90 / 90

第 90 天：作品集组装、出处审计与后续学习计划

为已验收的学习项目组装可复现的证据作品集，重跑公开验证命令，并以诚实、循证的计划为课程收官。

自主学习与 Python 480 分钟有效研习（8 小时）
01 120 分钟

作品集清单：出处、可复现性、主张与局限

Agent 辅助工程检查

前置要求

第 89 天掌握台账；已完成的学习项目链。

学习目标

定义每个作品集条目必须满足的清单——主张、证据路径、复现命令、观测结果、成本/授权状态、局限——以已验收项目证据和行业出处格式为据。

操作指引

1. 条目必备字段：{project, claim, evidence_path, verify_command, observed_result, status, limitations}。status 词表固定：verified-local（本机命令已验证）、verified-disposable（一次性环境内已验证）、blocked-awaiting-authorization（需要尚未授予的批准/付费/实时访问）。
2. 已验收证据到条目映射：
- Autograd 项目（工单 17）：projects/autograd engine + gradcheck；经 python3 projects/autograd/cli.py forward-backward 与 gradcheck 为 verified-local。
- 可续训预训练（工单 18/19）：projects/pretrain/artifacts/run_artifact.json——10 步 CPU smoke，train_loss 4.0316，val_loss 4.0372，7440 参数；verified-local。云 GPU 预训练证据：blocked-awaiting-authorization。
- 真LoRA 机制（工单 20/21）：projects/lora/artifacts——loss 4.0976→4.0945、perplexity 60.20→60.01、merge_verified true（仅机制验证）；云前后对比：依 comparison_contract.json 为 blocked-awaiting-authorization。
- 可评测 Agent（工单 22/23）：projects/agent_learning deterministic-v1 评分器+夹具；verified-local。真实模型评测：依 live_eval_contract.json 为 not-run-awaiting-authorization。
- CI 交付练习（工单 24）：scripts/ci_practice_runner.py + tests/engineering/test_ci_practice.py；verified-disposable。
3. 诚实规则（逐字）：不得把受阻项升级为已验证；每条主张须引用提交哈希与工件路径；smoke 规模结果不得当作收敛呈现。
4. 模型卡与数据说明书约定了业界期待的局限字段——借鉴其字段，而非抄其文本。

自主演练与记录

起草 checklist.md：列每条目必备 6 字段、3 个 status 值、2 条诚实规则，并标注每个已验收项目对应的 status。

独立掌握标准

清单覆盖全部 6 字段、3 种状态，并正确标注每个已验收项目的已验证与受阻待授权证据。

权威源资源与权利证明

linking CC BY 4.0

The Turing Way: A Handbook for Reproducible, Ethical and Collaborative Data Science

作者 / 机构: The Turing Way Community | 原始出处链接 ↗

对应章节 / 时间戳: Reproducible Research chapter: version control, testing, environment capture

验证版本与日期: Zenodo DOI 10.5281/zenodo.3233853, continuously updated (Verified: 2026-10-30)

入选理由:

Community-maintained CC-BY handbook codifying reproducibility checklists (versioned code, environment capture, rerunnable commands); directly defines the Day 90 reproduction-verification and provenance-audit workflow.

来源前置条件:

Basic version control and command-line familiarity.

安全回退方案（离线弹性）:

The portfolio checklist and verification log schema are authored locally; reproduction commands run without the handbook.

对比候选资源与选型分析 (2)
Wilson et al. Good enough practices in scientific computing (PLOS CB 2017) ↗ compared_alternative [CC BY 4.0]

Concise practices paper; The Turing Way's maintained, checklist-driven coverage of environments and reproducibility is broader and kept current.

Sandve et al. Ten Simple Rules for Reproducible Computational Research (PLOS CB 2013) ↗ compared_alternative [CC BY 4.0]

Foundational rules list; The Turing Way expands the same rules into actionable checklists used verbatim for the Day 90 verification run.

linking arXiv non-exclusive license / Open Access

Model Cards for Model Reporting

作者 / 机构: Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, Inioluwa Deborah Raji, Timnit Gebru | 原始出处链接 ↗

对应章节 / 时间戳: Section 4 Model Card Sections (4.1-4.9)

验证版本与日期: FAT* 2019 / arXiv:1810.03993v2 (Verified: 2026-10-30)

入选理由:

Primary standard for structured, transparent model reporting, mandating explicit disclosure of intended uses, evaluation slices, out-of-scope tasks, ethical considerations, and negative results.

