[Cherry-Pick][CI] Pin PaddlePaddle to release/3.3 last_commit build(#7547)

[EmmonsCurse]

Motivation

Due to recent framework-side changes, the default Paddle installation link used in CI for FD release/2.4 branches now points to the latest build from release/3.4 (previously release/3.3).

However, these branches are not yet compatible with Paddle release/3.4. Continuing to use the moving default link introduces compatibility risks and potential CI instability.

To ensure consistent and reliable CI behavior, the Paddle version needs to be fixed to a known compatible build.

Modifications

• Pinned Paddle version in CI for FD release/2.4 branches to the last commit build of release/3.3.
• Replaced floating installation links with fixed wheel package URLs.
• Prevented unintended upgrades to release/3.4 caused by upstream changes.
• Ensured CI stability and reproducibility by locking to a verified compatible version.
• Due to changes in [Cherry-pick][Performance Optimization] Rewrite GPU TopK kernel with radix-select and multi-tier sorting #78409 Paddle#78659, top_logprobs across different k are no longer guaranteed to be strict prefixes; larger gaps (e.g., 5 vs 10) may cause unstable ordering due to floating-point and non-deterministic effects, so we use close values (5 vs 7) for stability.

Usage or Command

N/A

Accuracy Tests

N/A

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!

[EmmonsCurse]

/skip-ci ci_iluvatar
/skip-ci ci_hpu
/skip-ci build_xpu

[PaddlePaddle-bot]

🤖 AI Code Review | 2026-04-21 23:32:49

📋 Review 摘要

PR 概述：将 FD release/2.4 分支 CI 中 PaddlePaddle 安装来源从浮动的 latest 链接固定到 release/3.3 的指定 commit 构建，同时同步修复 top_logprobs 测试的兼容性问题。
变更范围：8 个 CI Workflow 文件 + 1 个 CE 测试文件
影响面 Tag：[CI]

问题

级别	文件	概述
🟡 建议	tests/ce/server/test_compare_top_logprobs.py:108	注释最后一行语句被截断，表意不完整
❓ 疑问	.github/workflows/*.yml（8 个文件）	固定 URL 含具体日期 post20260420，未来版本更新时需同步修改 8 处，建议评估是否可抽取为 Workflow-level 环境变量以降低维护成本

总体评价

变更目标明确、改动最小，成功将 CI Paddle 依赖锁定至已验证兼容的 release/3.3 版本，消除了 release/3.4 意外升级的风险。测试注释存在一处截断，建议补全；固定 URL 分散在 8 个文件是已知维护成本，可后续优化。

[PaddlePaddle-bot]

🟡 建议 注释内容不完整，最后一行结尾被截断

# to be strict prefixes of each other. When k increases, slight variations 这行注释缺少后半句，读者无法理解「slight variations」具体指什么（是会影响结果？还是可以被接受？）。

建议补全注释，例如：

# to be strict prefixes of each other. When k increases, slight variations
# in probability values may appear, causing strict prefix checks to fail.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (release/2.4@f5e3d75). Learn more about missing BASE report.

Additional details and impacted files

@@              Coverage Diff               @@
##             release/2.4    #7548   +/-   ##
==============================================
  Coverage               ?   56.43%           
==============================================
  Files                  ?      333           
  Lines                  ?    42631           
  Branches               ?     6480           
==============================================
  Hits                   ?    24058           
  Misses                 ?    16701           
  Partials               ?     1872           

Flag	Coverage Δ
GPU	56.43% <ø> (?)

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.