Search: “dair-ai--x”

Nice paper on agent memory. It explores an effective routing mechanism for memory in agents. The overall finding is that you want a separate retrieval strategy for each type of memory instead of one e

Pay close attention to custom harnesses. This is a super interesting paper showing the impact of training inside the harness. Qwen3-14B goes from 22.2% to 54.8% on Spider 2.0-SQLite when it is trained

Recommended paper from Sakana AI on recursive self-improvement. They propose an interesting way to scale recursive self-improvement through multi-agent self-supervision. In this line of research, self

Banger paper from Meta Superintelligence Labs on self-improving agents. (bookmark it) It's hard to know exactly what drives self-improvement, since so many variables are at play (notes, skills, tool c

Banger paper from Meta Superintelligence Labs on user simulators for agent training. Agent RL setups usually let an assistant LLM play the user, so the simulated user is too cooperative and too explic

Build your own harness, folks. Reading papers like this makes me realize how underexplored harness engineering really is. The authors find that on whole-repository migration, GPT-5.6 Sol goes from 6.5

Bullish on this trend of making post-training more accessible. A new post-training era is upon us. If you work on agentic RL, long-context tasks (a big focus today) are expensive, inefficient, and don

Interesting paper from Sakana AI on memory for recurrent models. Recurrent models are good at state tracking, but they usually keep everything in one hidden vector, so short-term computation and long-

Super interesting NVIDIA paper on choosing base models for coding agents. It's actually a clever way to rank base checkpoints by how well each one is likely to do as a coding agent after post-training

Another great paper from NVIDIA. They find that only 0.6% to 1.2% of model weights change after each RL step in the six models they measured. Yet a standard refit copies the full checkpoint to the rol

Another great paper from Salesforce AI Research. The finding is that a 31B open Gemma-4 web agent scores 74.6% on the 9-app WebArena Infinity set, above Gemini 3 Flash with browser use at 70.1%. They

Great paper from Google on program repair agents inside CI. (bookmark it) They propose FlowAgent, which runs a ReAct-style generate-and-validate loop on pre-submit test failures and shows the fix insi

Huge release from @odysseyml. Odyssey-3 Pro sets a new top score on Physics-IQ Verified, a benchmark that asks models to continue videos of real physics experiments. The robotics results stood out to

Agent-to-agent communication with a personal agent on top is wild. Wow. I underestimated how well models like Opus 5.5 can coordinate already. This opens a whole new set of experiences with agents. Bu

Banger paper from NVIDIA. One exciting trend I am seeing is building verifiable environments for your agents and training the harness alongside the model. Co-evolving the harness and the model is a bi

Explore the top AI papers of the week with our brand-new MCP tools. Connect it to Codex, Claude Code, or Grok Bot. Introducing MCP tools for @dair_ai. Connect the MCP to your Codex/Claude/Grok Bot to

Introducing MCP tools for @dair_ai. Connect the MCP to your Codex/Claude/Grok Bot to discover and explore the top AI papers. We carefully curate the papers, so you can expect mostly bangers. The index

Reading eval results is now the slowest part of building agents. I'm Elvis, founder of @dair_ai. I lead research, build, and teach about AI agents. I run harness experiments every night, but reading t

This is a great paper from AWS AI Labs. They show that multi-agent coding teams raised their test pass rate by 28.2% and cut wall time by 16.5% without changing the model. The authors achieve this by

Useful paper on what an agent harness changes when the model stays the same. One takeaway: keep your harness prompt and tools small. In this study, that can cut costs by up to 3x without lowering accu

Banger paper from Meta AI on research agents that decide what to investigate next. (bookmark it) If you run long-horizon research agents, choosing the next investigation is hard to learn, because thos

Context compression is a huge bottleneck for long-running agents. This work finds that context compression hurts long-horizon agents at a few specific points. They propose PAIR, which replays the agen

