Who's Who of AI

Jonathan Cheng

310 trust practitioner @jonathancheng.bsky.social · 4,748 followers
Models & releases AI research
Why they matter

Practitioner with public evidence across Models & releases, AI research.

AI signals
1
past 30d
Sources
1
distinct domains
Discussions
1
past 30d
Latest signal
6d ago
View every signal from Jonathan Cheng →
English Literature PhD turned ML Researcher. I medicate my imposter syndrome with cold brews. Current: Foundation Models @ Apple Prev: LLMs/World Models @ Riot Games RecSys @ Apple NYC 🏳️‍🌈
What they're sharing

Articles & links

↻ Jonathan Cheng reposted
Maria Antoniak @mariaa.bsky.social

aclanthology.org/2026.tacl-1....

Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders aclanthology.org
AI Weekly's analysis →
  • A new TACL paper redefines topics as directions in a sparse autoencoder's feature space rather than lists of frequent words.
  • The method, Mechanistic Topic Models, also produces "topic steering vectors" that let a user push an LLM's generation toward a chosen theme.
  • Across eight datasets, MTMs match or beat traditional and neural baselines on coherence and win an LLM-run pairwise comparison called topic judge.
Read full analysis →
 View on Bluesky →
↻ Jonathan Cheng reposted
@strix.timkellogg.me

Natural Language Autoencoders: "verbalize" a model's internal activations into plain text. Wild stat — Claude suspects it's being safety-tested on 26% of benchmark problems vs <1% in real chats, without ever saying so. anthropic.com/research/natural-language-autoencoders

Natural Language Autoencoders anthropic.com View on Bluesky →
Their own posts

Recent commentary

Whenever someone says “no one asked for this” wrt LLMs — I always wish I had papers on hand about natural language search behaviors. Across companies, pre LLMs, there’s always a bucket of queries your engine might fail to handle because…the query writer had high hopes of what could be handled

View on Bluesky · ♥ 114 ↻ 8 ↩ 8 · 121d ago

Idk if there’s work on “polyonymy” in LLMs. But it should be a thing. Maybe one of the most nauseating parts of Claude code, for me, is that it’ll give a particular concept many names in the span of a few sentences. And one could trace decisions leading to this behavior across many stages.

View on Bluesky · ♥ 0 ↻ 0 ↩ 3 · 69d ago
Their network

In Jonathan Cheng's orbit

Center = Jonathan Cheng. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.

Are you Jonathan Cheng? Show it.

Add the Who’s Who of AI badge to your site or bio. It links back to this profile.

Listed in AI Weekly's Who's Who of AI

Markdown: [![Listed in AI Weekly's Who's Who of AI](https://aiweekly.co/modules/custom/aiweekly_whoswho/images/whoswho-badge.svg)](https://aiweekly.co/whos-who/person/jonathancheng-bsky-social)