Guoliang Shi

Retraining on the s3dis data had lower IoU

1

Hi, I attempted to retrain semantic segmentation on s3dis. When train on area 1,2,3,4,6 and test on area 5, I got same result in table 3 (mIoU 57.14). But, 6...

Thanks for you nice work. I just want to use DenseCRF to refine my point cloud segmentation methed. The input data of your DenseCRF is "*.pionts" and "probo.txt". Could you...

[NPUW] Fix Omni Thinker LLM with Two Outputs Issue

2

### Details: There is a previous PR (#30535) to fix 3D position input for Omin Thinker LLM. However, it still issue to get the right result caused by two outputs....

## Description Ticket: https://jira.devtools.intel.com/browse/CVS-175909 Fixes #(issue) ## Checklist: - [ ] Tests have been updated or added to cover the new code - [ ] This patch fully addresses the...

[NPUW] Add Eagle3 (top-1) pipeline support with new I/O

6

### Details: This PR introduces support for Eagle3 (top-1) speculative decoding in the NPUW. The main changes include: 1. Added a new `llm_eagle3_extension` module to handle Eagle3-specific input/output logic, including...