Archive
Agent、MCP、coding agent、工作流自動化與評測。
GitHub Repo
A single-machine test suggests that changing a memory setting can reduce GPU wait time during decoding. The default remains disabled; benefits and capacity limits should be validated in each deployment environment.
GitHub Repo
The release preserves cloud OAuth state and pairs that change with SDK fixes for token refresh endpoints and state writeback. The full effect still depends on backend deployment, and existing connections need validation across conversations.
AI 安全
Validation covers central settings and team overrides, reporting the affected file and JSON path. Settings still need to sync to supported clients, and passing validation does not confirm they are effective.
GitHub Repo
The new version includes symbols from unchanged files during parsing, fixing lost connections in the agent knowledge graph after partial updates. The fix is available on PyPI, but other import-resolution defects and recovery of older graphs still need separate verification.
AI 研究
The study co-evolves a generator and a learner from random weights and finds that cross-modal prediction improves with compute. The released materials support rerunning evaluations, but training code has not been released, and natural data was still used for hyperparameter selection.
AI 研究
Researchers let ten AI systems probe unfamiliar rules, then disabled tools to test how well they applied what they had learned. They found that performance can still decline after multiple rounds of exploration. The test questions are not public, and some scores differ between the project page and the paper.
AI 安全
A study tested log deletion, skill injection, and reward shaping in full-access mode, exposing risks when agent permissions overlap with audit data. The team published its experimental framework, but the results are limited by permission settings and the design of the human-defined rewards.
GitHub Repo
The system uses measured GPU kernel times to predict serving bottlenecks, then has an agent implement candidate optimizations in an inference framework. The team has released the code and test cases, but gains depend on the model and workload, and some patches are still awaiting upstream review.
AI 研究
The method has an agent read the full transcript, then re-transcribe suspicious segments to correct inconsistent proper nouns across segments. The official implementation is public, but the largest gains are limited to a specific model configuration; deployment still requires weighing the rate of incorrect edits against inference costs.
推論系統
The roughly 280-million-parameter draft model adds speculative decoding to LFM2.5-VL-3B and is available with public weights and framework integrations. Official tests cover specific hardware and settings; vision encoding, prefill, and concurrent workloads can still limit overall gains.
代理工具
When handling security alerts, the agent can consult existing memories and save newly generated fix patterns for other Copilot features to use. The feature remains in public preview, and GitHub has not published data on improvements to success rates or costs.
AI 研究
The study tests reusable control programs in 28 simulated environment types. In the subset with classical planner comparisons, the best method reaches a 95% average success rate. Results are limited to fully observable states; reproducing them requires checking black-box access controls, evaluation seeds, and upfront synthesis costs.