图像测试
- 作者仓库星标 16,429
- 许可证 NOASSERTION
- 作者更新于 实时读取
- 作者仓库 Megatron-LM
- 领域
- 工程开发
- 兼容 Agent
-
- Claude Code
- Cursor
- Cline
- Codex
- Windsurf
- Gemini CLI
- +20
- 信任分
- 94 / 100 · 已通过审计
- 作者 / 版本 / 许可
- @NVIDIA · NOASSERTION
- Token 消耗评级
- 低消耗
- 接入复杂程度
- 需手动接入
- 是否需要外部 API Key
- 需要 · Vendor-specific
- 兼容的系统
- Docker
- 底层运行要求
- Python · Docker
- 文件与系统权限
-
- 只读
- 允许写入 / 修改
- 网络行为
- 允许外网请求
- 安装命令数
- 26 条
档案由构建时根据 SKILL.md 与安装命令自动衍生，可能与作者实际意图存在差异。
需要注意： 未限定 allowed-tools，默认拥有全部工具权限。
---
name: cicd
description: CI/CD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering inter…
category: 工程开发
runtime: Python / Docker
---
# cicd 输出预览
## PART A: 任务判断
- 适用问题：代码实现、重构、调试或代码审查。
- 输入要求：目标材料、限制条件、期望输出和验收方式。
- 证据边界：围绕“CI Pipeline Structure / CI Test Scope Labels / Which label to attach when opening a PR”读取原文规则，不把推断写成作者承诺。
## PART B: 执行结果
- **01** 任务判断：确认你的需求是否属于代码实现、重构、调试或代码审查，并标出输入、限制和预期结果。
- **02** 执行计划：优先按“CI Pipeline Structure / CI Test Scope Labels / Which label to attach when opening a PR”拆成步骤，说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果：给出可复制的命令、文件改动、检查清单或内容草稿，并说明如何继续迭代。
- **04** 风险边界：结合 读取文件、写入/修改文件、会按任务需要访问外部网络、需要准备 Vendor-specific API Key 给出执行前确认项。
## Running Rules
- 读取文件、写入/修改文件；会按任务需要访问外部网络；需要准备 Vendor-specific API Key。
- 先小样例验证，再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。 先确认触发方式
原文没有稳定的斜杠命令要求。安装验证后通常全局生效，直接在对话里点名这个 Skill 并描述任务即可。
给清楚输入和边界
告诉 Agent 目标文件或材料、期望结果、不可改范围、是否允许联网或执行命令。本 Skill 的权限画像是：读取文件、写入/修改文件。
小样例验证后再放大
先用一个小任务确认它会围绕“CI Pipeline Structure / CI Test Scope Labels / Which label to attach when opening a PR”工作；涉及文件或命令时，先看 diff、日志、预览或测试结果。
复核后再交付
检查最终产物是否包含明确结果、必要证据和下一步动作；如果输出泛泛而谈，就补充输入、边界和验收标准后重跑。
---
name: cicd
description: CI/CD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering inter…
category: 工程开发
source: NVIDIA/Megatron-LM
---
# cicd
## 什么时候使用
- 把工程方向的常用动作沉淀成 Agent 可调用的技能 适合处理工程开发场景下的代码实现、调试、重构、测试或代码审查，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免 Agent 泛泛回答。 把任务拆成可执行、可检查、可继续迭代…
- 面向代码实现、重构、调试或代码审查，优先处理能明确输入、步骤和验收标准的工作。
## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。
## 执行规则
- 围绕「CI Pipeline Structure / CI Test Scope Labels / Which label to attach when opening a PR」组织步骤，不把推断写成作者事实。
- 读取文件、写入/修改文件；会按任务需要访问外部网络；需要准备 Vendor-specific API Key。
- 先跑小样例，确认结果可检查后再扩大任务范围。
## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown，不编造命令、平台或依赖。 证据边界与执行链路
作者原文负责流程事实；仓库文件负责来源和命令；流狐只补充适用场景、限制和质量判断。
skill "cicd" {
输入层 -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
上下文层 -> CI Pipeline Structure / CI Test Scope Labels / Which label to attach when opening a PR
规则层 -> SKILL.md 触发条件 / 执行顺序 / 输出格式
运行层 -> Python / Docker | 读取文件、写入/修改文件 | 会按任务需要访问外部网络
安全层 -> 需要准备 Vendor-specific API Key + 小任务验证 + diff / 日志复核
输出层 -> 可复制结果 + 检查清单 + 下一步迭代
}CI/CD Guide
CI Pipeline Structure
The main workflow is .github/workflows/cicd-main.yml. It triggers on pushes
to branches matching pull-request/[0-9]+ and deploy-release/*, on merge
groups, on a daily schedule, and on manual dispatch.
is-not-external-contributor
└─ pre-flight
└─ configure # determines scope, container tag, n_repeat
├─ linting
├─ cicd-container-build
│ ├─ cicd-parse-unit-tests → cicd-unit-tests-latest
│ ├─ cicd-parse-integration-tests-h100 → cicd-integration-tests-latest-h100
│ └─ cicd-parse-integration-tests-gb200 → cicd-integration-tests-latest-gb200 (maintainers only)
└─ Nemo_CICD_Test # final pass/fail gateImages are pushed to:
- AWS ECR:
766267172432.dkr.ecr.us-east-1.amazonaws.com/… - GCP Artifact Registry:
us-east4-docker.pkg.dev/nv-projdgxchipp-20260113193621/megatron-lm/…
CI Test Scope Labels
The CI pipeline reads PR labels to decide test scope, n_repeat, and container image.
