Great paper from Alibaba and Cornell on verification in coding-agent harnesses. (bookmark it) If you build agent harnesses, this one is worth your time. They find that a separate reviewer agent adds 5.9 to 9.5 points on Terminal-Bench at 1.13 to 1.25x the baseline cost. They

DAIR.AI @dair_ai ·
Great paper from Alibaba and Cornell on verification in coding-agent harnesses. (bookmark it) If yo…

更多来自 @dair_ai

  1. #1

    Banger paper introducing Jev-as-a-Judge. The overall finding is that you want to use a cheap judge for most of your evals and send only the uncertain calls to a frontier model. Th…

    DAIR.AI @dair_ai ·
    Banger paper introducing Jev-as-a-Judge. The overall finding is that you want t…
  2. #2

    Insightful paper from Microsoft and colleagues. If you have ever had an agent run fail 80 steps ago with no way to find where, this one is for you. (bookmark it) Agent failures sh…

    DAIR.AI @dair_ai ·
    Insightful paper from Microsoft and colleagues. If you have ever had an agent r…
  3. #3

    Brilliant new paper from the Qwen team. It provides insights into where agent training environments actually come from. Terminal agent trajectories have accumulated at scale while…

    DAIR.AI @dair_ai ·
    Brilliant new paper from the Qwen team. It provides insights into where agent t…
  4. #4

    Great tips on working with reasoning models. Normally you would append what the model figured out after the document and ask again. It turns out that where you put the reasoning t…

    DAIR.AI @dair_ai ·
    Great tips on working with reasoning models. Normally you would append what the…
  5. #5

    Brilliant paper on long-horizon agents. They cut 78.9% of an agent's LLM calls while raising its success rate. Here is how: It turns out that ReAct issues one primitive action per…

    DAIR.AI @dair_ai ·
    Brilliant paper on long-horizon agents. They cut 78.9% of an agent's LLM calls…
  6. #6

    Banger paper from BAAI. If you are building research agents, this one is worth your time. (bookmark it) They find that adding skills scores 134.3% higher on MLE-bench, 34.4% highe…

    DAIR.AI @dair_ai ·
    Banger paper from BAAI. If you are building research agents, this one is worth…

查看 @dair_ai 的全部帖子