fix(lazy_vlm): update HKVCacheFast and Qwen2.5VL model implementations#441
Merged
chenghuaWang merged 3 commits intoUbiquitousLearning:v2from Sep 12, 2025
Merged
fix(lazy_vlm): update HKVCacheFast and Qwen2.5VL model implementations#441chenghuaWang merged 3 commits intoUbiquitousLearning:v2from
chenghuaWang merged 3 commits intoUbiquitousLearning:v2from
Conversation
Collaborator
chenghuaWang
commented
Sep 12, 2025
chenghuaWang
commented
- Update HKVCacheFast::updateKVCache to use memcpy for faster cache updates
- Modify Qwen2_5VLForCausalLM and Qwen2_5VLText classes to handle position IDs and window indexing
- Update MultimodalRoPEOp to support different input shapes
- Adjust PagedAttnTest to include new test cases for query shapes
- Create C.hpp as the header file for MLLM FFI C API - Create PackedArgs.hpp for defining PackedArgs structure
… executables - Added `lazy_vlm_qwen2vl_fast` and `lazy_vlm_qwen2_5vl_fast` executables. - Updated `CMakeLists.txt` to include the new executables. - Modified `HKVCacheFast.cpp` and `HKVCacheFast.hpp` to add new methods for manual cache length update, visit hidden state cache, and get current sequence count. - Updated `LazyVLMQwen2_5VLFast.cpp` with the main function and necessary includes. - Updated `modeling_qwen2_5vl_fast.hpp` to include `MultimodalRoPEOp` and modify the `makeWindowIndex` function. - Updated `run_remote_android.py` to push the new executables to the remote Android device. - Updated `MultimodalRoPEOp.cpp`, `MultimodalRoPEOp.hpp`, and `MultimodalRoPE.cpp` to support different input types for the `MultimodalRoPEOp`. - Updated `MultimodalRoPEOp.hpp` and `MultimodalRoPE.hpp` to include the new input type enum and constructor parameters.
- Update HKVCacheFast::updateKVCache to use memcpy for faster cache updates - Modify Qwen2_5VLForCausalLM and Qwen2_5VLText classes to handle position IDs and window indexing - Update MultimodalRoPEOp to support different input shapes - Adjust PagedAttnTest to include new test cases for query shapes
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
1 participant
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.