Dmitry Nikolaev (@macleginn) header
Dmitry Nikolaev (@macleginn) profile picture

Lecturer at the University of Manchester. Computational linguistics, NLP, digital humanities, Celtic studies. He.

Joined July 2015
Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)Photo by Dmitry Nikolaev (@macleginn)
Needed an update.
2
76
Dmitry Nikolaev retweeted
Obituaries The demise of Jensen Huang was the end of a great many things, including Nvidia’s long-term prospects, the AI bubble, and the exceptional way in which the U.S. viewed itself in the world. stratechery.com/2026/obituar…
42
196
85
1,422
591,314
Dmitry Nikolaev retweeted
教授「c言語で配列を参照するにはa[i]と書きます」 自分[ふーん] 教授「ちなみにi[a]でもいいです」 ふーん[自分]
78
2,768
327
27,961
1,590,474
Revealed preferences
29
In more interesting AI news.
26
Dmitry Nikolaev retweeted
斯坦福CS329Z — Engineering AI Agents(2026 秋季)这门课程现在终于在youtube上线了。 youtube.com/watch?v=DGI4vucR… 这是一门全新的斯坦福研究生课程,由 Diyi Yang、Michael Ryan(DSPy 核心贡献者)和 John Yang(SWE-bench 共同创建者)共同授课。⁠ 核心定位是:从单一大模型转向复合 AI 系统(Compound AI Systems) 的工程实践。也就是把 LLM、检索器、工具、优化器、记忆等组件组合起来,构建从简单流水线到自主 Agent 的完整系统。 课程强调动手工程能力,而不是只讲概念: 先从零实现核心组件(RAG、工具调用、Agent Loop),再学习框架(如 DSPy)如何抽象这些模式。 重点训练:问题拆解、组件选型、数据收集与整理、评测体系搭建、设计权衡、可靠性与安全。 主要内容覆盖 Agent 基础与解剖结构(LLM 核心、推理、记忆、工具、环境) RAG(索引、检索器、Agentic RAG) 工具使用与函数调用(包括 MCP) 框架与编排(DSPy、LangGraph 等) Agent 设计模式(ReAct、Reflexion、CodeAct 等) 记忆架构、多智能体系统 优化(提示、权重、系统级) 评测(LLM-as-Judge、benchmark 设计、pass@k 等) 安全与防护(prompt injection、沙箱、权限模型) 专项主题:编程 Agent、主动式 Agent 还有嘉宾课,包括 Terminal-Bench、SWE-bench 团队,以及与 Andrew Ng 的 fireside chat。 作业与实践(已公开) HW1(Michael Ryan 设计):从零搭建 Agent Harness(类似迷你版 Claude Code)。要求不用现成框架,只用 chat completion API + 自己写代码。基于真实公司邮件语料做检索、摘要、工具调用、终端、记忆、人工审批等。成本很低(参考实现跑完测试大约不到 1 美元)。⁠ 仓库:github.com/cs329z/assignment… HW2(John Yang 设计):使用 Harbor 框架构建接近 Terminal-Bench 风格的真实 Agent 评测任务。学生要自己设计问题、运行环境和验证器。 仓库:github.com/cs329z/assignment… 课程网站:cs329z.stanford.edu/ YouTube 播放列表已开始更新(Lecture 1 已上线):youtube.com/watch?v=DGI4vucR… 整体来说,这是目前斯坦福最系统、最偏工程实践的 AI Agent 课程之一,重点在“怎么真正把 Agent 系统做出来并可靠运行”,而不是只调用现成框架。材料正在全面公开,现在是跟着学的好时机。
16
101
1
504
44,691
A somewhat weird aspect of the @ReviewAcl system is that one can get assigned a paper to review, start reading it, and only then find out that it was desk rejected. Isn’t the purpose of desk rejection to not send substandard papers to reviewers in the first place?
