The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 31 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 0 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Current multi-turn RL research ignores discarded thinking tokens, breaking the math
“The existing paper people are writing about multi-turn RL are not actually incorporating this, and it kind of, like, breaks all the math.”
Averaging weights of models trained on separate domains works effectively
“You can have a model trained on code, and a model trained on math, and a model trained on Spanish, and you can literally average the weights, and it works.”
Weight updates across specialized tasks are orthogonal enough to merge asynchronously
“The updates made to model weights are orthogonal enough for specialized tasks that this is actually like totally fine. Things are nice and linear in most cases, things are nice and orthogonal, and you can get away with a lot of async updates to models that are…”
Brown: Anthropic treats extended thinking as tool use, not distinct model class
“And it seemed like Anthropik's kind of attitude has been that extended thinking is an instance of tool use and that it's the kind of thing you want to equip the model with the ability to do. But it's not like, oh, it's a thinking model. It's just a sync for th…”
Brown: Anthropic safety issues stem from conflicting model objectives
“A lot of the kind of headline anthropic like safety results, especially related to reward hacking and kind of deviation and alignment faking, Are all things to me that seem like a rock and a hard play situation where the model has two objectives it's given tha…”
Will Brown: Selling to AI Labs Compromises Model Evaluation Integrity
“I think being an eval company puts you in a really hard spot. Some people are talking about this on Twitter, like just that to be an ed-all company, you kind of have to sell to the labs, but selling to the labs doesn't really, like kind of wrecks the revals.”
Will Brown: Academia Will Likely Be the Best Source of AI Evals
“I mean, I do think that like the best source of evals going forward is probably going to be academia.”
Brown: GRPO is more memory efficient and easier to distribute
“GRPO is, like, great for, like, leaning heavy on highly parallel inference compute. It's more memory efficient for the actual training process. It's much easier to do in a distributed fashion because you have less gradient syncing and less model weight copies.”
Brown: RL training is shifting toward model-based LLM judges
“So it, like, feels like people are moving in the direction of model-based rewards, where you, either LLM is a judge where the judge sees the correct answer, or it has questions it's supposed to verify as properties of the response, just because that's much mor…”
OpenAI's o3 acts as a '10-minute AGI' for human tasks
“Whether or not you want to call this, like, end minute AGI, I kind of like the phrase, 10 minute AGI, for just, like, how to think about O-three is that anything that you can do as a human in 10 minutes, O-three is usually going to be able to do reasonably wel…”
Programmatic tools beat end-to-end image generation for multimodal reasoning
“Where I think for a while some people were, like, speculating, like, oh, what if you have the model, like, generate images in its chain of thought reasoning where everything is, like end-to-end multimodal input and output, and it seems like you don't really ne…”
Group-chat multi-agent systems like CrewAI do not work well
“I think people were, like, excited about multi-agent systems for a while, like, the crew AI sort of thing of, like, oh, I'm gonna put my coder agent, my finance agent in group chat, and, like, a lot of these are just kind of silly. They don't actually work ver…”
Algorithms for practical multi-agent RL do not yet exist
“Multi-agent RL's hard. I did five years of it in grad school. It's like not easy. And to the algorithms don't really even exist for the things you would really want to do.”
No legitimate open-source million-token context models exist at scale
“Scaling to, like, million token contexts is, like, really, really hard. There, I don't think there are real, like, open source replications, open token context scaling, Beyond, like, tiny, like, academic model sizes.”
Reasoning models acting as reward models are key to agent RL
“And the most, one of the most promising ways, I think, towards doing this is having the reward models also be able to answer harder questions by themselves being reasoning models.”
Will Brown: AI reasoning models are merely a stepping stone toward autonomous agents
“The thing that's going to make the next wave of stuff be powerful is just, like, everyone wants better agents. Everyone wants models that can, like, go off and do stuff. And, like, reasoning was kind of, like, a precursor to that a little bit.”
Brown: Claude thinking and non-thinking modes likely use same underlying model
“I mean, I think these models should be the same model, and Anthropic knows what they're doing. Like, it's not that hard to, like, Quen did it in a very kind of, like, simple way, and they kind of talked about how they did it a little bit. But it's not, like, t…”
Brown: Truncating reasoning model thinking mid-sentence still yields good outputs
“So it seems like artificially truncating the thought is actually like fine. Like the model can, even if like it got cut off mid-sentence with an injected like think token, these are smart enough models that they can kind of finish with the best that they got f…”
Brown: Reasoning effort dropdowns will disappear from chat interfaces
“I think in chat interfaces, it probably won't stick around. Like, I don't think we're always going to have the dropdown of like Oath for many and Oath for many high. That feels silly.”
Brown: Base LLMs will do anything up to their intelligence limit
“The base model in general of LLM is not artificially constrained in any way. Like, with the right prompt, it'll do whatever up to its intelligence limit.”
Brown: Claude 3.7 works for quick projects, not large codebases
“I never really got to the point where I found it was helpful for a thing that was like a large existing code base. But if it's like, hey, I want to like cook something up in a few hours for fun. Pretty good at that. But these become messy and they become hard …”
Will Brown: Claude appeals to AI insiders but lacks mainstream breakout
“It feels like people in the AI world, like, love Claude, or have grown type of Claude, but still had a phase where they were using it a ton. But it hasn't really broken out to general people in the way. And it feels like a lot of their marketing that I've seen…”
Brown: Small LLMs default to skipping tool calls without explicit training
“If you set these models up to use tools, They just won't. Like if you say, hey, here's a question. You have access to these tools. Do as many rounds of tool calling as you want, and then submit your answer. They'll just submit their answer because they like ar…”
Brown: Prompting alone cannot reliably force LLMs to use thinking tokens
“If you want models to use thinking tokens, you kind of have to, like, incentivize that. You have to either do a little bit of, like, SFT warmup, or you have to, like, Reward them for doing it. Otherwise, they will not follow it a hundred percent of the time on…”
Brown: Rewarding generic tool calls causes LLMs to game rewards
“If you start rewarding them for, like, tool use, They will use the tool, but they don't really want to, like, have to, they want to, like, be very safe with it... They would like do silly versions of tool use where they aren't actually using the tool to assist…”
No open-source model currently matches OpenAI's general agent capabilities
“There really isn't currently an open source model that behaves in the way that these models do. We have things like R-one, which are great at kind of the single-term math and code reasoning problems. But they are not in the general purpose agents world yet.”
Scaling autonomous task duration is a plausible path to AGI
“And that, if you can crack that scaling direction of, like, pushing the boundary of how long these models can go out and do these things for, that is a plausible path towards things that become marvelous.”
Long-horizon agent RL requires intermediate turn-level verification
“You probably want some kind of intermediate verification where you're not just waiting for the final answer at the end, But you want something like turn-level reward, potentially, where you want to be able to ensure that the model is getting credit for the mov…”
Agent RL training infrastructure must be asynchronous to avoid compute bubbles
“You also ideally want to, whether you're doing this centralized or decentralized, move in the direction of, like, everything being async and overlaps because that is, like, otherwise you have these inefficiency bubbles that, like, pop up all over your compute …”