[Iluvatar] Fix cannot import name mtp_save_first_token

[wuyujiji]

Motivation

该pr导入了mtp_save_first_token和mtp_save_first_token_with_topk两个gpu算子，其他硬件执行到这里会报错，所以需要移到gpu分支里

Modifications

Pass

Usage or Command

Pass>

Accuracy Tests

Pass

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!

[wuyujiji]

/re-run all-failed

[wuyujiji]

/re-run all-failed

[EmmonsCurse]

/skip-ci ci_hpu
/skip-ci build_gpu
/skip-ci build_xpu

[PaddlePaddle-bot]

🤖 AI Code Review | 2026-04-21 11:35:41

📋 Review 摘要

PR 概述：将 mtp_save_first_token 和 mtp_save_first_token_with_topk 两个 GPU 算子的导入从顶层无条件导入移至平台判断的 GPU（else）分支内，修复非 GPU 硬件平台的导入报错。
变更范围：model_executor/pre_and_post_process.py
影响面 Tag：Iluvatar OP

📝 PR 规范检查

PR 描述中 Modifications 部分填写为 "Pass"，建议补充具体的修改内容说明。

描述建议（可直接复制）：

Modifications

将 mtp_save_first_token 和 mtp_save_first_token_with_topk 的导入从文件顶层的无条件导入（from fastdeploy.model_executor.ops.gpu import ...）移至 current_platform 条件分支的 else（GPU/CUDA）分支内，避免 Iluvatar、GCU、HPU 等非 GPU 平台执行到该导入时因找不到 GPU 算子而报错。

问题

未发现阻塞性问题。

总体评价

修复逻辑正确且影响面很小。mtp_save_first_token 和 mtp_save_first_token_with_topk 是 GPU 专属自定义算子（定义在 custom_ops/gpu_ops/ 下），原先被无条件导入会导致非 GPU 平台（如 Iluvatar）启动时报 ImportError。本次修改将其移入 else: 分支（即 CUDA/GPU 默认路径），与同分支内其他 GPU 算子（如 custom_numpy_to_tensor、speculate_limit_thinking_content_length 等）导入方式保持一致，修复合理。