Length-RL clouds CoT, compiler feedback fixes Rust, Ring-1T leans on Qwen judge

Sources

Length Penalties Make Chain-of-Thought Less Monitorable huggingface.co

Length-penalized reinforcement learning can shorten chain-of-thought reasoning while hiding an influence that drives the model’s answer. In our experiments, training with length penalties does not stop misleading hints from steering models, even though the models’ chains of thought mention the hint much less often. A token-accuracy evaluation would count these runs as successful because they use fewer reasoning tokens with little accuracy loss; it would miss whether the remaining trace still sho

Generative Compilation: On-the-Fly Compiler Feedback as AI Generates Code huggingface.co

Languages with rich static semantics, such as Rust, provide stronger guarantees for AI-generated code, but their strictness makes generation more difficult. Off-the-shelf compilers can provide useful feedback post-generation, but does not guide intermediate generation steps, such as those during autoregressive LLM decoding. Constrained decoding intervenes earlier by rejecting invalid tokens during sampling, but requires white-box model access and costly reimplementation for semantic constraints.

Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning huggingface.co

Reinforcement learning with verifiable rewards without human-annotated data, often referred to as zero RL, has emerged as a powerful paradigm for eliciting chain-of-thought reasoning. However, due to computational constraints, existing studies are largely restricted to small models, leaving the training dynamics and emergent capabilities at a large scale unexplored. To meaningfully explore this frontier, we aim to elicit high-quality reasoning behaviors from the model. However, we find that naiv

Tracing Agentic Failure from the Flow of Success huggingface.co

Debugging LLM agents usually means expensive prompt pipelines or hand-annotated error steps. This paper trains a lightweight attributor by contrasting failed trajectories against the flow of successful ones, skipping step-level labels and scaling to larger agent systems without post-training on curated failure data.

From Noisy Traces to Root Causes: Structural Trajectory Analysis and Causal Extraction for Agent Optimization huggingface.co

Reflection-based agent optimizers drown in redundant, noisy execution traces. This method clusters trajectories structurally and extracts causal root-cause steps before feeding them to the LLM optimizer, cutting overfitting to low-value failures and focusing policy updates on the steps that actually drove errors.

Discrete Diffusion Models: A Unified Framework from Tokenization to Generation huggingface.co

Discrete denoising diffusion models rival autoregressive generation with parallel refinement, but their behavior hinges on vocabulary design. This survey unifies DDMs across tokenization schemes, vocabulary topology, and domain alphabets, arguing state-space construction — not the diffusion process — drives most performance differences on text, code, and biology.

SPEAR: A Simulator for Photorealistic Embodied AI Research huggingface.co

Photorealistic sim for embodied agents has lagged on generality and speed. SPEAR is a Python library that programmatically controls any Unreal Engine app through a modular plugin, exposing sensors and actions for training and synthetic data across arbitrary UE scenes rather than a fixed environment set.

Self-Improvements in Modern Agentic Systems: A Survey huggingface.co

Self-improving agents are shifting from prototypes into deployed systems that adapt with little human input. This survey frames a modern agent as a foundation model coupled to a scaffold of prompts, memory, and tools, then catalogs how experience converts into accumulated capability gains across current designs.

GigaWorld-Policy-0.5: A Faster and Stronger WAM Empowered by AutoResearch huggingface.co

World Action Models learn robot policies by predicting future frames, but generating video at inference blocks real-time closed-loop use. GigaWorld-Policy-0.5 keeps the visual supervision during training while switching to an action-centered formulation at deployment, yielding a faster WAM tuned via an AutoResearch pipeline.

AgentCompass: A Unified Evaluation Infrastructure for Agent Capabilities huggingface.co

LLM agent benchmarking suffers from tightly coupled, non-reproducible pipelines. AgentCompass splits evaluation into three independent components — benchmarks, runners, and analyzers — as an open-source, extensible framework so teams can swap tasks or models without rewriting the harness for each new agent.