来源前置条件:

Evaluation reporting, demographic and performance slicing, model limitations and caveats.

安全回退方案（离线弹性）:

Complete Model Card Markdown templates, section schemas, and evaluation disclosure checklists are provided in the curriculum.

对比候选资源与选型分析 (2)
Bommasani et al. The Foundation Model Transparency Index ↗ compared_alternative [Open Access]

Corporate transparency audit index; Mitchell et al. provides the concrete technical artifact specification for individual model releases.

Pushkarna et al. Data Cards: Purposeful and Transparent Dataset Documentation for Responsible AI ↗ compared_alternative [Open Access]

Dataset documentation focus; Mitchell et al. provides the required specification for trained model artifact reporting.

02 120 分钟

组装作品集清单

独立掌握检查（无 AI 辅助）

前置要求

第 90 天活动 1 清单；辅助条件：独立。

学习目标

编写 portfolio_manifest.json，把每条学习主张映射到仓库工件路径、复现命令、诚实 status 与局限说明——无辅助完成，每条路径在仓库内可解析。

操作指引

1. 条目模式：{project, claim, evidence_path, verify_command, observed_result, status, limitations}。
2. 每个已验收项目（autograd、pretrain、lora、agent_learning、ci-practice）各写一条，另加一条课程本身。
3. 每个 evidence_path 必须能在仓库中解析；记录前用 ls 或 test -f 验证。
4. 每条主张引用观测值：如 autograd gradcheck 的 max-abs-diff 在容差内；pretrain run_artifact.json 的 total_parameters=7440、val_loss=4.0372；lora eval_results.json 的 loss_delta=0.0031、merge_verified=true；agent_learning 对 summarize-notes 夹具的评分输出；ci_practice_runner 的阶段证据。
5. limitations 字段必填且具体：如“10 步 CPU smoke 只验证管线而非收敛”“确定性适配器不是真实模型”“合成指令数据不是基准”“随机权重 smoke 机制不是质量证据”“契约校验不是已授权云运行”。

自主演练与记录

输出至少含 5 个项目条目+1 个课程条目的 portfolio_manifest.json；断言每个 evidence_path 存在且 status 使用固定词表。

独立掌握标准

清单解析到真实工件路径、诚实状态与每条主张的明确局限——无辅助产出。

权威源资源与权利证明

linking arXiv non-exclusive license / Open Access

Datasheets for Datasets

作者 / 机构: Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jennifer Wortman Vaughan, Hanna Wallach, Hal Daumé III, Kate Crawford | 原始出处链接 ↗

对应章节 / 时间戳: Section 3 Questions and Workflow (3.1 Motivation through 3.7 Maintenance)

验证版本与日期: Communications of the ACM 2021 / arXiv:1803.09010v8 (Verified: 2026-10-30)

入选理由:

Foundational literature introducing standardized documentation for machine learning datasets, specifying questions on dataset composition, collection mechanisms, preprocessing, and distribution rights.

来源前置条件:

Dataset provenance, curation motivation, preprocessing transparency, maintenance lifecycle.

安全回退方案（离线弹性）:

Datasheet questionnaires, provenance schemas, and dataset auditing checklists are completely integrated into lesson materials.

对比候选资源与选型分析 (2)
Bender & Friedman. Data Statements for Natural Language Processing: Toward Mitigating System Bias and Enabling Better Science ↗ compared_alternative [CC BY 4.0 (ACL Anthology, 2016 or later)]

Linguistic demographic documentation focus; Gebru et al. provides the universal engineering datasheet framework adopted across AI.

Holland et al. The Dataset Nutrition Label: A Framework to Drive Higher Data Quality Standards ↗ compared_alternative [Open Access]

Nutrition label metaphor; Gebru et al. is a widely cited framework for dataset documentation in both academic and industrial practice.

linking arXiv non-exclusive license / Open Access

Model Cards for Model Reporting

作者 / 机构: Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, Inioluwa Deborah Raji, Timnit Gebru | 原始出处链接 ↗

对应章节 / 时间戳: Section 4 Model Card Sections (4.1-4.9)

验证版本与日期: FAT* 2019 / arXiv:1810.03993v2 (Verified: 2026-10-30)

入选理由:

Primary standard for structured, transparent model reporting, mandating explicit disclosure of intended uses, evaluation slices, out-of-scope tasks, ethical considerations, and negative results.

来源前置条件:

Evaluation reporting, demographic and performance slicing, model limitations and caveats.

安全回退方案（离线弹性）:

Complete Model Card Markdown templates, section schemas, and evaluation disclosure checklists are provided in the curriculum.

