[None][doc] Add Glm4MoeForCausalLM to model support matrix

[venkywonka]

Summary

• Add Glm4MoeForCausalLM (GLM-4.5, GLM-4.6, GLM-4.7) to the supported models table
• Add feature support matrix entry for Glm4MoeForCausalLM

Test plan

• Documentation-only change, no functional changes
• Model authors to verify feature support matrix accuracy during review

Summary by CodeRabbit

Release Notes

• Documentation
  • Updated supported models documentation to include GLM-4-MoE model with architecture details and feature support information.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

Documentation update adding a new model entry Glm4MoeForCausalLM to the supported models reference. The change introduces the model into two existing tables documenting architecture information and feature support matrices.

Changes

Cohort / File(s)	Summary
Documentation Update docs/source/models/supported-models.md	Adds Glm4MoeForCausalLM entry to the Architecture table and Model-Feature Support Matrix, detailing model name, HuggingFace example, and feature support status (Overlaps, Untested, N/A).

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The PR title clearly and specifically describes the main change: adding Glm4MoeForCausalLM to the model support matrix documentation.
Description check	✅ Passed	The PR description follows the template structure with a clear Summary and Test Coverage section explaining the changes and testing approach.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[venkywonka]

Feature Support Summary for Glm4MoeForCausalLM

Requesting reviewers to validate the following feature support matrix entry:

Feature	Value	Evidence/Notes
Overlap Scheduler	Yes	Standard PyTorch backend feature
CUDA Graph	Yes	Standard PyTorch backend feature
Attention Data Parallelism	Yes	Verified: enable_attention_dp handling in modeling_glm.py
Disaggregated Serving	Untested	Similar architecture to DeepseekV3 but not explicitly verified
Chunked Prefill	Yes	Uses standard MHA (not MLA), no restrictions
MTP	Yes	Verified: Glm4MTP class in modeling_speculative.py
EAGLE-3 (One Model Engine)	Yes	Supports MTP_EAGLE_ONE_MODEL mode (num_nextn_predict_layers=1)
EAGLE-3 (Two Model Engine)	No	Separate draft model support not implemented
Torch Sampler	Yes	Standard feature
TLLM C++ Sampler	Yes	Standard feature
KV Cache Reuse	Untested	Should work (standard MHA) but not explicitly verified
Sliding Window Attention	N/A	Uses YARN RoPE, not sliding window
Logits Post Processor	Yes	Standard feature
Guided Decoding	Yes	Standard feature

Please confirm or correct any of the above values. The two "Untested" entries are conservative - I could not find explicit tests or documentation confirming these features for this model.

cc: @jhaotingc @symphonylyh @yuxianq @laikhtewari

[venkywonka]

/bot skip --comment "docs only change"

[tensorrt-cicd]

PR_Github #34262 [ skip ] triggered by Bot. Commit: ea810ed

[tensorrt-cicd]

PR_Github #34262 [ skip ] completed with state SUCCESS. Commit: ea810ed
Skipping testing for commit ea810ed