-
Notifications
You must be signed in to change notification settings - Fork 83
Pull requests: AI-Hypercomputer/maxdiffusion
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Speed up WAN 2.2 checkpoint load and compile (no output change)
#438
opened Jul 3, 2026 by
syhuang22
Collaborator
Loading…
Follow up fix custom flash block sizes fallback in Ulysses attention
pull ready
#437
opened Jul 2, 2026 by
ninatu
Collaborator
Loading…
Fixed-m softmax for pure-Ulysses WAN 2.2 attention (+ CFG batch-fold fix)
pull ready
#436
opened Jul 2, 2026 by
syhuang22
Collaborator
Loading…
Add MagCache inference acceleration for Wan2.2 (T2V + I2V)
#433
opened Jun 29, 2026 by
HadarIngonyama
Loading…
Refactor custom_splash_attention for improved clarity
pull ready
#431
opened Jun 29, 2026 by
Perseus14
Collaborator
Loading…
feat: experimental two-phase (head-chunked) Ulysses all-to-all
#428
opened Jun 24, 2026 by
csgoogle
Collaborator
Loading…
Fix get_dummy_wan_inputs for two-expert Wan2.2-I2V-A14B pipeline
pull ready
#427
opened Jun 24, 2026 by
ThomasNing
Loading…
fix(generate_wan): avoid process0-only gate for profiling run
pull ready
#426
opened Jun 22, 2026 by
edgexyz
Loading…
Perf(LTX2): Comprehensive XLA, Memory, and Transformer Code Quality Optimizations
gemini-review
#422
opened Jun 19, 2026 by
Perseus14
Collaborator
Loading…
refactor Flux transformer to use scanned blocks, dynamic checkpointing, and decoupled projections
gemini-review
pull ready
#417
opened Jun 12, 2026 by
prishajain1
Collaborator
Loading…
ProTip!
Filter pull requests by the default branch with base:main.