Great report from Google and MIT. Co-designing ML models and the accelerators is an exciting research direction. This report introduces Coco, an agent platform TPU architects use for hardware and mode

Like Jev, I believe RL will unlock several more like this, slashing the cost of critical agent operations. flow-1 is competitive in performance, but a huge cost-saver for finding failures in agent tra

Really nice paper on harness-aware distillation for small language model agents. (bookmark it) It's a super interesting distillation framework for small language model agents that are deployed with a

Banger paper from Microsoft and colleagues. If you run agents that search a large document collection, this one is worth your time. (bookmark it) They introduce CorpusMap, which resolves recurring ent

Does telling your agent to plan ahead actually help? It's standard practice to add a 'plan ahead' or 'think about the other players' instruction to an agent's prompt. In auctions and matching markets,

Great overview of context compression in LLM Agents And one interesting, unexpected finding. If your compaction policy is tuned to cut tokens, it may be making your agent run slower. Great study from

Huge release from EverMind AI. Raven is an open-source multi-agent system that builds a separate harness for each model and domain, evolves those harnesses from their failures, and coordinates them th

Learn to optimize your own harness, folks. You can squeeze much more performance and value from a harness. This work claims that a self-modified harness matched Codex for 4.03. Specifically, a coding

Own your personal agent, folks. Just built my own little personal agent with @pidotdev's Pi Durable. I wanted to see how quickly I could build an OpenAI dots clone with open-source tools. I am amazed

Very good to see more competent labs training open-weight models. This Reflection's new Beam model seems to have strong reasoning efficiency, which I think is a big deal for long-running agents. 3-4x

50K likes on this!? Most of what Karpthy has shared is things we have been doing for over a year. No surprises, as many replies are highlighting. But I still think this raises an interesting discussio

Banger paper from Google. It's standard practice to sample several agent rollouts and trust the answers they agree on. This Google paper shows that agreement can hide shared errors, while disagreement

Banger paper from NVIDIA on test-time compute for terminal agents. The finding is that you should sample several candidate shell commands, verify them before running one, and spend more on the verifie

Great paper from Microsoft and colleagues on optimizing agent harnesses. Current harness optimizers change how the harness is updated but keep the training scenarios fixed, so feedback keeps coming fr

The AI research community released a set of notable papers between September 28 and October 4, highlighted in a weekly roundup. Key contributions include JAZ, CASD, Agensh, Jev‑Mem, AutoGym, Taste‑Ben

Banger paper from CMU on harness learning. (bookmark it) Also, pay attention to this important new AI engineering skill of improving agents by editing their harness code instead of their weights. Seei

Banger paper from Meta Superintelligence Labs. They find something super interesting and unexpected. (bookmark it) Base models with a light harness often solve more agentic tasks than their RL post-tr

First AutoHarness, then AutoContext, now AutoCompact. I am seeing a rising trend of work that trains models to natively support more of what the harness does. This work specifically trains agents to d

Must-read paper from Meta on reliable auto-research agents. If you let coding agents run ML experiments, one silent bug like leaked eval data or a disconnected gradient can invalidate hours of trainin

Banger paper from Google Research on multi-agent proof discovery. (bookmark it) It's really interesting to see this emerging multi-agent pattern: not enforcing too much execution structure and pairing

Banger paper from NVIDIA on long running agents. A model can accept 128K tokens of context and still make more mistakes the longer it works through a task. If your agent loses its place partway throug

Learn to build a custom harness, folks. Mods in Claude Code, DeepSeek Harness, and now Pi Durable. What do these all have in common? Increased extensibility. Builders need coding agents and harnesses

More agents don't mean higher performance. There is a coordination bottleneck to consider. Not to mention the unnecessary costs. So how many of a multi-agent team's actions actually help it finish the

New paper from Microsoft on compressing agent context at test time. If your agent gets worse as its interaction history grows, this method is worth a look. FOCUS asks which past interactions the agent