I'm excited to sample the research frontier showcased at #COLM2026 in SF. I'm hosting our 2nd workshop on Social Sims with LLMs. sites.google.com/view/social-.... This area has blown up since our last workshop and we'll present the progress of our community building efforts alongside the submissions

I'm at #colm2026, Mon-Thurs

Supporting two papers from @ai2.bsky.social days
🐟Olmo Hybrid, 7B hybrid model
🐠Cracks in the Foundation, long context recipes

Hoping to have fun chats w folks about scaling LM data & evals, ☺️ unbiased on pretrain, posttrain, all the 🚂s

If you’re at #COLM2026, checkout our work on cost efficient LLM evaluation! my coworker Michael will be presenting it at poster session 2 tomorrow starting at 4:30.

arxiv.org/abs/2604.01418

In SF 🌉 for #COLM2026!
Presenting 👇 come say hi:

🗓️ Tue 16:30 | Growing Pains📊
Keeping benchmark scores comparable as benchmarks grow

🗓️ Wed 16:30 | Feelings to Metrics 🧪
How people actually vibe-test LLMs, and how to measure it

🗓️ Fri 9:15 | Growing Pains @ AIMS workshop

😄
Excited for #TADA2026 and #COLM2026.

I'm in the Bay Area this week! 🌊

Today: #TADA2026 @ Berkeley
Tues-Thurs: #COLM2026 @ SF
Fri: Giving an invited talk @ Berkeley

This year UMass Amherst CICS will be hiring tenure-track faculty in natural language processing! Job ad to be posted soon. For anyone at #TADA2026 or #COLM2026 this week, I or my colleague Hamed Zamani would love to chat or answer questions about it - just say hi or send me an email to meet!

If you're at #COLM2026 this week in San Francisco, stop by on Wednesday @4:30pm to learn more about our work on benchmark data for hybrid human-AI generated fake news.
colm.cc
preprint: lnkd.in/g-vdFAZf

We're heading to #COLM2026 next week! Four days of workshops, posters, & talks on our latest fully open AI research, from Olmo Hybrid to evals for AI-assisted scientific writing. 🧵

We'll present SWE-chat next week at COLM. Find us on Wednesday, Oct 7, 2026 during the poster session from 4:30 PM – 6:30 PM PDT in Grand Ballroom #111
Paper: arxiv.org/abs/2604.20779

#COLM2026

sadly I won't be at #COLM2026 this year but tinlab students Audrey and Jing will be there so talk to them! A little bit about what they'll be doing:

Was excited to be on this paper which offers a novel way to measure AI creativity focusing on narrative tension/suspense. LLMs are bad at this because they don't plan out stories step by step, they just generate. We built a tool to do this. Excited that my coauthors are presenting this at #COLM2026!

Did you know gifting four koi 🐟 is offensive in Japan (4 = "shi" = death 💀), or that chopsticks 🥢upright in rice is inappropriate in China (like funeral incense 🕯️)? Top VLMs don't!

Our #COLM2026 paper introduces 🌏NormViz to measure visual norm understanding in 16 countries.

Best model: 25.3% 🚨 🧵👇

(5/5) Fractured Chain-of-Thought Reasoning: examines how disruptions in chain-of-thought processes affect reasoning in language models.

arxiv.org/abs/2505.12992

#COLM2026

(4/5) RealUserSim: Bridging the Reality Gap in Agent Benchmarking via Grounded User Simulation: grounds agent benchmarking in realistic user simulation.

arxiv.org/abs/2605.20204

#COLM2026

(3/5) InterviewSim: A Scalable Framework for Interview-Grounded Personality Simulation: introduces a scalable framework for simulating personalities grounded in real interviews.

arxiv.org/abs/2602.20294

#COLM2026

(2/5) MTA-Agent: An Open Recipe for Multimodal Deep Search Agents: presents an open recipe for building multimodal deep search agents.

arxiv.org/abs/2604.06376

#COLM2026

(1/5) We are pleased to announce our participation in COLM 2026, the Third Annual Conference on Language Modeling, at the Hilton Union Square in San Francisco, October 6–9. Our researchers will present 4 accepted papers. Full list below ⬇️ #COLM2026

LLMs interact with environments through textual representations. In my newest work at #COLM2026, we explore how these textual representations can introduce unintended biases in LLM decision-making...(1/4)

Paper: arxiv.org/abs/2608.16707

If you're at #COLM2026, come check out our poster for "Spoiler Alert" (arxiv.org/abs/2604.09854). We ask why LLM fiction is so bad at holding narrative tension, and create a metric to measure tension in short stories. What improves LLM fiction on this metric, it turns out, is narrative planning.