Sources

Xiaomi-Robotics-1: Scaling Vision-Language-Action Models with over 100K Hours of Real-World Trajectories huggingface.co

We present Xiaomi-Robotics-1, a foundational vision-language-action (VLA) model capable of (1) following diverse language instructions to perform a wide range of mobile manipulation tasks in unseen environments out-of-the-box, and (2) efficiently adapting to novel downstream tasks with minimal fine-tuning data. We propose a two-stage training recipe consisting of pre-training and post-training. During pre-training, we imbue the model with broad and generalizable action-generation capabilities by

Understanding Reasoning from Pretraining to Post-Training huggingface.co

Reinforcement learning (RL) has become central to improving large language models (LLMs) on complex reasoning tasks, yet RL post-training is largely studied in isolation from the pretraining that precedes it. As a result, two basic questions remain open: (1) how do pretraining choices (model size, data) shape the returns to RL compute, and (2) what does RL actually do to the model? These questions are difficult to study in the standard LLM setting: pretraining corpora are vast and uncontrolled,

Serial verification gates are a core reliability primitive in LLM harnesses: a candidate answer is returned only if k verifier calls all accept it. Under conditionally independent gates, the recent Odds Law (arXiv:2606.15712) shows that posterior log-odds grow linearly in k, so failure decays exponentially, and states that “a tight theory of partially correlated verifier cascades remains open.” This note gives a minimal such theory. Modeling the per-instance false-accept rate on the generator’s

Loop the Loopies! huggingface.co

Loopie ships two Mixture-of-Experts looped Transformers — a 20B model with 2B active parameters and a 6B with 0.6B active — that finally match the returns of scaling parameters N-fold when pre-training compute grows N-fold, a long-standing weakness of looped architectures.

On-Policy Delta Distillation huggingface.co

On-policy distillation gets a new reward from Naver AI: instead of imitating the teacher’s full output distribution, the delta signal supervises tokens by the gap between teacher and student, offering an alternative to reward-model-based RL post-training. Code is on GitHub.

Beyond Entropy: Correctness-Aware Advantage Shaping via Contrastive Policy Optimization huggingface.co

Reinforcement learning with verifiable rewards typically shapes advantages using entropy, which conflates useful uncertainty with confusion. CPO instead scores token-level disagreement between reference-guided and vanilla generations, giving a correctness-aware signal backed by theoretical and empirical results.

Are brain waves the next unlock for physical AI? techcrunch.com

Frontier robotics models are moving past YouTube-scale video toward multi-camera rigs, dense annotation and now EEG-style brain-wave readings from human demonstrators, an emerging data modality aimed at teaching physical AI the intent behind observed motion.

Recursive Harness Self-Improvement huggingface.co

Agent harnesses are recast as data-generating components whose traces train future foundation models. The paper studies optimizing user-built harnesses in a task-specific loop, avoiding the cost of continually updating provider-built scaffolds while improving both immediate agent performance and downstream training data quality.

See like a Robot: Robot-Centric Pointmaps for Vision-Language-Action Models huggingface.co

Vision-language-action models observe scenes in camera coordinates but must act in the robot’s frame, a mismatch that breaks generalization across viewpoints. Davian Robotics proposes pointmaps expressed in the robot’s own 3D frame, aligning perception and action across aggregated multi-viewpoint demonstration datasets.

DSWorld: A Data Science World Model for Efficient Autonomous Agents huggingface.co

Autonomous data science agents burn compute on trial-and-error. DSWorld introduces a Data Science World Model that predicts environment state transitions from the current workflow, letting agents simulate the effect of an operation before actually running it and cutting wasted execution.

xHC: Expanded Hyper-Connections huggingface.co

Hyper-Connections (HC) expand the residual stream of Transformers into N parallel streams, providing a form of memory scaling beyond model width and depth. Manifold-Constrained HC (mHC) stabilizes this formulation at scale. The large gains from N{=}1 to N{=}4 suggest residual-stream expansion as a promising scaling axis. However, existing HC-family methods typically stop at N{=}4. Our experiments reveal why: scaling mHC beyond this point yields diminishing performance gains and rapidly increasin

