Harbor Framework

@harborframework

San Francisco, CA

harborframework.com

Joined January 2026

Following

2,229

Followers

Replies Reposts Media

Harbor Framework
@harborframework
Sep 21
Jev-as-a-Judge in Harbor Rewardkit!
Benedikt Stroebl
@benediktstroebl
Sep 21
Jev-as-a-Judge in Harbor Rewardkit is pretty amazing. It agrees with Fable and GPT and is very fast. When we test it on a real eval that uses an LLM judge, such as the @harvey LAB benchmark, Jev achieves perfect agreement but is 50x faster.
1
Harbor Framework
@harborframework
Sep 20
𝚑𝚊𝚛𝚋𝚘𝚛 𝚛𝚞𝚗 -𝚍 𝚑𝚊𝚗𝚍𝚜𝚑𝚊𝚔𝚎-𝚊𝚒/𝚊𝚝𝚕𝚊𝚜-𝚏𝚒𝚗𝚊𝚗𝚌𝚎
Jonas Mueller
@jomulr
Sep 15
Replying to @jomulr
In my view, ATLAS Finance is the closest benchmark yet to real end-to-end professional financial work. Code to run the benchmark as a @harborframework task suite: github.com/Handshake-AI-R… Data/environments: huggingface.co/datasets/hands…
Harbor Framework
@harborframework
Sep 16
docs drop!
Alex Shaw
@alexgshaw
Sep 16
New Harbor docs! We've released many exciting features that were previously undocumented like simulating users, streaming, and regrading trials. Check them out or give your coding agent the MCP. docs.harborframework.com
Harbor Framework
@harborframework
Sep 12
𝚑𝚊𝚛𝚋𝚘𝚛 𝚛𝚞𝚗 -𝚍 𝚋𝚎𝚜𝚙𝚘𝚔𝚎𝚕𝚊𝚋𝚜/𝚊𝚞𝚝𝚘𝚛𝚎𝚜𝚎𝚊𝚛𝚌𝚑-𝚎𝚡𝚊𝚖
Alex Dimakis
@AlexGDimakis
Sep 9
We are releasing AutoResearchExam, a benchmark on open-ended machine learning and engineering tasks. Our benchmark covers seven research areas including model training, data curation, AI safety and interpretability. benchmarks.bespokelabs.ai/autoresearchex… In each task, we give agents 24 hours
00:00
3
Harbor Framework
@harborframework
Sep 1
SkyRL + Harbor RL recipe!
Charlie Ruan
@charlie_ruan
Sep 1
Worked with Mercor Research and the SkyRL team, training Qwen3.5-397B-A17B on APEX-Agents (long-horizon office work) off-the-shelf data with SkyRL, improving Pass@1 from 16% to 27%. The post is more of a practical field guide for what to do given an RL dataset, de-risking step