对比候选资源与选型分析 (2)
Bommasani et al. The Foundation Model Transparency Index ↗ compared_alternative [Open Access]

Corporate transparency audit index; Mitchell et al. provides the concrete technical artifact specification for individual model releases.

Pushkarna et al. Data Cards: Purposeful and Transparent Dataset Documentation for Responsible AI ↗ compared_alternative [Open Access]

Dataset documentation focus; Mitchell et al. provides the required specification for trained model artifact reporting.

03 120 分钟

可复现性验证：重跑公开项目命令

Agent 辅助工程检查

前置要求

第 90 天活动 2 清单；辅助条件：agent_assisted（Agent 可协助捕获日志，但验证证据是命令输出本身）。

学习目标

在干净 shell 中重跑每个项目的公开验证命令、捕获输出，确认作品集主张可复现——如实记录环境细节与任何漂移。

操作指引

1. 先记录环境：python3 --version、uname -a（或等价物）、git rev-parse --short HEAD。
2. 运行各公开命令并捕获 stdout+退出码：
- python3 projects/autograd/cli.py forward-backward
- python3 projects/autograd/cli.py gradcheck
- python3 -m unittest tests.unit.test_autograd_engine
- python3 -m projects.agent_learning.cli evaluate --fixture summarize-notes
- python3 projects/lora/evidence/validate_contract.py（--require-authorized 在批准缺失时*应当*失败——把拒绝记为正确行为）
- python3 scripts/audit_calendar_ledger.py
3. 把每个输出与已记录工件字段对比（gradcheck 容差、台账合计、契约 status blocked-awaiting-authorization）。每条命令跑两次——一次冷启动、一次读过工件后——对输出做 diff。
4. 记录 verification_log.json：{command, exit_code, key_output, artifact_field_match: true|false, notes}。任何漂移（路径变更、版本差异、工件缺失）须记录不得隐藏。

自主演练与记录

输出覆盖全部 6 组命令的 verification_log.json；每行带实际观测的输出摘要与匹配判定。

独立掌握标准

全部复现命令已执行并捕获输出；预期拒绝正确标记；不匹配如实记录不作隐藏。

权威源资源与权利证明

linking CC BY 4.0

The Turing Way: A Handbook for Reproducible, Ethical and Collaborative Data Science

作者 / 机构: The Turing Way Community | 原始出处链接 ↗

对应章节 / 时间戳: Reproducible Research chapter: version control, testing, environment capture

验证版本与日期: Zenodo DOI 10.5281/zenodo.3233853, continuously updated (Verified: 2026-10-30)

入选理由:

Community-maintained CC-BY handbook codifying reproducibility checklists (versioned code, environment capture, rerunnable commands); directly defines the Day 90 reproduction-verification and provenance-audit workflow.

来源前置条件:

Basic version control and command-line familiarity.

安全回退方案（离线弹性）:

The portfolio checklist and verification log schema are authored locally; reproduction commands run without the handbook.

对比候选资源与选型分析 (2)
Wilson et al. Good enough practices in scientific computing (PLOS CB 2017) ↗ compared_alternative [CC BY 4.0]

Concise practices paper; The Turing Way's maintained, checklist-driven coverage of environments and reproducibility is broader and kept current.

Sandve et al. Ten Simple Rules for Reproducible Computational Research (PLOS CB 2013) ↗ compared_alternative [CC BY 4.0]

Foundational rules list; The Turing Way expands the same rules into actionable checklists used verbatim for the Day 90 verification run.

linking BSD-3-Clause

PyTorch Tutorial: Saving and Loading Models

作者 / 机构: PyTorch Documentation Team | 原始出处链接 ↗

对应章节 / 时间戳: Saving & Loading a General Checkpoint for Inference and/or Resuming Training

验证版本与日期: PyTorch 2.x Documentation (Verified: 2026-10-30)

入选理由:

Official documentation detailing state_dict serialization, optimizer state saving, and safe loading patterns.

来源前置条件:

Python, PyTorch tensors and nn.Module basics.

安全回退方案（离线弹性）:

All checkpoint serialization semantics, dictionary schema keys, and safe recovery steps are self-contained in lesson code.

对比候选资源与选型分析 (2)
TensorFlow tf.train.Checkpoint Guide ↗ compared_alternative [Apache 2.0]

TensorFlow trackable checkpoint architecture; PyTorch's explicit dictionary is our target implementation standard.

Hugging Face Safetensors Documentation ↗ compared_alternative [Apache 2.0]

Safetensors provides safe zero-copy tensor storage without pickle, compared directly when discussing pickle execution risks.

04 120 分钟

终审复习与循证后续学习计划

独立掌握检查（无 AI 辅助）

前置要求

第 90 天活动 1-3 完成；第 89 天掌握台账。

学习目标

产出最终课程评审与后续学习计划：锚定台账剩余缺口的间隔复习排期、接入保留的进阶年度路线、以及诚实的结课声明。

操作指引

1. 总结掌握台账：各域前后小计、带恢复路径的剩余缺口。
2. 为每个仍未达项安排 +1/+3/+7/+30 天的间隔检索——每次给具名资源与新变体任务（不是同一题）。
3. 把后续学习映射到保留的进阶年度路线（导航链接：历史年度路线 / legacy roadmap at /lessons/index.html）：扩展概率统计、更大规模预训练、部署与服务——90 天范围外主题。
4. 按其本意写下结课声明：完成 90 天课程产出的是已演示任务的学习证据——不是官方认证、不保证人人三个月掌握、辅助结果不能替代独立能力。未达标准保持可见未达。
5. 输出 next_learning_plan.json：{remaining_gaps[], scheduled_reviews[], annual_route_links[], honest_scope_statement}。

自主演练与记录

产出 next_learning_plan.json：为每个剩余缺口排好复习计划，并逐字记录诚实范围声明。

独立掌握标准

终审文档、间隔排期、年度路线链接与诚实范围声明完成——未达标准保持可见未达。

权威源资源与权利证明

linking Journal publisher copyright (all rights reserved) - lawful external linking only

Improving Students' Learning With Effective Learning Techniques: Promising Directions From Cognitive and Educational Psychology

作者 / 机构: John Dunlosky, Katherine A. Rawson, Elizabeth J. Marsh, Mitchell J. Nathan, Daniel T. Willingham | 原始出处链接 ↗

对应章节 / 时间戳: Psychological Science in the Public Interest 14(1): 4-58; practice testing and distributed practice sections

验证版本与日期: Psychological Science in the Public Interest, 2013 (Verified: 2026-10-30)

入选理由:

Detailed comparative review rating practice testing and distributed practice as high-utility techniques while rating highlighting and rereading low-utility; anchors the retrieval, correction, and spaced-review design of Days 88-90.

来源前置条件:

None beyond general academic reading.

安全回退方案（离线弹性）:

Curriculum restates the utility ratings and application conditions in original prose; every task remains runnable offline.

对比候选资源与选型分析 (2)
Pashler et al. Learning Styles: Concepts and Evidence ↗ compared_alternative [Publisher copyright (originally Wiley; now SAGE Publications)]

Important negative result on learning-styles matching but narrower scope; Dunlosky et al. rates a broader set of techniques directly usable for this curriculum.

Brown, Roediger & McDaniel: Make It Stick: The Science of Successful Learning ↗ compared_alternative [Publisher copyright (Harvard University Press)]

Strong trade-book synthesis of the same evidence base; the peer-reviewed review is preferred for citation precision and technique-by-technique utility ratings.

linking CC BY-SA 4.0 (book text) / modified MIT (sample code) share-alike

Dive into Deep Learning (d2l.ai)

作者 / 机构: Aston Zhang, Zachary C. Lipton, Mu Li, Alexander J. Smola | 原始出处链接 ↗

对应章节 / 时间戳: Section: Model Selection, Underfitting, and Overfitting; Chapter: Optimization Algorithms

验证版本与日期: Current online edition (Verified: 2026-10-30)

入选理由:

Primary open reference tying evaluation discipline (model selection, underfitting, overfitting) and optimization algorithms (gradient descent, minibatch) to the rest of the curriculum.

来源前置条件:

Linear algebra, calculus, and probability at the Days 19-28 level.

安全回退方案（离线弹性）:

All required formulas and worked numbers are reproduced as original explanations; the book is further reading.

对比候选资源与选型分析 (2)
Deep Learning (Goodfellow, Bengio, Courville) ↗ compared_alternative [MIT Press; free HTML for reading, all rights reserved]

Detailed treatment of optimization difficulties (chapter 8) including saddle points and plateaus; all-rights-reserved text kept as a linking-only alternative.

fast.ai: Practical Deep Learning for Coders ↗ compared_alternative [Course site copyright / Apache-2.0 (fastai code)]

Strong practitioner framing, but organized around the fastai library and video lessons rather than self-contained mathematics.