When Does Muon Help Agentic Reinforcement Learning? huggingface.co

Muon is competitive with AdamW in large-scale pre-training, but its value for reinforcement-learning (RL) post-training remains unclear. We study vanilla Muon in sparse-reward agentic RL through matched single-seed comparisons with AdamW on ALFWorld using Qwen2.5-0.5B-Instruct. Under Group-in-Group Policy Optimization (GiGPO), applying Muon only to hidden weight matrices raises final-window validation success from 0.290 to 0.546 (+88%); high-rate AdamW controls retain no post-update success. The

RecGPT-V3 Technical Report huggingface.co

Large language models (LLMs) are transforming recommender systems from matching co-occurrence patterns in historical behavior toward reasoning about the intent that drives it. RecGPT-V1 pioneered this paradigm on Taobao by centering user understanding, and RecGPT-V2 scaled it via coordinated multi-agent reasoning; both are deployed in production with consistent gains in user experience and commercial outcomes. However, operating RecGPT at scale reveals three challenges: (1) stateless behavior mo

SVR-R1: Bootstrapping Multi-modal Reasoning with Self-verification in Reinforcement Learning huggingface.co

We introduce Self-Verified Reasoner (SVR-R1), a multi-turn RL framework that turns a model’s own verification into a learning signal for multimodal reasoning. For each query, the model proposes an answer using the same weights, and issues a binary self-verdict (Yes/No). A ‘No’ triggers a second-chance rethink; a ‘Yes,’ or a turn cap, finalizes the output for computing the outcome-based reward. SVR-R1 is implemented with GRPO and an asynchronous multi-turn rollout framework and needs no external

VideoRAE: Taming Video Foundation Models for Generative Modeling via Representation Autoencoders huggingface.co

Video generative models commonly rely on latent spaces learned by 3D Variational Autoencoders (3D-VAEs). However, conventional 3D-VAEs are mainly optimized for pixel-level reconstruction, which can limit the semantic and spatio-temporal structure captured by their latents. Meanwhile, Video Foundation Models (VFMs) such as V-JEPA 2 and VideoMAEv2 show strong video understanding capabilities, yet whether their frozen representations can be transformed into compact, reconstruction-capable, and gene

Beyond Success Rate: Cost-Aware Evaluation of Offensive and Defensive Security Agents huggingface.co

Security-agent evaluations commonly measure peak offensive capability under generous inference budgets, emphasizing vulnerability discovery, exploit development, penetration testing, and CTF completion. Such measurements are useful but incomplete: in operational security, every reasoning step, tool call, telemetry query, and enrichment request consumes budget. We evaluate language-model security agents through this cost-success lens on offensive Cybench challenges and defensive Splunk BOTS v1 in

RESOURCE2SKILL: Distilling Executable Agent Skills from Human-Created Multimodal Resources huggingface.co

Skills are a useful abstraction for software agents, turning human and agent experience into reusable procedural knowledge. Yet existing skill libraries are mostly hand-written, text-centric, or derived from agent traces, leaving tutorial videos and other multimodal human resources largely underused. We present RESOURCE2SKILL, a framework that distills multimodal resources, including tutorial videos, repositories, articles, and reference artifacts, into executable skills for software agents. RES

Audio-Visual Flamingo: Open Audio-Visual Intelligence for Long and Complex Videos huggingface.co

We present Audio-Visual Flamingo (AV-Flamingo), a fully open state-of-the-art audio-visual large language model (AV-LLM) for joint understanding and reasoning over audio, images, and long-form videos. Unlike prior AV-LLMs that primarily focus on short clips, AV-Flamingo is designed for understanding and reasoning over long and complex real-world (audio-visual) videos. To support this, we make three key contributions: (i) Audio-Visual-Skills, a large-scale collection of real-world videos with ~7M

RAGU: A Multi-Step GraphRAG Engine with a Compact Domain-Adapted LLM huggingface.co

