Thomas Broadley tbroadley
-
METR
- Berkeley
- https://thomasbroadley.com
- in/thomasbroadley
- @htbroadley
Stars
Data visualization for Inspect AI large language model evalutions.
CVE-Bench: A Benchmark for AI Agents’ Ability to Exploit Real-World Web Application Vulnerabilities
vscode extension for filepath completion in shell scripts
A reference implementation for the specification that can create and configure a dev container from a devcontainer.json.
Python client for interacting with Helm 3.
Collection of evals for Inspect AI
eBPF-based Networking, Security, and Observability
A Kubernetes sandbox environment for use with inspect_ai
Package for calling different models with same interface
:octocat: Github action to retrieve all (added, copied, modified, deleted, renamed, type changed, unmerged, unknown) files and directories.
Monaco Editor Copilot is a plugin for the Monaco Editor that integrates OpenAI's GPT-based code completion engine to provide a seamless and intelligent coding experience.
CodeMirror extension to add GPT autocompletion like GitHub's Copilot.
Vivaria is METR's tool for running evaluations and conducting agent elicitation research.
Inspect: A framework for large language model evaluations
GitHub Action to expose GitHub runtime to the workflow
Generate and auto-execute Python scripts in the cli
Lightweight dependency injection container for JavaScript/TypeScript