 wan_causal: causal DiT (CausalWanTransformer3DModel) GPU bring-up
SolitaryThinker
2026-06-18 02:34:41 +00:00
-
1594f8e6be
[fix] v2 tests: make no-torch-import guards GPU-aware (skipif torch installed)
SolitaryThinker
2026-06-17 23:30:17 +00:00
-
64cadaa0bf
[feat] v2 wan21: backend-aware latent geometry + checkpoint stamping
SolitaryThinker
2026-06-17 23:30:17 +00:00
-
750fc1245b
[feat] v2 cuda backend: real fastvideo construction + risk A-E fixes (Wan2.1 verified on H100)
SolitaryThinker
2026-06-17 23:30:17 +00:00
-
7725998b0c
[docs] add v2/HANDOFF.md for GPU-side bring-up of the torch backend
SolitaryThinker
2026-06-17 14:24:12 -07:00
-
4b61dedc43
[fix] correct GPU adapters against real fastvideo API (cross-check findings)
SolitaryThinker
2026-06-17 14:16:23 -07:00
-
1e58d90d02
[feat] real torch/CUDA backend (written-not-run) behind the cuda cells
SolitaryThinker
2026-06-17 13:52:28 -07:00
-
47f7a04e09
[feat] static-buffer capture form for the cudagraph step body (Path A)
SolitaryThinker
2026-06-17 13:28:56 -07:00
-
9b8838834c
[feat] piecewise CUDA-graph capture/replay at the step boundary (Path A)
SolitaryThinker
2026-06-17 13:14:40 -07:00
-
634f0828a2
[feat] route all diffusion loops + RL recompute through the kernel table
SolitaryThinker
2026-06-17 12:55:52 -07:00
-
2f044c02dd
[feat] multi-backend dispatch substrate (device/arch/kernel registries)
SolitaryThinker
2026-06-17 12:26:11 -07:00
-
431f4daddb
[feat] Adapter plane, non-linear workflows, RL→distill flywheel
SolitaryThinker
2026-06-16 23:51:32 -07:00
-
6220d02746
[feat] #8b speculative (draft-verify) decoding — exact + lower-latency AR
SolitaryThinker
2026-06-16 23:32:01 -07:00
-
9489c6c1dd
[feat] Five more stress tests: LTX-2 A/V, weight-sync, served reward, cache-dit, nested workflows
SolitaryThinker
2026-06-16 23:20:23 -07:00
-
69c9871154
[examples] Add v2_examples/{training,omni,workflows}/ — runnable examples
SolitaryThinker
2026-06-16 22:13:43 -07:00
-
18dd295e8d
[examples] Add v2_examples/inference/ — runnable Wan2.1 inference examples
SolitaryThinker
2026-06-16 22:04:56 -07:00
-
4c333e0509
[docs] Update v2/README to designv4 + current scope (127 tests)
SolitaryThinker
2026-06-16 21:24:25 -07:00
-
32a7a6b87b
[feat] Three stress tests: interactive sessions, workflow RL, heterogeneous co-scheduling
SolitaryThinker
2026-06-16 19:44:40 -07:00
-
58223c0c41
[feat] Register cross-model workflows as first-class named servables
SolitaryThinker
2026-06-16 19:13:36 -07:00
-
b254d1affe
[feat] Cross-model T2I→I2V workflow + N-way joint RL over arbitrary experts
SolitaryThinker
2026-06-16 18:38:10 -07:00
-
c7e0a8e894
[feat] Add Qwen-Omni thinker→talker→vocoder model (3 experts, 3 loops)
SolitaryThinker
2026-06-16 15:00:12 -07:00
-
b3ddf6014d
[feat] UniRL/PromptRL joint LM+generator RL stress test + designv4
SolitaryThinker
2026-06-16 13:37:18 -07:00
-
a9e5f6ee7a
[refactor] rename package mini_fastvideo → v2
SolitaryThinker
2026-06-16 10:57:01 -07:00
-
7467076d72
[fix] mini-fastvideo serving: address adversarial-review findings (capacity/credit leaks, robustness)
SolitaryThinker
2026-06-16 09:53:34 -07:00
-
01dc0c3377
