git
ask
hub
Privacy
Terms
Sign in with GitHub
benchflow-ai/skillsbench-trajectories
↗
Stars ·
6
Language ·
Python
License ·
AGPL-3.0
Simplify and Visualize This Repo
Scan the safety of this repo
Self Host this repo
Check out more work of Developer: benchflow-ai
Ask anything about this repo to start.
Read the summary data file in this repository and generate an HTML table comparing the 18 AI model setups by pass rate across the 87 tasks.
Write a Python script that downloads the detailed trajectories from the linked HuggingFace dataset and cross-references them with the summary file here to spot-check pass/fail counts.
Using the data in this repo, identify which AI models improved the most when given access to curated Skills versus when they did not, and summarize the findings for a non-technical audience.
Build a simple web page that reads the processed summary file from this repository and displays the scores and pass/fail counts in a dashboard format.
Full explanation on explaingit →
📎
Send
By chatting or signing in you agree to the
Terms
and chat-message logging (revocable in
History
).