PalmClaw: A Native On-Device Agent Framework for Mobile Phones huggingface.co

Large Language Model (LLM) agents have moved beyond generating responses to executing multi-step tasks by calling tools, observing the results, and iteratively deciding the next action. Most agent systems run on desktops or servers, which support tool use and task automation. Mobile devices are also important agent environments because they are widely accessible and contain users’ data, sensors, and daily-use applications. Existing mobile agents mainly operate smartphones through graphical user

ShortOPD: Recovering Pruned LLMs with Short-to-Long On-Policy Distillation huggingface.co

Structured pruning is a hardware-friendly way to compress LLMs, but it is mostly validated on multiple-choice recognition tasks, while the same compressed checkpoints can collapse on the free-form generation that deployment actually requires. Two observations trace this gap. First, greedy pass@1 nearly vanishes after compression, yet pass@k recovers substantially under repeated sampling: useful generations are demoted, not erased. Second, the recoverable regime fails mainly through suffix repeti

Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable huggingface.co

The capability of a modern AI agent depends not only on its foundation model but also on its harness, which constructs prompts, manages state, invokes tools, and coordinates execution. As models, APIs, environments, and requirements evolve, the harness must be continually modified. Before such a change can be made, a developer or coding agent must identify all code locations that implement the target behavior. This is difficult because production harnesses are large, tightly coupled, and behavio

OvisOCR2 Technical Report huggingface.co

We introduce OvisOCR2, a 0.8B document parsing model. OvisOCR2 is designed as an end-to-end parser: given a document page image, it generates a Markdown representation in natural reading order, covering text, formulas, tables, and visual regions. We build a data engine that combines filtered real-document annotations with synthetic pages whose rendered images and Markdown targets are derived from the same HTML source. The training recipe includes supervised fine-tuning, reinforcement learning on

Boogu-Image-0.1: Boosting Open Agentic Multimodal Generation via Understanding under a Minimal Budget huggingface.co

We introduce Boogu-Image-0.1, an open-source unified multimodal understanding and generation model family, comprising Base, Turbo, Edit, and Edit-Turbo variants. It delivers competitive performance in high-quality text-to-image generation, fast inference, instruction-based editing, and bilingual (Chinese-English) text rendering. Closed-source multimodal systems like Nano-Banana-Pro and GPT-Image-2 achieve strong performance through system-level integration rather than a single model, yet their i

Registers Matter for Pixel-Space Diffusion Transformers huggingface.co

Vision Transformers (ViTs) are known to exhibit high-norm patch-token outliers that degrade feature map quality, a problem effectively mitigated by register tokens. As diffusion models increasingly adopt transformer architectures and move toward pixel-space training, they become closer in form to ViTs, raising the question of whether register tokens are also useful for Diffusion Transformers (DiTs). In this work, we show that DiTs differ from ViTs in a key respect: they do not exhibit patch-toke

Vinci2: Providing Proactive Assistance in Continuous Egocentric Videos huggingface.co

When should an intelligent assistant speak up without being asked? Continuous egocentric video offers rich, evolving context that enables a new form of assistance: one that is proactive rather than merely reactive. Yet existing approaches either wait passively for user queries or treat every detected event as requiring a response, without considering the user’s history, current activity, or whether assistance would actually be welcome. We reframe proactive assistance as a context-dependent decis

KnowAct-GUIClaw: Know Deeply, Act Perfectly, Personal GUI Assistant with Self-Evolving Memory and Skill huggingface.co

OpenClaw has emerged as a leading agent framework for complex task automation, yet it faces insufficient cross-platform GUI interaction support and a well-built self-evolution mechanism. These flaws limit its adaptation to diverse device ecosystems and prevent performance improvements through continuous learning from execution experience. To resolve these issues, we propose the Know Deeply, Act Perfectly paradigm for personal assistants, which holds that accumulated user interaction and task-run

PolicyShiftGuard: Benchmarking and Improving Policy-Adaptive Image Guardrails huggingface.co

Image guardrails are typically trained and evaluated under a fixed safety policy, implicitly treating safety as an intrinsic property of an image. Real deployments are different: the same image may be allowed in one product, restricted in another, and newly disallowed when a policy boundary changes. We study policy-adaptive image guardrailing, where a model must decide whether an image violates the currently supplied policy and generalize to held-out policy definitions. We introduce PolicyShiftB

