Intelligence.Log
2026-05-17
Extracted: 49 items. Sources: GitHub, Bluesky, X.
++ AI OVERVIEW ++
Today's trending signals point to a dual focus on foundational tooling and the cultural shape of AI development. On GitHub, Python infrastructure is getting a sharp upgrade, with **psf/pypistats.org** (a PyPI analytics dashboard) and **mschwager/cohesion** (a class cohesion measurer) both gaining traction via simonw’s stars, while the ever-dominant **llama.cpp** crosses 110K stars, cementing its role as the go-to for local LLM inference. On Bluesky, Ethan Mollick sparked a lively debate by revealing GPT-5.5 Pro’s attempt at humor analysis produced “scrotum snorkel” and “tuba subpoena,” underscoring that even advanced models struggle with nuanced creativity. Nathan Lambert’s call for more “zagging” in AI—valuing independent thought over Silicon Valley monoculture—resonated, while Emily M. Bender took a sharp jab at misplaced graduation speeches, and Amy Zhang celebrated Kevin Feng’s dissertation on designing interactive systems for “generally capable AI,” marking a milestone for the Social Futures Lab. The throughline: a community balancing rigorous engineering with a growing appetite for critical, human-centered perspectives on where the technology is heading.
◆ Signal
Co-Starred · Last 7 days
Repos independently starred by multiple AI leaders in the week ending 2026-05-17. Stronger signal = more overlap.
antirez/ds4
×2 starrers▲ 8/10★ 9.0k
DeepSeek 4 Flash local inference engine for Metal and CUDA
grep TOPIC=
grep SOURCE=
sort --by=
GH
LLM inference in C/C++
“llama.cpp enables efficient LLM inference in C/C++ with minimal dependencies, supporting a wide range of models including LLaMA, Mistral, and GPT-2. It features quantization, GPU acceleration, and a lightweight server for local deployment.”
GH
A tool for measuring Python class cohesion.
“Cohesion is a flake8 plugin that measures Python class cohesion using the Lack of Cohesion of Methods (LCOM) metric. It helps developers identify classes that may be doing too much and should be refactored, improving code maintainability and adherence to single responsibility principle.”
GH
PyPI downloads analytics dashboard
“PyPI Stats provides a public dashboard for viewing download statistics of Python packages from PyPI. It offers insights into package popularity and trends over time, useful for developers and maintainers.”
BSKY
Being out of SF has lowered my information proximity but with the big upside of giving me space to cultivate my own beliefs and values around ai. We need more people zagging in AI, the monoculture just helps the incumbents win at this point.
❤️ 40 Likes|
BSKY
GPT-5.5 Pro faces its hardest academic challenge: to apply the technique from a paper analyzing which word pairs were funny & why to come up with its own It came up with scrotum snorkel, tuba subpoena, waffle coffin, toad commode, diarrhea tiara, banana tribunal & muffin ruffian.
❤️ 39 Likes|[LLM]
BSKY
“Data centers create economic activity, especially in directly related sectors and during construction, and they are associated with larger county-level income aggregates. They also raise electricity prices and are associated with higher house prices,” www.nber.org/papers/w35194
❤️ 28 Likes|[Infra]
BSKY
I know the upcoming film version of the Odyssey is controversial, so I whipped together a completely accurate take that I think will be happily accepted by everyone as the most definitive version since Homer's original, if not more so.
❤️ 68 Likes|
BSKY
The kids are alright. Quite on top of everything else, what kind of bozo makes a graduation speech that isn't actually about the graduates?
❤️ 154 Likes|
BSKY
Congrats to @kjfeng.me on passing his dissertation defense, titled "Designing for Interactive Systems Powered by Generally Capable AI"! Kevin is @socialfutureslab.bsky.social's 3rd PhD graduate! Thanks to his all-star committee for thoughtful Qs: David McDonald @emtseng.bsky.social Dan Weld @mako.cc
❤️ 8 Likes|
X
LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating
[LLM][RAG][Tooling]
“DeepSeek Summary: Karpathy uses LLMs to build personal knowledge bases for research topics, shifting token usage from code manipulation to knowledge manipulation.”
X
I'm being accused of overhyping the [site everyone heard too much about today already].
[LLM]
“DeepSeek Summary: Karpathy responds to criticism of overhyping a popular site, likely related to AI or tech.”
