Add modular pipeline for HunyuanVideo 1.5 #13389
yiyixuxu
merged
19 commits into
huggingface:main
from
akshan-main:modular-hunyuan1.5
Add modular pipeline support for HunyuanVideo 1.5
48b4f3fd
Fix I2V latent/cond spatial dimension mismatch
f029fc57
Fix guidance_scale default to 7.5 matching ClassifierFreeGuidance
5cc2ed13
Fix tokenizer type: use Qwen2TokenizerFast to match model
db41b410
Fix system message string formatting to match standard pipeline
bfbb6b45
Rewrite HunyuanVideo 1.5 modular: use standard pipeline methods directly
aec5b385
Remove I2V exports (T2V only for now)
d9ad0efa
Fix encoder: use static methods directly instead of encode_prompt
5923b5e5
Inline all standard pipeline methods, remove runtime dependency
1691e313
Add HunyuanVideo 1.5 image-to-video modular blocks
d189ef19
Fix missing FrozenDict import in before_denoise.py
cfd5de4e
auto-generated docstrings via #auto_docstring
acada9be
Fix ruff lint and format issues
6ec688dd
use InputParam/OutputParam templates and fix ruff
93d0426e
Address LTX review feedback here like add AutoBlocks, refactor I2V la…
86fe24aa
Add workflow map, workflow tests, auto docstrings, export only AutoBl…
77cd7d68
yiyixuxu
approved these changes
on 2026-04-15
Merge branch 'main' into modular-hunyuan1.5
458c97cb
Address Claude CI review
da4f4ef7
yiyixuxu
commented
on 2026-04-16
yiyixuxu
commented
on 2026-04-16
yiyixuxu
commented
on 2026-04-16
Address claude CI review 2
2bfc708b
yiyixuxu
approved these changes
on 2026-04-16
Reviewers
Assignees
No one assigned
Labels
close-to-merge
tests
size/L
modular-pipelines
utils
Milestone
No milestone
Login to write a write a comment.
Login via GitHub