[Test][HunyuanImage3] Add e2e offline inference smoke tests for I2T, T2I#2986
Closed
TaffyOfficial wants to merge 2 commits into
Closed
[Test][HunyuanImage3] Add e2e offline inference smoke tests for I2T, T2I#2986TaffyOfficial wants to merge 2 commits into
TaffyOfficial wants to merge 2 commits into
Conversation
|
Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits. |
TaffyOfficial
changed the title
TaffyOfficial
force-pushed
the
feat/hunyuan-image3-e2e-tests
branch
3 times, most recently
from
6f49a9d to
48a494f
Compare
TaffyOfficial
changed the title
…nd T2I pipelines Add missing e2e smoke tests for HunyuanImage-3.0 offline inference: - test_hunyuanimage3_i2t.py: Image-to-Text (single AR stage, text output) - test_hunyuanimage3_t2i.py: Text-to-Image (DiT-only) Both tests are marked @pytest.mark.advanced_model and require >= 4 CUDA GPUs (TP4). Signed-off-by: TaffyOfficial <2324465096@qq.com> Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
TaffyOfficial
force-pushed
the
feat/hunyuan-image3-e2e-tests
branch
from
48a494f to
4819192
Compare
Contributor
Author
|
@hsliuustc0106 This is the test case that was overlooked during the last merge |
Contributor
Author
|
@hsliuustc0106 need review |
3 tasks
…re to nightly CI
- Pin first 20 chars of AR output ("The image is a solid") against the HF greedy
reference captured via scripts/bench/hf_i2t_pr2986_baseline.py. 20 chars is the
longest stable common prefix; vllm-omni vs HF is not bitwise-alignable past
that point (sampler tie-break on temp=0 + MoE numerics).
- Add nightly Buildkite step under nightly-diffusion-x2iat-group running both
test_hunyuanimage3_i2t.py and test_hunyuanimage3_t2i.py on 4x H100, gated by
the group's existing labels (NIGHTLY / nightly-test / diffusion-x2iat-test).
Signed-off-by: zuiho-kai <wu15922848573@outlook.com>
Signed-off-by: TaffyOfficial <2324465096@qq.com>
TaffyOfficial
force-pushed
the
feat/hunyuan-image3-e2e-tests
branch
from
9517ac7 to
0ad2e82
Compare
Collaborator
|
how does this relate to #3332 as I2T is already contained? |
Contributor
Author
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Add missing e2e smoke tests for all three HunyuanImage-3.0 offline inference pipelines.
Currently only
test_hunyuanimage3_text2img.py(CLIP-based accuracy test) exists — thereare no basic smoke tests that verify each pipeline topology can load and produce output.
Changes
Tests
tests/e2e/offline_inference/test_hunyuanimage3_i2t.py— Image-to-Text (single AR stage, text output)tests/e2e/offline_inference/test_hunyuanimage3_t2i.py— Text-to-Image (DiT-only)All tests are marked
@pytest.mark.advanced_modeland require ≥ 4 CUDA GPUs (TP4).Examples
examples/offline_inference/hunyuan_image3/image_to_image.pyfor IT2I pipelineimage_to_text.pyto usebuild_prompt()andstage_configs_pathprompt_utils.pyTest Plan