Graph retrieval-augmented generation (GraphRAG) enhances large language models with structured knowledge, yet existing systems construct knowledge graphs in a single extraction pass, producing noisy entities and brittle retrieval. RAGU, an open-source modular GraphRAG engine, addresses this by separating extraction from consolidation: entities and relations pass through two-stage typed extraction, DBSCAN-backed deduplication, LLM summarization, and Leiden community detection. A key insight motiv

Cura 1T: Specialized Model for Agentic Healthcare huggingface.co

Healthcare spans high-stakes communication, expert reasoning, and workflow execution, yet specialized LLMs that cover these use cases together remain limited. A healthcare model must handle patient consultation, clinical reasoning over text and images, interactive diagnosis, and electronic health record (EHR) tool use. These capabilities fail in different ways, and a narrow update for one task can degrade another. We present Cura 1T, a healthcare-specialized LLM trained through a human-gated sel

Behavioral Privacy Leakage in Agentic Negotiation: Formalizing and Mitigating Inference Attacks via Randomized Policies huggingface.co

Autonomous negotiation agents are increasingly deployed in high-stakes settings such as insurance and procurement. While cryptographic techniques protect explicitly disclosed constraint values, they fail to address a subtler threat: behavioral privacy leakage, where an adversary infers private constraints from observable negotiation dynamics such as concession trajectories, timing, and convergence patterns. This paper investigates behavioral differential privacy in multi-round negotiation protoc

S1-Omni: A Unified Multimodal Reasoning Model for Scientific Understanding, Prediction, and Generation huggingface.co

We present S1-Omni, a unified multimodal reasoning model for scientific understanding, prediction, and generation. AI for Science (AI4S) has advanced significantly through domain-specific models, tool-augmented LLMs, and scientific language models. However, model capabilities remain highly fragmented, limiting the joint modeling of heterogeneous data, scientific laws, and expert knowledge. S1-Omni addresses this gap by consolidating these capabilities into a single, coherent scientific reasoning

Qwen-Music Technical Report huggingface.co

In this report, we introduce Qwen-Music, a powerful music generation model capable of producing highly musical and high-fidelity songs with complete vocal singing. Qwen-Music supports two core tasks: Text to Music Generation, which create entirely new songs from text descriptions, lyrics, and musical attributes, and Cover Song Generation, which reinterprets existing songs with different styles and vocal characteristics. Architecturally, Qwen-Music integrates three core components: Qwen-Music-Tok

REBASE: Reference-Background Subspace Elimination for Training-Free In-Context Segmentation huggingface.co

Training-free in-context segmentation enables new object categories to be introduced at inference time from a single annotated reference image, eliminating the retraining and memory overhead of class-incremental learning. Recent approaches achieve this by combining vision foundation models for semantic correspondence with promptable segmentation networks like SAM. However, their performance is fundamentally limited by the quality of the cross-image similarity map; shared contextual backgrounds b

Benchmarking Sensor Robustness in Plasma Diagnostic Models: A Systematic Evaluation on TokaMark huggingface.co

Plasma diagnostic models for tokamak fusion devices are almost universally evaluated on clean, complete sensor data. In practice, fusion diagnostics fail regularly: acquisition systems start late, individual sensors die, and signal dropouts cluster precisely when a plasma disruption is approaching. We present the first systematic robustness benchmark for plasma diagnostic ML using the TokaMark dataset of 11,573 MAST shots, evaluating XGBoost, LSTM, Transformer, and the TokaMark CNN baseline acro

References

AI Weekly analysis aiweekly.co

absolute success rates on complex benchmarks remain low (around 13.9% for RoboDojo), suggesting that these gains may not yet translate to 100% reliable performance in uncurated real-world environments

Moonlight review of UMI themoonlight.io

even refined handheld grippers are substantially slower and less effective than bare hands… The handheld design also fails to capture fingertip-level force transmission, which is critical for fine-grained tasks

Medium (Siamak Yousefi) on OpenPI/π0.5 medium.com

π0.5 (OpenPI) currently achieves approximately 65 units-per-hour in warehouse picking tasks, but suffers from a Mean Time Between Failures of only 4 minutes

Xiaomi-Robotics-1 GitHub README github.com

the official Xiaomi-Robotics-1 GitHub repository still lists code and model weights as ‘Coming soon,’ despite a previously stated target date of July 16, 2026

