<aside> 📌

personal website: krtk.app (looking for internships summer/winter 27)

linkedin: https://www.linkedin.com/in/karthik-gangula/

github: https://github.com/gangula-karthik

</aside>

TODO:

• continuous quantization work: https://github.com/vllm-project/vllm-omni/issues/1854
• lora support for MM models: https://github.com/vllm-project/vllm/issues/31479

feature proposal / issues raised:

• dpsark spec decode in llama.cpp: https://github.com/ggml-org/llama.cpp/issues/25096
• disaggregated diffusion proposal in mooncake: https://github.com/kvcache-ai/Mooncake/issues/2873

benchmaring:

• benchmark torch vs rust preproc vllm: ‣

pull requests raised:

• sarvamai/shuka-1 model support: https://github.com/vllm-project/vllm/pull/47957
• longcat-next model support:
  • longcat-next model support: https://github.com/vllm-project/vllm-omni/pull/5299
  • loncat-next MLA/dual-attention recognition: https://github.com/vllm-project/vllm/pull/49407
• lora support in LlavaNext: https://github.com/vllm-project/vllm/pull/49788

pull requests merged 🎉:

• lora support in LlavaNextVideo: https://github.com/vllm-project/vllm/pull/48594
• update docstring for invert function numpy: https://github.com/numpy/numpy/pull/26646