AI Papers of the Week
Every paper worth reading in AI, hand-picked one week at a time.
Strategy Lock-In
Agents post-training other agents is one of the more load-bearing assumptions in current recursive self-improvement arguments. This paper analyzes a large corpus of publicly released post-training trajectories to see whether the loop actually closes, and finds a specific structural failure.
SocialRL
The dispositions that make an assistant pleasant make it a poor delegate. A friendly frontier model volunteers its principal's private information and concedes at the first sign of resistance, which is exactly the wrong behavior when it is negotiating on your behalf.
Skaling
Standard neural scaling laws assume model size and training data act on loss independently. That assumption bakes in a cross-derivative of exactly zero, and it is why the Chinchilla form drifts at the data-scarce and heavy-overtraining edges of the grid, which is exactly where deployment now happens.
Stealing Reasoning Traces
Frontier providers hide chain-of-thought and hand the client an encrypted block instead, which the client returns with every subsequent request. This work identifies an architectural flaw in that design and turns it into a scalable extraction attack across three providers.
Reason Wide, Not Deep
Reasoning modes beat non-reasoning modes on multi-step agentic tasks and charge a 3x to 6x output-token premium on every single episode. Much of that spend goes into re-deriving procedures the model already worked out on earlier episodes in the same domain, which means the cost is recurring by accident rather than by necessity.
Cracks in the Foundation
You might assume architectural variations within the dense transformer paradigm barely move accuracy, and in the short-context setting you would be right. This work shows four minor decisions, normalization, GQA, pretraining context length, and sliding window attention, each made by at least one of the Olmo, Llama, and Qwen dense families, have a compoundingly negative effect on long-context extensibility. Any one alone is minor, but combining three or more drops downstream long-context performance by up to 47%, and none of it is detectable from short-context loss or validation sets, which is precisely how these choices survive into shipped models. Applying context extension early in pretraining exposes the problem cheaply. After over 170,000 GPU hours the authors release OlmPool, 26 comparable 7B models with checkpoints before and after extension, including several architectures that beat the Llama 3 architecture on long-context extensibility.
Model or Harness
Agent evaluations mostly report system-level outcomes, so a failed run leaves the repair unassigned. The same visible failure might call for model post-training, harness engineering, environment redesign, or benchmark repair, and outcome labels cannot separate those cases.
Harness-R1
Agents accumulate interaction trajectories during deployment and then leave them unused, because their behavior stays fixed. Those trajectories can improve the harness that constructs context, mediates tools, validates actions, and recovers execution, and this work makes that editing a learned capability.
Rehearse
Autoresearch loops propose changes, run full training jobs, and keep whatever improves the metric. Their efficiency depends on judging, before spending a run, whether a proposed modification is likely to work, and this paper studies how that judgment holds up over a trajectory.
ReOPD
On-policy distillation for agentic tasks is expensive because every update needs fresh student rollouts through the environment plus teacher queries at each visited history. Microsoft Research and the University of Amsterdam propose reusing pre-collected teacher trajectories instead.
Invisible Reasoning
Chain-of-thought monitoring rests on the assumption that a model expresses its reasoning in its output tokens. This work demonstrates a concrete failure of that assumption in models shipping today.
Molt
Agentic RL research is constant algorithm modification, and in mainstream frameworks each change threads through layers of trainer, distributed backend, and rollout glue. NVIDIA's Molt is a PyTorch-native training framework built to make that cost small.
ACM
Production agents accumulate context every turn. The usual fix compresses on a token threshold and throws the remainder away, so the trigger fires for reasons unrelated to what the agent is working on. Meta and CMU hand the decision to the agent instead.
Beyond AdamW
Higher-order optimizers have promised faster convergence than AdamW for a while, with computational cost and numerical stability as the standing objections. This NVIDIA work adapts them for large-scale pretraining, identifying instabilities in SOAP at large batch sizes and eliminating the loss spikes with per-step QR orthogonalization and improved preconditioning, then running a unified study of SOAP, Muon, and AdamW under update-RMS matching for fair learning rate transfer. On multi-billion-parameter models trained over trillions of tokens, SOAP and Muon consistently beat AdamW, and at batch sizes up to 100M tokens for next-token prediction they hold stability and quality while AdamW degrades. A layer-wise distributed optimizer compatible with Megatron-LM balances memory and hides communication without approximating the optimizer math.
