[Feature] 为 FusedMoE 添加 hidden_size 显式参数支持

[zhoutianzi666]

Motivation
解耦 FusedMoE 对 fd_config.model_config.hidden_size 的强依赖，使类设计更加灵活，便于未来扩展...]

Modifications
在 FusedMoE.init 中新增 hidden_size 参数，默认值为 -1
移除原有从 fd_config.model_config.hidden_size 获取 hidden_size 的逻辑
更新所有调用方显式传入 hidden_size=fd_config.model_config.hidden_size
更新相关测试文件

Usage or Command

Accuracy Tests

Checklist

• Add at least a tag in the PR title.
  • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
  • You can add new tags based on the PR content, but the semantics must be clear.
• Format your code, run pre-commit before commit.
• Add unit tests. Please write the reason in this PR if no unit tests.
• Provide accuracy results.
• If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

[paddle-bot]

Thanks for your contribution!

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

“liuruian” seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[chang-wenbin]

LGTM

[PaddlePaddle-bot]

🤖 AI Code Review | 2026-04-13

📋 Review 摘要

PR 概述：为 FusedMoE 类添加 hidden_size 参数，使调用方可以显式指定 hidden_size 值，而不再从 fd_config.model_config.hidden_size 自动获取。

变更范围：model_executor/layers/moe/moe.py、model_executor/models/ 下的 6 个 MoE 模型文件、4 个测试文件

影响面 Tag：[Models] [OP]

📝 PR 规范检查

问题：

1. PR 标题缺少官方 Tag
2. PR 描述未填写 Motivation 和 Modifications

标题建议（可直接复制）：

• [Feature] 为 FusedMoE 添加 hidden_size 显式参数支持
• [OP] 修改 FusedMoE 构造函数，支持传入 hidden_size 参数

描述模板（可直接复制）：

Motivation

[填写修改原因，例如：解耦 FusedMoE 对 fd_config.model_config.hidden_size 的强依赖，使类设计更加灵活，便于未来扩展...]

Modifications

1. 在 FusedMoE.__init__ 中新增 hidden_size 参数，默认值为 -1
2. 移除原有从 fd_config.model_config.hidden_size 获取 hidden_size 的逻辑
3. 更新所有调用方显式传入 hidden_size=fd_config.model_config.hidden_size
4. 更新相关测试文件

问题

级别	文件	概述
🔴 Bug	moe.py:156	hidden_size 参数默认值为 -1（magic number），且缺少验证逻辑

总体评价

PR 的变更意图合理，将 hidden_size 从自动获取改为显式传入，提高了接口的灵活性。所有现有的调用方都已正确更新，测试文件也同步修改。

但存在一个潜在的安全问题：hidden_size 参数默认值为 -1，且没有验证逻辑。如果未来有新代码忘记传入该参数，会导致 self.hidden_size = -1，可能在后续代码（如 forward_chunked_moe 中的 paddle.empty）中引发错误或意外行为。

[PaddlePaddle-bot]

🔴 Bug hidden_size 参数默认值为 -1（magic number），且没有验证逻辑。

如果未来有代码忘记传入 hidden_size 参数，会导致 self.hidden_size = -1，这可能在后续代码中（如 forward_chunked_moe 中的 paddle.empty(shape=[0, self.hidden_size], ...)）引发错误或意外行为。

建议修复方式：

方案 1（推荐）：移除默认值，强制所有调用方显式传入

def __init__(
    self,
    fd_config,
    hidden_size: int,  # 移除默认值
    ...
):

方案 2：添加验证逻辑

def __init__(
    self,
    fd_config,
    hidden_size: int = -1,
    ...
):
    super().__init__()
    ...
    assert hidden_size > 0, f"hidden_size must be positive, got {hidden_size}"
    self.hidden_size = hidden_size

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (develop@225fc8d). Learn more about missing BASE report.

Additional details and impacted files

@@            Coverage Diff             @@
##             develop    #7361   +/-   ##
==========================================
  Coverage           ?   74.10%           
==========================================
  Files              ?      383           
  Lines              ?    53591           
  Branches           ?     8406           
==========================================
  Hits               ?    39713           
  Misses             ?    11181           
  Partials           ?     2697           

Flag	Coverage Δ
GPU	74.10% <100.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.