Archive
GitHub 熱門專案、開源工具、框架、runtime 與重要 release。
最新模型
The new framework lets a coding agent maintain rules and long-term consequences, then compiles the state into depth and semantic proxy videos that constrain MiniMax-H3’s visual generation. The prototype demonstrates five sets of controlled scenarios but provides no quantitative benchmarks, ablation studies, or real-time generation results.
最新模型
ByteDance Seed’s method avoids applying a terminal reward directly to every denoising step. Instead, it uses reward gradients to construct positive and negative clean-image prediction targets. It achieved the best result in 19 of 20 paired tests across two backbones, although the findings still rely primarily on automated reward models and measurements conducted by the authors.
推論系統
The release also adds Decode Context Parallel, fused FlashKDA kernels, and adaptive speculative token budgets, while extending DeepSeek V4’s sparse MLA across standard and speculative decoding. Some performance figures cover only individual kernels or specific DSpark configurations, and upgrades must account for breaking changes to defaults and plugin interfaces.
代理工具
MLCommons has introduced an end-to-end RAG inference benchmark that separately measures vector database construction and multi-hop question-answering throughput. To control uncertainty in generation, performance tests replay prerecorded inputs for each stage, so they do not fully represent the dynamic behavior of online RAG systems.
最新模型
GlanceWAM generates a latent representation of future frames in the background roughly every three seconds, allowing it to be reused across multiple 0.8-second action chunks. The paper reports a 72.2% success rate on RoboCasa, but the evidence currently remains limited to simulation and the authors’ own measurements.
代理工具
Alibaba’s MAI team tests central planning agents for mobile devices using mutable application databases, permission errors, and tool dependencies, moving beyond static function-call matching. The best model is stronger at direct tool use, but still lags significantly in memory application and cross-agent collaboration.
AI 安全
Prime Intellect has released a full technical report explaining how it preserves context, subagents, and a modifiable Harness in separate layers while calculating costs across the entire agent tree. The experiments also reveal the downside of persistent learning: the agent once saved a specification exploit as a reusable Skill.
AI 安全
The new benchmark packages complete vulnerable programs, fuzzing interfaces, and sanitizers into offline Docker images, scoring models by distinct, reproducible crash signatures. Its design gives credit for defects that models discover unexpectedly, but the current difficulty coefficients are derived from the same models being evaluated, leaving the rankings circular.
代理工具
NVIDIA has added agent-driven workflows to COMPASS, enabling coding agents to help validate environments, prepare scenes, train residual policies, and compare checkpoints. The agent handles development orchestration only; actual navigation is still performed by the exported vision policy and ROS 2 controllers.
代理工具
A Kyoto University team created MemUse from a four-month real-world deployment, separately evaluating whether a model can retrieve a memory and whether it uses that memory in an appropriate response. The results show that expanding memory capacity substantially improves direct QA, but does not produce corresponding gains in natural integration or overall user satisfaction.
AI 安全
A team including the Shanghai AI Laboratory has released StepGuard, a 4B-parameter model that uses paired safe/dangerous trajectories and Balance-GRPO to calibrate overblocking and missed detections. It substantially reduces attack success rates in two agent environments, but its training corpus and data generator have not yet been released, and it still misses some policy violations.
推論系統
SemiAnalysis has released AgentX, featuring long-context, multi-turn, and concurrent subagent workloads so inference systems are no longer compared solely using fixed-length prompts. The dataset preserves request timing and KV-cache reuse patterns, but its sources are concentrated on Claude Code, and it measures serving-system efficiency rather than programming-task correctness.