Repository navigation
Read the dtype from text_config in convert_ggml_to_int8.py - #2946
Conversation
Multimodal configs such as GLM-5.3-Flash's keep dtype only in text_config, so the converter failed with KeyError: 'torch_dtype'. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
📝 Walkthrough
Merge Risk: ⚪ Minimal · up to
|
| Check name | Status | Explanation | Resolution |
|---|---|---|---|
| Docstring Coverage |  |
Docstring coverage is 20.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 5 functions across 2 files. | Write docstrings for the functions missing them to satisfy the coverage threshold. |
-
Check name Status Explanation Description Check Check skipped - CodeRabbit’s high-level summary is enabled. Title check The title clearly and concisely describes the main change: reading the dtype from text_config in convert_ggml_to_int8.py. Linked Issues check Check skipped because no linked issues were found for this pull request. Out of Scope Changes check Check skipped because no linked issues were found for this pull request. Security Anti-Patterns The pull request changes only examples/hf_ptq/convert_ggml_to_int8.pyand its test. The added code contains notorch.load(..., weights_only=False), pickle-enabled NumPy loading, hardcoded `trust_r…Full details: Security Anti-Patterns
-
Explanation
The pull request changes only
examples/hf_ptq/convert_ggml_to_int8.pyand its test. The added code contains notorch.load(..., weights_only=False), pickle-enabled NumPy loading, hardcodedtrust_remote_code=True,eval,exec, or# nosec. It adds no dependency-file changes. No listed security anti-pattern is introduced.
-
✨ Finishing Touches: 💡 1
-
📝 Generate docstrings: 💡
-
- Commit to this branch
- Create a new PR
🧪 Generate unit tests (beta)
-
- Commit to this branch
- Create a new PR
-
- Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
Comment @coderabbitai help to get the list of available commands.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
meenchen
left a comment
meenchen
left a comment
There was a problem hiding this comment.
Bot review (gpt-6.1-sol) — DM the bot to share feedback.
Approve: the fallback preserves top-level dtype precedence and fixes text-only dtype lookup, with a regression test checking the converted scale precision.
No action needed.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #2946 +/- ##
=======================================
Coverage 79.67% 79.67%
=======================================
Files 659 659
Lines 74466 74466
=======================================
Hits 59331 59331
Misses 15135 15135
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
|
What does this PR do?
Type of change: Bug fix
convert_ggml_to_int8.py(#2903) read the scale dtype from the top-leveldtype/torch_dtypeofconfig.json. Multimodal configs such as GLM-5.3-Flash's keep it only intext_config, so converting a GGML export of GLM-5.3-Flash failed withKeyError: 'torch_dtype'. The converter now falls back totext_config.Usage
Unchanged:
Testing
test_dtype_is_read_from_text_configintests/examples/hf_ptq/test_convert_ggml_to_int8.pymoves the tiny export's dtype intotext_configand converts it. It fails on main withKeyError: 'torch_dtype'and passes with the fix. All 4 tests in the file pass locally (the new one runs in 0.2 s).Before your PR is "Ready for review"
CONTRIBUTING.md: N/AAdditional Information
Follow-up to #2903.
🤖 Generated with Claude Code
Summary by CodeRabbit