X
Very interested in what the coming era of highly bespoke software might look like. Example from this morning - I've become a bit loosy goosy with my cardio recently so I decided to do a more srs, regimented experiment to try to lower my Resting Heart Rate from 50 -> 45, over https://t.co/EDULdIpWmE
[Agent][Tooling]
“DeepSeek Summary: Karpathy explores bespoke software for personal health experiments, aiming to lower his resting heart rate.”
X
Vibe coding is irresponsibly building software through dice rolls, not caring what code is produced
[LLM][Safety]
“DeepSeek Summary: Simon Willison criticizes 'vibe coding' as an irresponsible approach to software development.”
X
A short note that the predictions that LLMs would favor 'boring technology' that's
[LLM][Evaluation]
“DeepSeek Summary: Simon Willison notes that LLMs may not favor boring technology as predicted.”
X
I am not excited about visual workflow builders 1. Not simple enough for the average user
[Tooling][Agent]
“DeepSeek Summary: Harrison Chase expresses skepticism about visual workflow builders, citing lack of simplicity for average users.”
X
We launched LangSmith Agent Builder this week as a no-code way to build agents. A key part of Agent builder is it's memory system.
[Agent][Tooling]
“DeepSeek Summary: Announcement of LangSmith Agent Builder, a no-code agent builder with a focus on memory systems.”
X
In the hot path as the agent is running. The agent can decided to (or the user can prompt it to) update its memory as it is working on the core
[Agent][Infra]
“DeepSeek Summary: Describes how agents can update memory during execution, either autonomously or via user prompt.”
X
Your harness, your memory ... The “best” way to build agentic systems has changed dramatically over the past three years. When ChatGPT came out,
[Agent][LLM]
“DeepSeek Summary: Reflects on the evolution of best practices for building agentic systems since ChatGPT's release.”
X
It gives me a lot of comfort knowing that we are the last generation without advanced robots everywhere.
[Agent]
“DeepSeek Summary: Reflects on the imminent ubiquity of advanced robots, suggesting a future where robots are commonplace.”
X
Resource constraints are a beautiful thing. Survival instinct in a cut-throat AI competitive land.
[Agent]
“DeepSeek Summary: Views resource constraints as a positive force driving innovation and survival in AI competition.”
X
In this context, I define world modeling as predicting the next plausible world state (or a longer duration of states) conditioned on an action.
[Multi-modal]
“DeepSeek Summary: Defines world modeling in the context of AI and robotics as predicting future states based on actions.”
X
Here's a complete unedited video of asking Grok for its views on the Israel/Palestine situation. It first searches twitter for what Elon thinks.
[LLM][Safety][Deployment]
“DeepSeek Summary: Demonstrates how Grok's response is influenced by Elon Musk's tweets on a sensitive geopolitical topic.”
X
Here's what I would prefer to see:
[LLM]
“DeepSeek Summary: Expresses a preference for something, possibly related to AI or tech policy.”
X
reading 'AI News' (previously Smol Talk) is probably the highest-leverage 45 mins
[LLM][Tooling]
“DeepSeek Summary: Soumith recommends 'AI News' as a high-leverage use of 45 minutes.”
X
we've been working on democratizing fast kernel writing on the @PyTorch team. try
[Infra][Deployment]
“DeepSeek Summary: Soumith highlights PyTorch team's efforts to make fast kernel writing more accessible.”
X
Sometimes we forget that NVIDIA wins because it's a software company.
[Infra]
“DeepSeek Summary: Soumith argues NVIDIA's success is due to its software, not just hardware.”
X
I think it's clear that for many smaller companies that invested in deep learning, it turned out
[Evaluation]
“DeepSeek Summary: Smaller companies that invested in deep learning faced challenges.”
X
Very excited to share @theworldlabs 's latest research work RTFM!! It's a real-time, ...
[Multi-modal]
“DeepSeek Summary: Fei-Fei Li announces World Labs' RTFM research, focusing on real-time spatial intelligence.”
X
HuggingFace CEO Clem Delangue said we're in an “LLM bubble“ that might burst next year, arguing the industry's obsessed with building one massive model when...
[LLM]
“DeepSeek Summary: Delangue warns that the current AI hype is specifically an LLM bubble, not a broader AI bubble, and it may burst soon.”
