Fix using VAE in quantization mode #1090
Conversation
WalkthroughRecent changes in the codebase involve the removal of conditional logic concerning the OneFlow backend in Changes
Poem
Thank you for using CodeRabbit. We offer it for free to the OSS community and would appreciate your support in helping us grow. If you find it useful, would you consider giving us a shout-out on your favorite social media? TipsChatThere are 3 ways to chat with CodeRabbit:
Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (invoked as PR comments)
Additionally, you can add CodeRabbit Configuration File (
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Files selected for processing (2)
- onediff_comfy_nodes/modules/booster_cache.py (2 hunks)
- onediff_comfy_nodes/modules/oneflow/booster_quantization.py (3 hunks)
Additional context used
Ruff
onediff_comfy_nodes/modules/oneflow/booster_quantization.py
135-135: No explicit
stacklevelkeyword argument found(B028)
Additional comments not posted (2)
onediff_comfy_nodes/modules/booster_cache.py (1)
54-64: Evaluate the impact of removing OneFlow backend checks.The removal of OneFlow backend checks simplifies the code but may affect functionality related to caching and backend compatibility. Ensure that this change aligns with the overall architectural goals and does not introduce regressions.
Consider verifying the impact of these changes on the system's behavior, especially in environments where OneFlow might be used.
onediff_comfy_nodes/modules/oneflow/booster_quantization.py (1)
130-139: Ensure the newVAEhandling aligns with quantization objectives.The new method for handling
VAEin theexecutemethod is a valuable addition. Ensure that this approach aligns with the overall quantization strategy and does not introduce inconsistencies.Consider testing the behavior of this method with various
VAEmodels to ensure it functions as expected.Tools
Ruff
135-135: No explicit
stacklevelkeyword argument found(B028)
| @execute.register(VAE) | ||
| def _(self, model: VAE, **kwargs): | ||
| # TODO: VAE does not support quantization and patch compatibility | ||
| from .booster_basic import BasicOneFlowBoosterExecutor | ||
|
|
||
| warnings.warn( | ||
| "TODO: VAE does not support quantization and patch compatibility", | ||
| UserWarning, | ||
| ) | ||
| return BasicOneFlowBoosterExecutor().execute(model, **kwargs) |
There was a problem hiding this comment.
Add stacklevel to the warning for better traceability.
The warning about VAE not supporting quantization lacks a stacklevel argument, which could help identify where the warning originates.
Apply this diff to improve the warning:
warnings.warn(
"TODO: VAE does not support quantization and patch compatibility",
UserWarning,
+ stacklevel=2
)Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| @execute.register(VAE) | |
| def _(self, model: VAE, **kwargs): | |
| # TODO: VAE does not support quantization and patch compatibility | |
| from .booster_basic import BasicOneFlowBoosterExecutor | |
| warnings.warn( | |
| "TODO: VAE does not support quantization and patch compatibility", | |
| UserWarning, | |
| ) | |
| return BasicOneFlowBoosterExecutor().execute(model, **kwargs) | |
| @execute.register(VAE) | |
| def _(self, model: VAE, **kwargs): | |
| # TODO: VAE does not support quantization and patch compatibility | |
| from .booster_basic import BasicOneFlowBoosterExecutor | |
| warnings.warn( | |
| "TODO: VAE does not support quantization and patch compatibility", | |
| UserWarning, | |
| stacklevel=2 | |
| ) | |
| return BasicOneFlowBoosterExecutor().execute(model, **kwargs) |
Tools
Ruff
135-135: No explicit
stacklevelkeyword argument found(B028)
Summary by CodeRabbit
VAEmodels in the booster quantization process, with a warning about current limitations.