1. BenchmarksScore82

    BRIE: A Living Benchmark for EHR Information Retrieval

    Researchers have developed a scalable framework to automatically generate question-answer pairs from electronic health record (EHR) notes, creating a continuously maintainable evaluation dataset called BRIE. This new benchmark addresses the limitations of manually curated datasets, which are costly to update and quickly become obsolete.

    This information comes from the research paper "A Living Benchmark for Information Retrieval from Electronic Health Records" by Jordan L. Cahoon et al., published on arXiv. Full analysis
  2. AI ToolsScore79

    Accelerating Vision-Language Models with LFM2.5-VL-DSpark

    LiquidAI has introduced LFM2.5-VL-DSpark, a new dataset designed to accelerate the training of vision-language models. The dataset aims to improve the efficiency and performance of these models.

    Source: Hugging Face blog post on LFM2.5-VL-DSpark. Full analysis
  3. Developer ToolsScore79

    Codex CLI 0.157.0 Released

    OpenAI has released version 0.157.0 of its Codex Command Line Interface (CLI). This update focuses on tooling for developers interacting with OpenAI's Codex models.

    Source: OpenAI developers changelog Full analysis