feat : Add Support of Qwen2.5Omni Model and MiniCPM-o-4_5 Model#612

Open

KKkai0315 wants to merge 21 commits intoUbiquitousLearning:mainfrom

Commits on Jan 23, 2026

feat:add Qwen2.5omni text modal processing
KKkai0315
committed
add qwen2.5omni vision, audio modal
KKkai0315
committed
fix: Enhance quantization modules. Introduced FixedActivationQDQ for fixed quantization parameters, updated ActivationQDQ to use MovingAverageMinMaxObserver, and adjusted eps values for better prec…

chenghuaWang
authored and
KKkai0315
committed
fix: Suppress deprecated comma-subscript warnings in CMake and remove debug print statements from Qwen3DecoderLayer

chenghuaWang
authored and
KKkai0315
committed
feat(qualcomm): Add installation targets for flatbuffers and MllmQNNBackend in CMake, enhance PTQPass with unsolved tensor value checks, and update quantization specifications in RMSNorm and model …

chenghuaWang
authored and
KKkai0315
committed
feat(qualcomm): Refactor Qwen3 model to integrate ConcatObserver for improved quantization, enhance rotate_half function to utilize observers, and ensure consistent scale and zero_point across conc…

chenghuaWang
authored and
KKkai0315
committed
feat(cpu): Implement fill operations for various data types including zeros, ones, specific values, arange, and random fills. Introduce a new fill-inl.hpp file for optimized implementations and upd…

chenghuaWang
authored and
KKkai0315
committed
feat(qnn): Enhance QNNBackend initialization with improved logging and error handling; update default log level to verbose. Add QEmbedding class for quantized embedding operations in PyTorch. Intro…

chenghuaWang
authored and
KKkai0315
committed
feat(qnn): Update quantization handling and embedding output data types; ensure position-independent code for flatbuffers. Enhance context creation with existing context checks and improve weight q…

chenghuaWang
authored and
KKkai0315
committed
feat(qwen3): Integrate QEmbedding for quantized embeddings and refine input layer normalization handling in Qwen3DecoderLayer. Update weight conversion logic in training script to address model com…

chenghuaWang
authored and
KKkai0315
committed
fix
KKkai0315
committed
fix
KKkai0315
committed
Merge remote-tracking branch 'refs/remotes/origin/main'
KKkai0315
committed

Commits on Jan 24, 2026

add ConvTranspose1dOp & TanhOp
KKkai0315
committed

Commits on Jan 25, 2026

fix: fix Tanh op and add test for Tanh Op and ConvTranspose1d Op
KKkai0315
committed

Commits on Feb 23, 2026

add minicpmo45
KKkai0315
committed
merge
KKkai0315
committed

Commits on Feb 24, 2026

add
KKkai0315
committed

Commits on Mar 5, 2026

Commits on Mar 12, 2026

Merge branch 'main' into main
oreomaker
authored