Skip to content

Conversation

yangwang201911
Copy link

@yangwang201911 yangwang201911 commented Sep 9, 2025

Integrating a CDPruner module for visual token pruning into VLM pipeline.
Paper: CDPruner (arXiv)
Code: GitHub Repository

Tickets: CVS-173220

liangali and others added 30 commits August 1, 2025 10:06
…e configuration.

Refactor CDPruner to use visual tokens percentage instead of count for pruning configuration
…ode and remove unused visual token pruning methods
…e arguments and update GenerationConfig structure
…ross codebase for consistency in CDPruner configuration
…in_percentage"

- Updated Python scripts to reflect the corrected parameter name in argument parsing and configuration settings.
2.  Added unit tests for the FastGreedyDPP class to ensure proper functionality and selection behavior based on the visual tokens retention percentage.
…nhance cache performance with manual loop unrolling
…lves for improved relevance computation and DPP selection
…ing invalid values

Refactor performance logging in CDPruner: consolidate timing summary and metrics output
…ent vector subtraction and multiplication functions
…model compilation and enhance token selection with visual token splitting for improved performance

Add support for OpenVINO ops in CDPruner configuration: introduce 'use_ops_model' flag for enhanced pruning computation
@github-actions github-actions bot added category: llm_bench Label for tool/llm_bench folder category: visual language Visual language pipeline category: continuous batching Continuous batching category: sampling Sampling / Decoding algorithms category: cmake / build Cmake scripts category: Python API Python API for GenAI category: CPP API Changes in GenAI C++ public headers no-match-files category: VLM samples GenAI VLM samples labels Sep 9, 2025
…r function for efficient token selection across kernel matrices
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
category: cmake / build Cmake scripts category: continuous batching Continuous batching category: CPP API Changes in GenAI C++ public headers category: llm_bench Label for tool/llm_bench folder category: Python API Python API for GenAI category: sampling Sampling / Decoding algorithms category: visual language Visual language pipeline category: VLM samples GenAI VLM samples no-match-files
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants