ai-algorithm SKILL.md skillA
ai-algorithm from yuecao365/offercome is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add yuecao365/offercome/ai-algorithm@git:20260920.a4843f5Pins git:20260920.a4843f5 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- yuecao365/offercome
- Path
src/lib/mock-interviews/skills/ai-algorithm/SKILL.md- Versions recorded
- 1
- Latest version
- git:20260920.a4843f5, committed Sep 20, 2026
- Size
- 11,817 bytes
- Repository stars
- 24
- License
- MIT
大模型算法怎么面：Transformer、预训练、后训练与对齐、RL、微调、Embedding、评测。LLM 算法岗读。.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
## 面试官在意什么 大模型算法工程师（LLM 算法、大模型研究、后训练 / 对齐、多模态算法）做的是模型本身：造数据、定训练目标、跑预训练或后训练、做对齐、建评测、分析失败样本。和大模型应用岗的区别是：应用岗问"你怎么知道它效果好、坏了怎么查"，算法岗问"为什么这个目标、这份数据、这个算法能让模型变好，你怎么证明是它起的作用"。真实面试三到四轮：一面考 Transformer 与训练基础加手撕（attention、交叉熵、RoPE、DPO loss 这类组件），二面深挖项目里的数据配比、训练曲线、消融与评测，三面考对预训练 / 后训练 / RL 路线的判断与近一年方法的理解，主管面看方向匹配。 面试官最在意三件事：第一，基础能否推导而不是背结论——缩放点积为什么除以根号 d、RoPE 怎么外推、DPO 从 RLHF 目标怎么推出来、GRPO 去掉 critic 靠什么估计优势；第二，实验是否可信——数据有没有泄漏进评测集、消融有没有做、涨点是不是噪声、基线选得公不公平；第三，对"为什么有效"的判断——这份数据为什么该加、这个奖励为什么会被 hack、这个方法在什么规模下会失效。 校招候选人的项目多是复现、比赛、实验室课题或实习里的一段流程，不期待大规模训练经验，看基础推导、对实验的诚实评估、能否解释每个设计的理由；社招必须交代规模（多少卡、多少 token、多少步）、数据管线、评测体系、一次真实的训练事故（loss 尖峰、奖励 hack、评测污染）和当时怎么处理。 怎么问才像这个方向的面试官： - 基础题追到公式或机制层面（推导、复杂度、梯度的方向），只会背名字按不理解处理；手撕以模型组件与损失为主。 - 项目题的核心是实验可信度：评测集与训练集有没有重叠、基线是否公平、涨点是否超过噪声、有没有消融，任何一项答不出都要深挖。 - 不考论文时效性名词（"最新的 SOTA 是什么"），考方法的动机、假设与失效条件；候选人熟悉的方法与题目不同时用他的方法替代。 - 每个训练决策都追"怎么验证"：数据配比、奖励设计、超参，说了任何改动立刻追评测与对照。 - 结合规模出题：几张卡微调 7B 的项目不问千卡并行与 MoE 负载均衡；参与过预训练的必须追数据管线、稳定性与 scaling 判断。 - 2026 年后训练的主线已经从"SFT + RLHF 偏好标注"转到"SFT + 可验证奖励的 RL（GRPO / DAPO 一类）+ 合成数据自博弈"：面试官会追候选人知不知道为什么 RLHF 的人工偏好标注不再是主流、可验证奖励在什么任务上不成立、GRPO 去掉 critic 的代价是什么。只会背 RLHF 三段的按过时处理。 ## 项目 / 实习怎么深挖 …
Read the whole file at its exact version.
How to install
mdr add yuecao365/offercome/ai-algorithm@sha256:745851c085ac5c0eThe line at the top of this page pins the label git:20260920.a4843f5, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_my76d26zrx3mux7z)
5 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 1 version of this file, committed on Sep 20, 2026.
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (11817 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
yuecao365/offercome · 24 stars · license MIT · pushed 2026-10-01 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_my76d26zrx3mux7z GET https://markdownregistry.com/api/v1/resolve?ref=yuecao365/offercome/ai-algorithm GET https://markdownregistry.com/api/v1/blob/745851c085ac5c0e7f83e567e0b20ed7367e1e806fe38702b66933c65872a056
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.