1
45
Dmitry Nikolaev retweeted
You can tell because, in the rare event that you choose a task for which they have no relevant RL data, it's not that hard to catch up to even the largest and most effective frontier models just by SFT and RL of a very small open model
2
4
116
3,187
Having finished a prep session for the Life in the UK test (the arts, culture, and leisure section) I open X to find that
🚨 NEW: The following porn is illegal to possess or publish in England and Wales from tomorrow: - Strangulation or suffocation, such as restricting breathing through facesitting or hand over mouth - Penetrative sex between family members or people pretending to be related (like father and daughter), including step and foster-family scenarios where one person is or appears under 18 - Adults roleplaying as children under 16, such as wearing school uniform Anyone who currently possesses the content above must delete it before tomorrow or face up to 3 years in prison, or 5 years if they publish it
1
88
I like how the air-pollution section of the Met Office forecast is always oblivious of whatever else is happening at the moment.
1
1
36
What kind of x-axis is this?
1
31
We came to the point when exactly the same things are being said about music and mathematics. ft.com/content/2c21c250-d186…
1
1
74
There is a meme abt answering your own SO/Reddit question incorrectly to provoke ppl into giving the real answer. I’ve suddenly realised that this is my one and only strategy for making ppl explain things to me: ask a seed question, rephrase the answer, rile them up, repeat.
1
50
Over the years, I bought three Bialetti products (a mid-size moka pot, a larger one, and a cafetiere/French press). Two out of three were of very low quality, so my global market is now lost to them.
1
52
I have always been somewhat puzzled by how/why people cite my work, but now that hallucinated references have started coming in, I can finally put these thoughts to rest.
3
60
Dmitry Nikolaev retweeted
I don't think this is true: it will just select for a very particular kind of mathematician: the G. H. Hardy. If you read C. P. Snow's foreword to the Apology, it becomes clear that Hardy's most substantive mathematical contributions came about from taking the highly creative and technically impressive (but otherwise utterly inexplicable) accomplishments of Littlewood and Ramanujan, and figuring out how to explain, interpret, and communicate them in a manner that was simultaneously rigorous, precise, and beautiful. It's ironic given his derogatory comments about mathematical exposition in the Apology, but Hardy's real skill lay in the beauty and clarity of his prose and his explanation. He was always the one who wrote up the final manuscript for publication, but he was almost never the one who provided the underlying results or ideas. (Again, Snow's foreword makes clear that the Apology is, amongst other things, a work of immense self-hatred.) In a world where Littlewoods and Ramanujans are commodified, be a Hardy.
>this means mathematicians job is now to interpret AI results to figure that shit out yall realize this is going to make mathematics research soul crushingly boring to do so the field will collapse in less than a generation because nobody will want to be a mathematician anymore
35
54
11
600
36,940
Dmitry Nikolaev retweeted
10 remarks on AI and math that can be true all at once: 1. The recent progress is stunning and exciting. 2. The Quasi-Riemann Hypothesis is arguably Fields-medal level. 3. Mathematics is being transformed and there is no way back. 4. OpenAI doesn't give a fuck about the future of mathematics. 5. Half-baked unreadable preprints claiming major results can effectively stall mathematical progress. 6. Some mathematicians are responding with naive, overly defensive comments. 7. The social-media wankers framing this as "OpenAI tramples gatekeeping" have no experience of mathematical research. 8. The product of mathematics is clarity and understanding. Not theorems, by themselves. 9. Mathematics is a critical infrastructure of human cognition. The human need for understanding isn't obsolete and it can't be offloaded to machines. 10. We haven't "solved" mathematics and we won't get there before the Sun burns out.
158
465
58
2,964
160,013
We don't know much about the prehistorical state of mankind, but probably there was much less economical coercion, bc there was no economy. Then there was all of recorded history, which mostly consisted of various types of economic coercion, and now we're back to light coercion.
RIP Chris Wickham. Unlike so many historians who emphasise particularities and contingencies, he was the master of constructing grand, coherent, beautifully rendered worlds out of his evidence. This, from his 2021 Past and Present essay, is a masterpiece of historical writing.
1
1
142
So humans are seemingly evolutionary-psychologically not very well suited to economic coercion (although this may vary by population?). And now we are not doing that well with light economic coercion (taxes for welfare) either, but what are the alternatives.
1
41