X
Clem Delangue, CEO of Hugging Face, is making open-source AI practical, scalable, and ethical. See why he's one of BigDATAwire's 2026 People
[Infra][Safety]
“DeepSeek Summary: Delangue is recognized for advancing open-source AI with a focus on practicality, scalability, and ethics.”
X
congrats to OpenAI on winning the Turing Test
[LLM][Evaluation]
“DeepSeek Summary: Max Woolf sarcastically congratulates OpenAI on passing the Turing Test, likely referencing ChatGPT's launch.”
X
Some personal news: I recently joined Cursor. Cursor is a small, ambitious team, and they've created
[Tooling]
“DeepSeek Summary: Sasha Rush announced joining Cursor, highlighting the team's ambition and product.”
X
Been reflecting a bit on the Harvard news. This paper from 2017 was ... Didn't realize at the time how lucky for us Americans to work with incredible people from around the world.
[Evaluation]
“DeepSeek Summary: Reflection on Harvard news and gratitude for international collaboration.”
X
today i woke up to a living version of a phd student's nightmare. a new paper in my inbox: a detailed reproduction of a paper i wrote
[Evaluation]
“DeepSeek Summary: Sasha Rush humorously describes receiving a reproduction of his own paper, a common academic anxiety.”
X
Classical Jensen math. Unidirectional bandwidth is topped at 450GB/s, and then there comes a protocol overhead of two digit percentage.
[Infra]
“DeepSeek Summary: Discusses bandwidth limitations and protocol overhead in GPU interconnects.”
X
Hear, hear, I'm excited to introduce a new performance metric: Maximum Achievable Matmul
[Evaluation][Infra]
“DeepSeek Summary: Introduces a new performance metric for matrix multiplication.”
X
Thanks to an awesome contribution from @omarnomad The Machine Learning Engineering Open book now can
[Tooling]
“DeepSeek Summary: Announces a contribution to the Machine Learning Engineering Open Book.”
X
If you're trying out FA4, you're likely to run into not being able to load cutlass.cute
[Infra][Tooling]
“DeepSeek Summary: Warns about a common issue with FA4 and cutlass.cute loading.”
X
After working on releasing the v5, this is the latest release from the Transformers team at
[Infra][Deployment]
“DeepSeek Summary: Sayak announces a new release from the Transformers team after working on v5.”
X
Thanks @AnthropicAI. Thanks @huggingface for letting me work on Diffusers and other open-source projects across the fleet.
[Tooling][Deployment]
“DeepSeek Summary: Sayak expresses gratitude for opportunities to work on Diffusers and other open-source projects.”
X
My presentation at the @PyTorch Conf EU is now live. It's an exciting piece given its
[Evaluation][Deployment]
“DeepSeek Summary: Sayak shares that his PyTorch Conference EU presentation is now available.”
X
Ethan Mollick (@emollick). 124 likes 7 replies.
[LLM]
“DeepSeek Summary: Tweet with 124 likes and 7 replies, but content not fully captured.”
X
On the plus side with Opus 4.7, if it does decide to think it produces BY FAR the best
[LLM]
“DeepSeek Summary: Mollick praises Opus 4.7's thinking capability for producing the best results.”
X
From a pure 'do good for the world' mission perspective, having the acting like a solid personalized tutor is one of the better uses of AI.
[LLM]
“DeepSeek Summary: Mollick argues that AI as a personalized tutor is a top use case for societal good.”
X
New preprint! Everyone loves causal interp. It's coherently defined! It makes testable predictions
[Evaluation][Safety]
“DeepSeek Summary: Announces a new preprint on causal interpretation, emphasizing its coherent definition and testable predictions.”
X
For the first time in almost a decade, I'm teaching a class on learning and control.
[Evaluation]
“DeepSeek Summary: Ben Recht is teaching a class on learning and control after a long hiatus.”
X
Everyone knows actions are fundamentally different than predictions, but it's hard to write this
[Evaluation]
“DeepSeek Summary: Distinguishes between actions and predictions in machine learning contexts.”
X
Assembling a reading list for a class on the theory of engineering architecture.
[Evaluation]
“DeepSeek Summary: Ben is curating materials for a course on engineering architecture theory.”
-- END OF LOG --
[STATS] 49 items · Filter applied
Powered by Horizon + DeepSeek