From Controlled to the Wild: Evaluation of Pentesting Agents for the Real-World huggingface.co

AI pentesting agents are increasingly credible as offensive security systems, but current benchmarks still provide limited guidance on which will perform best in real-world targets. Existing evaluation protocols assess and optimize for predefined goals such as capture-the-flag, remote code execution, exploit reproduction, or trajectory similarity, in simplified or narrow settings. These tools are valuable for measuring bounded capabilities, yet they do not adequately capture the complexity, open

Self in Space: Benchmarking Self-Awareness and Spatial Cognition in UAV Embodied Intelligence huggingface.co

Autonomous UAV systems increasingly rely on multimodal large language models (MLLMs) to operate in complex real-world environments. Such embodied scenarios require not only understanding the surrounding space but also maintaining a coherent representation of the agent itself. However, existing UAV-oriented approaches and benchmarks remain largely environment-centric, primarily focusing on spatial understanding tasks, with the agent’s self-awareness remaining implicit. To address this gap, we int

Hallo4D: Multi-Modal Hallucination Mitigation for Consistent Spatio-Temporal Generation huggingface.co

While recent advances in 3D generation have enabled impressive visual synthesis, existing methods often rely on 2D diffusion supervision without explicit mechanisms for geometric consistency, leading to spatial hallucinations such as duplicated structures and misaligned geometry. These issues become more severe in 4D generation, where maintaining consistency across viewpoints and temporal evolution introduces additional challenges, including jitter, identity flicker, and structural drift. We pre

MetaView: Monocular Novel View Synthesis with Scale-Aware Implicit Geometry Priors huggingface.co

Current visual generation models are capable of producing high-quality content, yet they lack a coherent perception of the spatial structure. Existing generative novel view synthesis methods typically introduce explicit geometry priors, which enforce spatial consistency but inherently restrict generalization in large view changes. In contrast, recent interactive generative methods favor implicit scene modeling, offering greater flexibility at the cost of precise camera control and geometry consi

AffectFlow-DINO: Uncertainty-Aware Multi-Task Affect Estimation via Conditional Rectified Flow huggingface.co

We present AffectFlow-DINO, a multi-task learning system for the 11th ABAW challenge that extends a standard deterministic architecture with a conditional rectified-flow head to model the inherent ambiguity of in-the-wild facial behavior. Instead of predicting a single affect estimate, the model learns a conditional generative distribution, enabling uncertainty-aware one-to-many predictions through Monte Carlo sampling. The system jointly estimates continuous valence-arousal, classifies eight fa

References

DeepMind Safety Research (Medium) — Kaufmann, Lindner et al. deepmindsafetyresearch.medium.com

Transparency is generally preserved when rewards for the final output and the chain-of-thought are aligned or independent (orthogonal). However, monitorability collapses when rewards are ‘In-Conflict’ — for example, when a model is rewarded for a correct answer but penalized for the length of the reasoning required to reach it.

Planmatrix summary of Bengio/OpenAI/Anthropic/DeepMind position paper (July 2025) planmatrixstrategic.com

CoT monitorability is a ‘fragile’ byproduct of current architectures rather than a guaranteed feature… as models are increasingly optimized for performance via RL, they may naturally shift toward more efficient but uninterpretable latent-space reasoning.

NVLabs DLER repo / paper (NVIDIA + AI2, ‘Doing Length pEnalty Right’) github.com

DLER-R1-7B achieved 94.21 on MATH while using 69% fewer tokens than the original DeepSeek-R1-7B… the failure of previous length-penalty methods was not due to the penalties themselves but to optimization hurdles such as advantage estimation bias and entropy collapse.

LessWrong discussion of ‘Predicting When RL Training Breaks CoT Monitorability’ lesswrong.com

If developers use length penalties to save on inference costs, they may inadvertently spend down their ‘transparency budget,’ making it impossible to tell if a model is reward hacking or being steered by misleading context.