Structured Output Collapses Diversity
Teams benchmark models in chat, then ship them behind JSON schemas for tools, extraction, and routing. This study of 44 language models shows that the structured surface you deploy is measurably more homogeneous than the chat surface you evaluated on.
RoboTTT
Recent robot foundation models run on single-step or short-history context, a strange way to attempt a five-minute assembly task. RoboTTT, from NVIDIA with Stanford and UT Austin, integrates test-time training into vision-language-action policies to scale visuomotor context to 8K timesteps, three orders of magnitude past prior policies, without growing inference latency. The longer context unlocks one-shot in-context imitation from human video, on-the-fly policy improvement, and robustness to perturbations. It improves overall performance by 87% over a single-step baseline, fully completes a ten-stage assembly task that no baseline finishes, and gains 62% from pretraining with 8K rather than 1K timesteps.
Self-Improving Agents Survey
Self-improving agents are moving from research demos into deployed systems, and this survey gives the trend a clean formalism. It frames a modern agent as a foundation model coupled with an operational scaffold of prompts, memory, tools, and control logic, then treats self-improvement as a self-induced update that commits changes to either the weights or the scaffold.
GFlowRL
Reward-maximizing RL tends to collapse large reasoning models onto a single dominant mode, and GFlowNet-style training is appealing because it matches reward distributions and keeps diverse reasoning paths. GFlowRL scales this to modern post-training by replacing the hard-to-learn partition function with an in-batch Monte Carlo estimate computed from the rollout group the pipeline already produces. It is the first GFlowNet-style RL algorithm to train stably across both dense and sparse architectures, reaching a 2048 Codeforces rating at 14B and outperforming prior methods on math, code, and adversarial red-teaming benchmarks like AdvBench and HarmBench.
LingBot-VLA 2.0
LingBot-VLA 2.0 is an open-source generalist embodied model from Robbyant, trained across 20 robot configurations from single-arm rigs to humanoids like Unitree G1 and Fourier GR-2. It packs 60,000 hours of curated data, 50,000 hours of real-robot trajectories plus 10,000 hours of egocentric human video, into one policy that also predicts future depth and semantic features before it acts. On 9 GM-100 tabletop tasks it beats π0.5 and GR00T N1.7 across two robot platforms and stays ahead on long-horizon mobile tasks, running at about 130 ms on a single RTX 4090D with open-sourced post-training code.
Verification as a Scaling Axis
Verification is emerging as a distinct scaling axis alongside pre-training and test-time compute, and this Stanford, NVIDIA, and UC Berkeley collaboration builds a training-free verifier that reads a continuous, calibrated score straight off the scoring-token logits instead of trusting a discrete pass or fail grade.
Puzzle-75B
Bigger mixture-of-experts models keep winning on quality, but serving them at interactive latency is still hard. NVIDIA compresses the hybrid MoE Nemotron-3-Super into Puzzle-75B-A9B and roughly doubles interactive server throughput while holding quality.
ReContext
Models now support 128K context windows yet still fail to use evidence already sitting in the prompt. ReContext is a training-free inference harness for long-context reasoning that uses model-internal relevance signals to build a query-conditioned evidence pool, then replays it right before final generation while preserving the full original context.
RLVR Meets Human Likeness
RL with verifiable rewards only optimizes what you can objectively score, so style, structure, and diversity quietly collapse and reward hacking creeps in. This MIT work adds an adversarial discriminator trained on human demonstrations as a learned proxy for the human output distribution, and the generator maximizes both task accuracy and that human-likeness signal. Across bug fixing, story generation, and a reward-hacking benchmark, it preserves RLVR's accuracy gains while restoring the fuzzy properties it usually destroys, with misbehavior nearly disappearing.
AutoMem
Memory for LLM agents is usually a fixed module bolted onto the model, but knowing what to encode, when to retrieve, and how to organize notes is itself a skill. AutoMem, from Stanford, treats memory management as a trainable cognitive ability, a capacity cognitive science calls metamemory. ---