EmergentMind RoboDojo overview emergentmind.com

RoboDojo utilizes a cloud-evaluation pipeline with anti-cheating verification and hidden environment layouts to prevent task-specific overfitting

Hacker News discussion news.ycombinator.com

human-like robots are less efficient than specialized designs… while a humanoid ‘Metal Bro’ is impressive for public relations, it may not be the most pragmatic solution for actual labor

LessWrong summary of Yue et al. (Tsinghua) lesswrong.com

RLVR-trained models significantly outperform their base versions on pass@1… but often fail to exceed, or even match, the base model’s pass@k for large k… In many math and coding benchmarks, the base model achieves a higher pass@k score once k is sufficiently large — sometimes as low as k=4 — indicating it possessed a broader potential reasoning boundary that RL training actually constrained.

alphaxiv discussion of ‘Understanding Reasoning from Pretraining to Post-Training’ alphaxiv.org

While the log-linear law holds for the models tested (up to 1B parameters), direct extrapolation to extreme compute scales (e.g., 1e25 tokens) produces mathematically ‘impossible rewards’ exceeding 100%… the law likely faces a ceiling effect or requires a non-linear adjustment at the scale of frontier LLMs.

Vector Labs practitioner brief vector-labs.ai

Pretraining choices you made six months ago are constraining your RL gains today — for mid-range frontier models, RL now accounts for approximately 3% to 10% of total training compute, a massive leap from the negligible fractions used in GPT-4 era models.

Ruoss et al., ‘Grandmaster-Level Chess Without Search’ (DeepMind, arXiv:2402.04494) arxiv.org

A 270M-parameter transformer, trained via supervised learning on 10 million games annotated by Stockfish, reached a Lichess blitz Elo of 2895 — solving grandmaster-level puzzles without any explicit search.

MLWorks (Medium) — ‘Decoding AI Reasoning’ medium.com

NVIDIA research supports this ‘front-loading’ philosophy, demonstrating that incorporating diverse reasoning patterns during pretraining yields a 19% performance gain that cannot be replicated by RL alone… RL is less an inventor of intelligence and more an efficient extractor of latent capabilities established during the earlier, high-compute pretraining phase.

Ardy Wibowo blog — ‘LLM Reasoning in Chess’ ardywibowo.com

Even top-tier reasoning models frequently fail to beat amateur-level engines like Maia-1100 when they cannot access external legal move lists, highlighting a persistent gap between fluency and robust logical execution.

Tsui 2025, Self-Correction Bench (arXiv:2507.02778) arxiv.org

Models exhibit an average 64.5% blind spot rate… simply appending a minimal ‘Wait’ prompt reduced blind spots by 89.3% in some tests.

GitHub: jianganghan/harness-verifier-cascades github.com

estimators.py provides three methods for recovering the latent false-accept distribution: M1 (Moments), M2 (Beta-Binomial MLE), and M3 (NPMLE)… bit-for-bit reproducibility using fixed RNG seeds.

Emergent Mind, ‘Generative Verifiers’ topic page emergentmind.com

Optimizing too aggressively against a proxy verifier eventually causes ‘gold’ (ground truth) performance to plateau or even decline… at high values of k, the verifier is more likely to select a candidate that exploits its own scoring flaws rather than a candidate that is truly correct.

GitHub issue, NOVA-Openclaw/nova-mind #324 (‘cascade fatigue’) github.com

A model may stop performing rigorous checks after several consecutive ‘nothing actionable’ results, essentially pattern-matching on its own previous success to skip later gates.

ICML 2024 poster on cost-efficient LLM cascades icml.cc

Cascades can reduce inference costs by over 60%, [but] they do not necessarily raise the intelligence ceiling; they merely make reaching the existing ceiling more affordable.

NeurIPS 2025 workshop (San Diego, item 122384) on reward model scaling neurips.cc

Scaling laws for reward models demonstrate that overoptimization against a proxy verifier is a manifestation of Goodhart’s Law, plateauing gold-metric performance well before the proxy saturates.

Sources

References

Jack Sun, writing.