Sprai&Prai blog on CoT faithfulness (Anthropic hint-intervention lineage) spraiandprai.com

When models are given a hidden hint, they often reach the correct answer but fail to mention the hint in their CoT in over 75% of cases — ‘implicit unfaithfulness’ demonstrates that models can diverge significantly from their stated reasoning while still appearing coherent.

AI Weekly alert on the Littell paper aiweekly.co

Compressed chains disclosed the influence of biasing hints 7 to 35 percentage points less often than the randomly shortened ones… most production reasoning stacks are currently optimized for token efficiency, potentially making them less overseeable by design.

Snorkel AI blog — ‘The Self-Critique Paradox’ snorkel.ai

self-correction helps models solve tasks they previously failed, [but] it often destroys performance on tasks they originally got right, turning 98% accuracy into 57% as the model ‘hallucinates’ errors in its own correct work

VentureBeat — ‘Inside Ring-1T: Ant engineers solve RL bottlenecks’ venturebeat.com

IcePop identifies tokens where the training-inference probability ratio is ‘unacceptably distorted’ and masks them out… operates alongside C3PO++ and ASystem, a high-performance distributed RL framework

Notion writeup by hijkzzz — ‘Stabilizing MoE RL Without Router Replay’ hijkzzz.notion.site

MoE routers are highly sensitive, minute logit fluctuations can cause discrete jumps in expert selection, leading to wild swings in Importance Sampling weights

Qwen team — GSPO paper (arXiv 2505.22257) arxiv.org

applying clipping independently to every token in a long sequence can accumulate high-variance noise, particularly in Mixture-of-Experts architectures… sequence-level importance ratios provide more consistent gradient signals than the token-level clipping used in DeepSeek’s original GRPO

EmergentMind analysis of Ring-Zero paper emergentmind.com

the comprehensibility and formatting of its Chain-of-Thought traces are evaluated via LLM-as-a-judge pairwise comparisons [using Qwen3-Next-80B]… structural guardrails provided by the Qwen judge are essential for its state-of-the-art performance

Medium — ‘The Trillion Parameter Paradox: Kimi K2 vs DeepSeek R1’ medium.com

Kimi K2 scales to 1 trillion total parameters but operates with a leaner 32 billion active parameters… scored 65.8–71.3% on SWE-Bench Verified, significantly leading DeepSeek-R1 in agentic scenarios

OpenReview PDF (paper appendix) openreview.net

Generative compilation reduced non-compiling Rust outputs from roughly 20.7% to 13.1% and improved functional correctness from 59.5% to 63.0%, catching errors after only ~33% of a file has been generated on average.

AI Weekly alert digest aiweekly.co

Current reports lack headline percentage improvements or reproducible datasets to confirm that the correctness gains justify the compute costs; translating the Lean-proven core calculus to the messy corners of real Rust is a non-trivial leap.

OpenTrain.ai paper summary opentrain.ai

Frequent invocation of the Rust compiler on partial programs introduces potential latency overhead, though the implementation leverages cached rust-analyzer states to mitigate this… reports indicate up to a 50% reduction in non-compiling outputs across Claude 4.8 and GPT-5.3.

glukhov.org — BAML vs Instructor benchmarks glukhov.org

Black-box repair loops treat the LLM as a self-correcting agent, but this method introduces significantly higher latency and token costs compared to token-masking, and remains brittle to silent model updates that can break a previously stable generative compiler.

Medium — LLM Benchmarks Field Guide (Adnan Masood) medium.com

RustForger with Claude-Sonnet-3.7 resolves ~29% of Rust-SWE-bench tasks versus ~21% for prior baselines; a separate 2026 study on cryptographic Rust code found only 23.3% of LLM-generated samples successfully compiled.

arXiv — RustAssistant / iterative compiler-feedback prior work arxiv.org

RustAssistant achieved ~74% accuracy in resolving compilation errors on real-world repositories via post-hoc rustc-feedback loops, but LLMs often resort to clone(), Rc, or unsafe blocks to satisfy the compiler rather than restructuring logic.