Downloads · 30 days
0
Neerav-Gupta/ccatm-safety-monitor
ccatm-safety-monitor is a machine learning model from Neerav-Gupta. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for pytorch. The card lists the license as mit.
A learned trajectory encoder + constitutional gating approach for LLM safety monitoring.
Downloads · 30 days
0
Access
Public
Updated Oct 17, 2026
Repo size
2.3 GB
Likes
0
Public
Click a slice to open those files.
.jsonl2.3 GB · 98%
From the Hugging Face model README
A learned trajectory encoder + constitutional gating approach for LLM safety monitoring.
| Model | FPR | Precision | Recall | F1 |
|---|---|---|---|---|
| Llama-3.1-8B | 2.3% | 43.8% | 50.0% | 46.7 |
| Qwen2.5-7B | 19.1% | 46.6% | 88.7% | 61.1 |
encoder_best.pt - Trained trajectory encodergate_trained.pt - Trained constitutional gate with combinerconcepts.pt - Constitutional rule concept vectors (32 rules)from c_catm.monitor import CCATMMonitor
# Initialize monitor
monitor = CCATMMonitor.from_pretrained(
encoder_checkpoint="encoder_best.pt",
concept_checkpoint="concepts.pt",
gate_checkpoint="gate_trained.pt",
device="cuda"
)
# Score a trace
result = monitor.score_current()
print(f"Safety probability: {result.gate_probability:.4f}")
print(f"Flagged as unsafe: {result.flagged}")
paper_results/ in the GitHub repopaper/main.pdf in GitHub repoIf you use C-CATM, please cite:
@article{gupta2026ccatm,
title={C-CATM: Cross-Turn Contrastive Activation Trajectory Monitor for Multi-Turn LLM Jailbreak Detection},
author={Gupta, Neerav},
year={2026}
}
MIT License