deployment_guide: link the official GLM-5.3-Flash recipes; explain --trust-remote-code and the vLLM patch

#2
Intern Large Models org Hugging Face CLI

Only deployment_guide.md changes.

  • Links the official GLM-5.3-Flash recipes for vLLM and SGLang (same architecture) for installation and tuning, and lists what Sage-S2.1-Flash needs on top of them.
  • New section "Required for Sage-S2.1-Flash": --trust-remote-code on every engine, with what goes wrong without it, and the vLLM patch.
  • vLLM: explains why the patch is needed (images in tool results returned out of call order are attached to the wrong results) and links the upstream fix in progress, vllm-project/vllm#67108; notes that the official recipe's Rust frontend does not accept images yet, and the known non-streaming parsing issue vllm-project/vllm#64326.
haijunlv changed pull request status to merged

Sign up or log in to comment