Update on the development branch

[kaiyux]

Hi,

The TensorRT-LLM team is pleased to announce that we have pushed an update to the development branch (and the Triton backend) this March 4, 2025.

This update includes:

• Model Support
  • PyTorch workflow
    • Added Llava-Next support.
    • Added BERT support.
  • Added Qwen2-Audio support. Refer to examples/qwen2audio/README.md.
• Features
  • Added Language-Adapter support. Refer to examples/language_adapter/README.md.
  • Added AutoDeploy MoE Pattern Matcher for the fused_moe op.
  • [BREAKING CHANGE] Added dynamic logits processor support, and deprecated static logits processor in LLM API.
  • Added Deepseek eager mode AllReduce fusion support.
• Bug fixes
  • Fixed incorrect batch slot usage in addCumLogProbs kernel. Thanks to the contribution from @aotman in Fix Incorrect Batch Slot Usage in addCumLogProbs Kernel #2787.

Thanks,
The TensorRT-LLM Engineering Team