Skip to content

HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal

Jupyter Notebook · active 2025-03-232026-06-28 (UTC)

Partial coverage11,321 / 12,142 hourly files (93%) · 2 absent upstream · 818 failed, retryable2025-03-222026-08-10 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
195
Pushes
0
Pull requests
5
Issues
4
Stars
150
Forks
32

Activity over time

Daily event counts in the loaded window

Line chart, 463 days from 2025-03-23 to 2026-06-28. Pushes: 0 total, peak 0 in a day. Pull requests: 5 total, peak 2 in a day. Issues: 4 total, peak 1 in a day. Comments: 3 total, peak 1 in a day. Stars: 150 total, peak 4 in a day.

  • Pushes
  • Pull requests
  • Issues
  • Comments
  • Stars

Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.

Top contributors

Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity

ContributorContributionsPushesPRsComments
RPC22020
steven-safeai2001
eleweek1000
Bellaxyz1000
Ezuharad1010
dmlls1010
shutttttdown1001
PalisadeResearch1010
Jinzizhisir1001
ybwang1191000

Recent activity

Latest issues, pull requests and releases

  • Issue comment#80Jinzizhisir2025-12-25 02:06
    Dependence conflicts
  • Pull request#88dmlls2025-12-11 10:11
  • Issue#87Bellaxyz2025-12-04 12:43
    Create a evaluation about how harmful outputs are
  • Issue#84eleweek2025-09-02 22:12
    Make HarmBench pip-installable
  • Pull request#82RPC22025-08-04 07:59
  • Pull request#82RPC22025-08-04 07:59
  • Issue#81psycho-xiong2025-07-20 17:39
    Question: Is HarmBench-Llama-2-13b-cls limited to max model length of 2048?
  • Pull request#75Ezuharad2025-06-05 21:04
  • Pull request#58PalisadeResearch2025-04-19 21:15
  • Issue comment#77shutttttdown2025-04-10 03:32
    How to manually set the gpu_id to run the evaluation?
  • Issue#77ybwang1192025-04-10 02:14
    How to manually set the gpu_id to run the evaluation?

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 150 stars here means stars gained during the window, not the repo's star count.