sglang
mtp
speculative-decoding
draft-head
qwen3_5_moe

Commit History

16 days ago

  1. Link four MLX precisions and add local API usage where applicable

    Verified signature
    808a5e3
  2. Unify Occamy model cards, checkpoint navigation and deployment guide

    Verified signature
    73efbbb

  1. Preserve BF16 MTP head in NVFP4 assembly and document community vLLM setup

    Verified signature
    2fd68d3
  2. Add validated native MTP3 runtime and H200 results

    Verified signature
    e563cc5
  3. Release Occamy 1.0 MTP

    755da9f