[feat] mini-fastvideo serving + fleet (our own version, Dynamo-optional)
SolitaryThinker
2026-06-16 09:25:12 -07:00
-
f1dc587c74
[feat] mini-fastvideo phase 2: omni/MoT — Cosmos3 + canonical vllm-omni (BAGEL/lance)
SolitaryThinker
2026-06-16 04:40:10 -07:00
-
098bcf014a
[fix] mini-fastvideo: address adversarial-review findings (admission liveness + §7.1 cache key)
SolitaryThinker
2026-06-16 04:33:53 -07:00
-
270fae959d
[feat] mini-fastvideo: model-native runtime per design_v3 (Wan2.1/LTX2 + 4 training methods)
SolitaryThinker
2026-06-16 04:21:10 -07:00
-
4a14c1afa3
update
SolitaryThinker
2026-06-15 16:09:35 -07:00
-
f1c19050c3
design
SolitaryThinker
2026-06-13 15:39:21 -07:00
-
-
-
aea5cadb5f
[feat] WanTrack val: MotionStream-aligned joint text+motion CFG
shaoxiongduan
2026-07-04 03:22:52 +00:00
-
a199a76f4b
[feat] Add NVIDIA Veo training-data generator
SolitaryThinker
2026-07-03 15:18:11 -07:00
-
-
-
51ed1ea423
[build] Bump fastvideo-kernel pin to 0.3.2 (#1541)
William Lin
2026-07-03 14:48:15 -07:00
-
00ec3e7388
[bugfix]: compute VSA topk from padded blocks (#1517)
Mac Lee
2026-07-04 05:16:51 +08:00
-
0d626ef2d1
[kernel] Bump fastvideo-kernel pin to 0.3.1 and version the FA4 tile_mn port as 0.3.2 (#1539)
William Lin
2026-07-03 14:11:08 -07:00
-
b36d0ef085
[infra] Add DGX Spark and multi-architecture CUDA support (#1447)
William Lin
2026-07-03 13:41:43 -07:00
-
c5444621f9
[feat] WanTrack val: paired no-track counterfactual generation
shaoxiongduan
2026-07-03 08:54:31 +00:00
-
c6d0233eab
[bugfix] WanTrack: track subsampling was a no-op when max_points == N
shaoxiongduan
2026-07-03 08:18:44 +00:00
-
de8e9e20b4
[feat] WanTrack: EPIC-200 / static-EPIC-200 / Wan-200-sampler training configs
shaoxiongduan
2026-07-03 07:08:33 +00:00
-
4fe7638101
[feat] WanTrack: track-informativeness, sampler-viz, static-window dataset tooling
shaoxiongduan
2026-07-03 07:08:19 +00:00
-
6c23b4552a
[feat] WanTrack: low-rank informativeness track sampler + weights plumbing
shaoxiongduan
2026-07-03 07:08:03 +00:00
-
10c8c5df4d
[misc] reorg: relocate ui/ and performance_dashboard/ under apps/ (#1537)
William Lin
2026-07-02 20:29:17 -07:00
-
7c6f8f3f23
[feat] WanTrack: SAM object segmentation for track coverage + viz
shaoxiongduan
2026-07-02 07:14:23 +00:00
-
21f25791e2
[feat] WanTrack: MotionStream-aligned track sampling + motion-CFG + Wan-200 cfg
shaoxiongduan
2026-07-02 07:05:14 +00:00
-
d203b5f557
[feat] WanTrack: dataset-inspection dashboard + interactive draw-trace upgrades
shaoxiongduan
2026-07-01 06:31:42 +00:00
-
b588be05d1
[feat] WanTrack: DROID track-conditioning dataset + overfit config
shaoxiongduan
2026-06-30 10:12:58 +00:00
-
42d4f524a0
[feat] WanTrack: controllability eval + diagnostics tooling
shaoxiongduan
2026-06-30 10:12:33 +00:00
-
7349e07cec
[feat] WanTrack: Fun-Control init to break the track-encoder deadlock
shaoxiongduan
2026-06-30 10:06:20 +00:00
-
8af0c43a65
[fix] WanTrack: enforce I2V via Fun-InP base + CLIP cross-attention
shaoxiongduan
2026-06-27 17:06:59 +00:00
-
34a0a81be1
[feat] WanTrack: bidirectional I2V + point-track (MotionStream) finetune + eval
shaoxiongduan
