I’ll be at COLM in SF from Oct 5–9. It’ll be my first in-person conference in 6.5 years, and my first extended stay in SF in 8 years! I'd love to meet mutuals & folks working on LLM post-training at established labs & neolabs. Please DM or reply here and we'll coordinate 💛

What do LLMs get up to when left to have free-form conversations with copies of themselves or other models? And how can we even quantify what models are doing in free-form conversations? We ran a number of free-form discussions between LLMs and observed the dynamics of the conversation in representation space. There we can quantify a few quite interesting behaviors: 1) When we control for topics, the remaining representation space trajectories for models are quite predictable, so models 'end up' in a similar space across conversations when talking to a copy. 2) More interestingly, if talking to another model, we can observe that the dynamics circle the end points of both models, i.e. the models are being attracted by one another, taking on features and styles from the other. 3) This attraction is highly model-specific. For example Haiku is strongly attracting other models, while barely budging... A few more details and plots (and an arxiv link below):

Excited to share that 𝘾𝙎𝙄𝘾𝙇 is accepted to #COLM2026 🎉 We introduce code-switching in-context learning, an inference-time cross-lingual representation alignment mechanism that gradually bridges non-En languages with En representations. 📄 arxiv.org/abs/2510.05678 (1/🧵)

Can LLMs generate diverse outputs for open-ended questions? Is it helpful if we ensemble outputs from multiple models? We study 18 LLMs on 4 datasets and find that no single model is best at generating diverse outputs 👇/ 🧵

COLM is going to San Francisco for 2026! 🗓️Dates: October 6-9, 2026 🏨Venue: Hilton San Francisco Union Square Website and CFPs for papers and workshops coming up soon!