Will Lin fc3332550c [feat] v2: Wan2.2-I2V-A14B (MoE i2v) — combine boundary-routed experts + i2v conditioning ...

Reuses everything: 2 WanTransformer3DModel experts + BoundaryTimestepRouting (from the A14B MoE pattern),
the CLIP image encoder + first-frame [mask|cond] conditioning + the i2v program (from the Fun-InP i2v
port), and the shared WanDenoiseLoop (i2v hooks + the boundary expert). No new adapter. CPU-verified: the
toy MoE i2v runs end-to-end (2 experts + boundary + conditioning -> finite video); resolves with i2v caps.
Structural (GPU-pending: 2x14B, like the A14B T2V). Wan family now largely covered (T2V 1.3B/14B/TI2V-5B/
A14B, causal SF, i2v 1.3B/14B/A14B). CPU mini 211/2.

2026-07-04 17:15:34 +00:00

..

dataset/vidprom

[self-forcing] [3/n] Text embed only preprocessing (#797)

2025-09-13 14:03:53 -07:00

distill

[misc] Rename MatrixGame to MatrixGame2 (#1357)

2026-05-17 22:57:03 -07:00

inference

[feat] v2: Wan2.2-I2V-A14B (MoE i2v) — combine boundary-routed experts + i2v conditioning

2026-07-04 17:15:34 +00:00

serving

[infra] Dreamverse 09/14: Add Docker and launch scripts (#1332)

2026-05-12 18:01:11 -07:00

train

[feat]: make VSA tile cache configurable for training (#1444)

2026-06-28 02:18:31 -07:00

training

[docs]: add raw-video preprocess script for the QAD MixKit recipe (#1487)

2026-06-28 18:01:13 -07:00