2026-06-27 16:40:44 +00:00
-
7b0c070ae3
feat: port job queue/card/create slice to Next.js
will/studio-l1-jobs
SolitaryThinker
2026-06-30 16:26:31 -07:00
-
b1901894ab
[studio] add CreateJobModal stub for parallel feature units
SolitaryThinker
2026-06-30 16:09:27 -07:00
-
0772b96572
[studio] Next.js foundation: scaffold, Tailwind v4, shadcn, stores, shell
SolitaryThinker
2026-06-30 16:04:27 -07:00
-
6c04372e17
reorg
SolitaryThinker
2026-06-30 15:06:40 -07:00
-
-
-
31e26abec4
[chore] release fastvideo-kernel 0.3.1 (#1520)
William Lin
2026-06-30 12:25:27 -07:00
-
9e83ba630c
[kernel] Extract VSA utility functions into fastvideo_kernel (#1408)
sumyyyyyandSolitaryThinker
2026-07-01 03:19:25 +08:00
-
fc02a9ce8e
[ci] aarch64 kernel wheel: build for Blackwell (sm_100a/sm_120a), not Hopper (#1516)
William Lin
2026-06-29 16:10:42 -07:00
-
3d36160fc4
[bugfix] -fsigned-char so ThunderKittens compiles on aarch64 (Grace Hopper) (#1515)
William Lin
2026-06-29 14:05:38 -07:00
-
a11ec43de2
[ci] build + publish aarch64 (Grace Hopper) kernel wheels alongside x86_64 (#1514)
William Lin
2026-06-29 12:28:34 -07:00
-
2656d6530c
[bugfix] compile FP4 (attn_qat_infer) kernels for sm_120a only via per-arch split (#1508)
William Lin
2026-06-29 11:07:27 -07:00
-
e3f54e7169
[bugfix] Fix causal attention mask for Blackwell FP4 MMA column layout (#1506)
Kevin Lin
2026-06-28 20:36:40 -07:00
-
4ba5681307
[bugfix] guard ThunderKittens Hopper kernels for non-sm_90a device passes (#1507)
William Lin
2026-06-28 20:07:43 -07:00
-
8c23c86994
[docs]: add raw-video preprocess script for the QAD MixKit recipe (#1487)
alexzms
2026-06-28 18:01:13 -07:00
-
93a375c654
[bugfix]: resolve InterleaveThinker review findings
fix/interleave-thinker-review
SolitaryThinker
2026-06-28 02:32:34 -07:00
-
78d606a3a7
[feat]: make VSA tile cache configurable for training (#1444)
William Lin
2026-06-28 02:18:31 -07:00
-
c4e108de78
[docs]: modernize documentation build (#1503)
William Lin
2026-06-28 02:02:09 -07:00
-
16bf2eaf77
[feat] Relativistic RoPE re-indexing for long causal rollouts (#1454)
Kaiqin Kong
2026-06-28 01:43:00 -07:00
-
1d15d974fa
[misc]: cleanup outdated or unneeded agent infrastructure (#1504)
William Lin
2026-06-27 19:37:49 -07:00
-
5e57868b76
[ci] train-framework model coverage: Cosmos/MatrixGame2 finetune grad-norm + Cosmos/LongCat/MatrixGame2 loading smokes (#1497)
alexzms
2026-06-27 17:16:00 -07:00
-
6be280c914
[bugfix]: fix docs build (#1502)
William Lin
2026-06-27 13:00:09 -07:00
-
3ccdec9798
[bugfix] Make enable_torch_compile_vae actually compile the Wan VAE (#1498)
KyleNeverGivesUpandClaude Opus 4.8
2026-06-27 12:30:59 -07:00
-
8658f774f3
[docs] Restructure contributing CI/CD and testing docs (#1501)
Satyam Srivastava
2026-06-27 11:46:18 -07:00
-
83de8caab4
[docs]: update InterleaveThinker branch name
Mac Lee
2026-06-27 11:28:40 +00:00
-
7540d1181f
[docs]: record InterleaveThinker training dry-runs
Mac Lee
2026-06-27 11:24:26 +00:00
-
9758d931d3
[docs]: record role model abstraction cleanup
Mac Lee
2026-06-27 07:18:43 +00:00
-
7219b3b743
[refactor]: add non-diffusion role model base
Mac Lee
2026-06-27 07:18:10 +00:00