Decision tree (first match wins):
| Condition | scope |
n_repeat |
lightweight |
Notes |
|---|---|---|---|---|
| Merge group | mr-github |
1 | false | Automatic, no label needed |
Label: Run tests |
mr-github |
1 | true | Trains 4 steps, no golden-value compare |
Label: Run functional tests |
mr-github |
5 | false | Trains 100 steps, golden-value compare |
| (no label) | mr-github-slim |
5 | false | Slim subset only |
Orthogonal image label:
| Label | Effect |
|---|---|
container::lts |
Use the LTS base image instead of dev (combinable with any scope label) |
Run MBridge tests |
Also triggers the MBridge L1 test suite |
Which label to attach when opening a PR
| Changed paths / nature of change | Label to attach |
|---|---|
Docs only (docs/, *.md, docstrings) |
(none) |
CI/tooling only (.github/, tools/, Makefile) |
(none) |
Test files only (tests/) — existing tests, no new golden values |
Run tests |
| New test cases added (no golden values exist yet) | Run functional tests |
Re-enabling a disabled test (scope -broken → active) |
Run functional tests |
| Non-numerical library code (logging, error handling, CLI flags, refactors) | Run tests |
| Could affect training numerics (model arch, attention, optimizer, distributed, MoE routing) | Run functional tests |
Container or dependency changes (docker/, pyproject.toml, uv.lock) |
Run tests + container::lts |
| Touches MBridge integration | add Run MBridge tests |
Rule of thumb: default to Run tests. Always use Run functional tests when the PR adds new test cases (golden values must be generated) or when the change could plausibly shift loss curves.
Triggering Internal CI
Use tools/trigger_internal_ci.py to push the current branch to the internal
GitLab remote and trigger a pipeline — without touching the GitLab UI.
Full setup and usage details: @tools/trigger_internal_ci.md.
Prerequisites (one-time):
# 1. Add the internal GitLab remote
git remote add gitlab git@<gitlab-hostname>:ADLR/Megatron-LM.git
# 2. Create a personal access token with 'api' scope on your GitLab profile,
# then store it:
export GITLAB_TOKEN=glpat-<your-token>Usage:
python tools/trigger_internal_ci.py \
--gitlab-origin gitlab \
[--functional-test-scope mr] \
[--functional-test-repeat 5] \
[--functional-test-cases all] \
[--dry-run]The script force-pushes the current branch as pull-request/<branch> and
prints the resulting pipeline URL.
CI Failure Investigation
CI branches always follow the pattern pull-request/<number>.
Locating the PR from a CI Branch
# Extract PR number from the current branch
PR_NUMBER=$(git rev-parse --abbrev-ref HEAD | grep -oP '(?<=pull-request/)\d+')
# Fetch the PR metadata (title, labels, author, base branch)
gh pr view "$PR_NUMBER" --repo NVIDIA/Megatron-LM
# Show the changeset for that PR
gh pr diff "$PR_NUMBER" --repo NVIDIA/Megatron-LMReading CI Job Logs
# List recent workflow runs for the PR
gh run list --repo NVIDIA/Megatron-LM --branch "pull-request/$PR_NUMBER"
# Stream failing job output
gh run view <run-id> --repo NVIDIA/Megatron-LM --log-failedFull per-rank logs are not in the runner stdout. They are uploaded as
GitHub artifacts named logs-<test_case>-<run_id>-<uuid>.
# 1. Find artifact name
gh run view <run-id> --repo NVIDIA/Megatron-LM --json artifacts \
--jq '.artifacts[].name'
# 2. Download the artifact zip
gh run download <run-id> --repo NVIDIA/Megatron-LM \
--name "logs-<artifact-name>" -D ./ci-logs
# 3. Locate which rank logs contain errors
grep -r -l "ERROR\|Traceback\|FAILED\|fatal" ./ci-logs/
# 4. Log files can exceed 10 000 lines — never read a full log at once.
wc -l ./ci-logs/<test>/<attempt>/attempt_0/<rank>/stderr.log
sed -n '1,200p' ./ci-logs/.../stderr.log # read in chunksIdentifying Failure Root Cause
- Linting failure — re-run
tools/autoformat.shlocally; the diff shows exactly what needs to change. - Container build failure — inspect the
cicd-container-buildjob log. - Unit test failure — the failing bucket is in the
cicd-unit-tests-latestjob matrix. - Functional test failure — look at the
cicd-integration-tests-*job. Start withstdout.logfor rank 0. - Flaky test — the runner retries automatically up to 3 times. If all retries exhausted and the pattern matches a known transient (NCCL, ECC, segfault), it is infrastructure noise.
Correlating a Failure with the PR Changeset
# Find unit tests that cover a changed source file
grep -r "from megatron.core.transformer.attention" tests/unit_tests/ -l
# Check CODEOWNERS for reviewer assignment
cat .github/CODEOWNERS | grep "<changed-path>"
先判断是否适合
作者设计意图
作者的方法与取舍
边界和复核