Skip to content

Fix int8 loras causing lower quality requant with wrong settings. - #14650

Merged
comfyanonymous merged 2 commits into
masterfrom
comfyanonymous-patch-2
Jun 26, 2026
Merged

Fix int8 loras causing lower quality requant with wrong settings.#14650
comfyanonymous merged 2 commits into
masterfrom
comfyanonymous-patch-2

Conversation

@comfyanonymous

Copy link
Copy Markdown
Member

No description provided.

@socket-security

Copy link
Copy Markdown

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff Package Supply Chain
Security
Vulnerability Quality Maintenance License
Updatedpypi/​comfy-kitchen@​0.2.12 ⏵ 0.2.13100 +110010010080

View full report

@coderabbitai

coderabbitai Bot commented Jun 26, 2026

Copy link
Copy Markdown

Review Change Stack

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 6c826dc8-936e-4c17-9751-a86b0e9fe061

📥 Commits

Reviewing files that changed from the base of the PR and between 7cb784e and 43dc208.

📒 Files selected for processing (2)
  • comfy/ops.py
  • requirements.txt

📝 Walkthrough

Walkthrough

comfy/ops.py now requantizes existing quantized tensor instances in post_cast and MixedPrecisionOps.Linear.set_weight instead of creating new QuantizedTensor objects from float. Both paths use scale="recalculate" and seeded stochastic_rounding. requirements.txt updates comfy-kitchen from 0.2.12 to 0.2.13.

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name Status Explanation Resolution
Docstring Coverage ⚠️ Warning Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Write docstrings for the functions missing them to satisfy the coverage threshold.
Description check ❓ Inconclusive No pull request description was provided, so there is no meaningful description to evaluate. Add a brief description of the change and why the requantization logic was adjusted.
✅ Passed checks (3 passed)
Check name Status Explanation
Title check ✅ Passed The title accurately summarizes the main fix for int8 LoRA requantization quality.
Linked Issues check ✅ Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check ✅ Passed Check skipped because no linked issues were found for this pull request.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@comfyanonymous
comfyanonymous merged commit 470ac36 into master Jun 26, 2026
16 checks passed
@comfyanonymous
comfyanonymous deleted the comfyanonymous-patch-2 branch June 26, 2026 23:41
zhangp365 pushed a commit to zhangp365/ComfyUI that referenced this pull request Jun 29, 2026
…mfy-Org#14650)

* Update comfy-kitchen

* Support requantizing with same settings as orig quant.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants