fix qwen3-30-a3b lcb-code score#4142
Merged
lvhan028 merged 6 commits intoInternLM:mainfrom Nov 26, 2025
Merged
Conversation
yao-fengchen
force-pushed
the
fix_accuracy
branch
from
17c386e to
9ece6a8
Compare
Contributor
There was a problem hiding this comment.
Pull request overview
This PR fixes accuracy issues in the dlinfer backend by improving numerical precision in rotary embeddings, updating tensor type conversions, and upgrading dependencies to stable releases.
- Refactored rotary embedding inverse frequency calculation to use native float types and eliminate unnecessary type conversions
- Updated Ascend NPU backend tensor operations to use explicit int32 conversions for improved compatibility
- Added support for grouped MoE routing with
n_groupsparameter - Upgraded CANN and torch-npu dependencies from release candidates to stable versions
Reviewed changes
Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| lmdeploy/pytorch/backends/dlinfer/rotary_embedding.py | Improved numerical precision by changing base parameter to float and refactoring inv_freq calculation to avoid intermediate int64 conversions |
| lmdeploy/pytorch/backends/dlinfer/moe.py | Added n_groups parameter support to align with base SoftmaxTopKBuilder interface |
| lmdeploy/pytorch/backends/dlinfer/ascend/op_backend.py | Updated tensor type conversions to explicitly use int32 for Ascend NPU compatibility |
| docker/Dockerfile_ascend_a3 | Upgraded CANN from 8.3.rc1.alpha002 to 8.3.rc1 and torch-npu from 2.8.0rc1 to 2.8.0 |
| docker/Dockerfile_ascend_a2_300i | Upgraded CANN from 8.3.rc1.alpha002 to 8.3.rc1 and torch-npu from 2.8.0rc1 to 2.8.0 |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
jinminxi104
approved these changes
Nov 26, 2025
grimoire
approved these changes
Nov 26, 2025
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
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.
No description provided.