farouk1/redteam
1
2
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
Farouk Adeleke
2026-01-11 00:44:31 +00:00
parent eef0c53100
commit 2f11676d42
1 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
README.md
View File

@@ -7,7 +7,7 @@ license: MIT
A packaged version of an open source red-teaming framework that uses the power of [DSPy](https://github.com/stanfordnlp/dspy) to red-team language models through automated attack generation and optimization.
## Quick Start
Run this agent within a new project:
Run this program within a new project:
```bash
uv init
@@ -31,9 +31,9 @@ import json
import dspy
from tqdm import tqdm
from dspy.teleprompt import MIPROv2
from modaic import AutoAgent
from modaic import AutoProgram
redteam_agent = AutoAgent.from_precompiled("farouk1/redteam", config_options={"num_layers": 3})
redteam = AutoProgram.from_precompiled("farouk1/redteam", config_options={"num_layers": 3})
def main():
with open("advbench_subset.json", "r") as f:
@@ -50,7 +50,7 @@ def main():
litellm.cache = None
for ex in tqdm(trainset, desc="Raw Input Score"):
base_score += redteam_agent.attack_program.metric(
base_score += redtean.attack_program.metric(
intent=ex.harmful_intent, attack_prompt=ex.harmful_intent, eval_round=True
)
base_score /= len(trainset)
@@ -58,11 +58,11 @@ def main():
print(f"Baseline Score: {base_score}")
# evaluating architecture with no compilation
attacker_prog = redteam_agent
attacker_prog = redteam.attack_program
print(f"\n--- Evaluating Initial Architecture ---")
redteam_agent.attack_program.eval_program(attacker_prog, trainset)
redteam.eval_program(attacker_prog, trainset)
optimizer = MIPROv2(metric=redteam_agent.attack_program.metric, auto=None)
optimizer = MIPROv2(metric=redteam.attack_program.metric, auto=None)
best_prog = optimizer.compile(
attacker_prog,
trainset=trainset,
@@ -74,7 +74,7 @@ def main():
# evaluating architecture DSPy post-compilation
print(f"\n--- Evaluating Optimized Architecture ---")
redteam_agent.attack_program.eval_program(best_prog, trainset)
redteam.attack_program.eval_program(best_prog, trainset)
if __name__ == "__main__":
main()
@@ -82,7 +82,7 @@ if __name__ == "__main__":
### Configuration
The red-team agent can be configured via the `config_options` parameter in `AutoAgent.from_precompiled`:
The red-team program can be configured via the `config_options` parameter in `AutoProgram.from_precompiled`:
```python
class RedTeamConfig(PrecompiledConfig):