[UNIDY2002]

Motivation

Part of #33522.

Some models require a post-load logic. This PR also added support for MXFP4, so models like K3 can be loaded by weight cache.

Modifications

• Added _post_load_weights() to the IPC loading lifecycle.
• Added compressed-tensors to IPC_QUANT_ALLOWLIST
• Some K3-specific patches,

Accuracy Tests

K3 and DSV3 can pass e2e.

Speed Tests and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review and Merge Process

1. Ping Merge Oncalls to start the process. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • Common commands include /tag-and-rerun-ci, /tag-run-ci-label, /rerun-failed-ci
4. After green CI and required approvals, ask Merge Oncalls or people with Write permission to merge the PR.

---

CI States

Latest PR Test (Base): 🚫 Run #34124802751
Latest PR Test (Extra): ✅ Run #34124802402
Latest PR Test (AMD ROCm 7.2): ❌ Run #34124802709

[UNIDY2002]

/tag-run-ci-label extra

[UNIDY2002]

/rerun-failed-ci