Archive
GitHub 熱門專案、開源工具、框架、runtime 與重要 release。
代理工具
Agent Behavioral Contracts II analyzed 18,000 preregistered task runs and found highly correlated failures between two agents using the same model. The study instead uses linear programming over joint distributions to calculate finite-sample lower bounds, though the current evidence remains limited to two-agent handoff settings.
最新模型
DreamX-Phi injects the rigid-body transformations of two robotic arms directly into attention and constrains video generation using depth, object masks, and V-JEPA features. It scored 60.65 on a WorldArena 2.0 snapshot, but its weights and inference code will not be released until the challenge concludes, and generalization to real robots remains unverified.
GitHub Repo
The open-source MARC v1 uses YAML to define extraction, reasoning, and output agents, with support for the Gemini API or local deployment through Ollama. It improves error localization and data control, but presents only three types of use cases and does not demonstrate that a multi-agent approach is more accurate than a single prompt.
最新模型
A new study finds that sparse autoencoder ablation evaluations typically allow each dictionary to select its own highest-activation token, meaning ostensibly identical latents are tested at different positions. After fixing a shared position, the 7.6% and 11.9% of variance previously attributed to dictionary differences in controlled experiments nearly disappeared.
代理工具
VALG breaks machine learning theory research into problem formalization, proof dependency graphs, local derivations, and multi-perspective review, then determines whether to repair a proof or open a separate relaxed branch based on the level of failure. Across nine COLT 2026 open subproblems, the system produced 22 internally finalized candidates, but only two runs yielded results within the scope of the original problems.
最新模型
Researchers filtered 88 billion tokens of K–5 material from FineWeb‑Edu and trained 0.6B, 1.3B, and 5B models from scratch alongside compute-matched control models. Experiments show that scaling the models, applying SFT＋GRPO, and providing in-context examples primarily elicit existing capabilities rather than clearly pushing beyond the pretraining knowledge boundary.
AI 安全
Synthetic Persona Pretraining embeds first-person reflections generated from a “values constitution” into roughly 10% of pretraining documents, then uses post-training to bind that persona to the assistant identity. Newly released full experiments, weights, and data show that early intervention mainly improves value generalization, while resisting conventional jailbreaks may not require training from scratch.
最新模型
PlayWorld uses an agent that observes generated frames and dynamically adjusts control commands, preventing differences in movement magnitude across world models from making fixed trajectories incomparable. Results from 171 scenarios show that although current models can generate fluid visuals, they still struggle to preserve global spatial structure and states that continue evolving outside the field of view.
AI 基礎設施
Instead of replicating an entire model, OpScale scales only the operators currently creating bottlenecks—such as attention and MLP—based on traffic and sequence length. The authors replayed 929,000 requests on A100 and GB200 clusters, using up to 36.3% fewer GPUs while meeting the same latency targets, but the prototype code has not yet been released.
最新模型
Using 600 verifiable Go life-and-death problems, TsuGO measures whether models allocate reasoning resources to the correct branches rather than merely counting correct answers and tokens. After four multiple-choice candidates are removed, both accuracy and search efficiency fall sharply across several reasoning models, leaving a clear gap with the neural-network-guided KataGo.
最新模型
LongEarth-Bench organizes 117,000 remote-sensing images into approximately 120,000 long-sequence question-answer pairs, requiring models to track changes in landforms, anomalies, and missing states. LongEarth-R1, built on Qwen2.5-VL-7B, adds frame numbering, structured reasoning, and spatiotemporal rewards, but its data, weights, and code have not yet been released.
代理工具
AutoDesign uses an outer agent to analyze execution traces and modify the production system, with an independent development set determining whether updates are accepted. The open-source DesignHarness outperformed several commercial systems on the authors’ PosterBench, but automated scores had a correlation coefficient of only 0.34 with human preferences.