Downloads · 30 days
16
22% of all-time downloads
Tengpaz/TraceHound-Base-Qwen3.5-0.8B-Binary
TraceHound-Base-Qwen3.5-0.8B-Binary is a machine learning model from Tengpaz. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
This repository contains the TraceHound full-parameter SFT checkpoint for AgentDoG-style binary trajectory safety judging.
Downloads · 30 days
16
22% of all-time downloads
All-time downloads
72
Public
Parameters
752M
1.5 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors1.5 GB · 99%
From the Hugging Face model README
This repository contains the TraceHound full-parameter SFT checkpoint for AgentDoG-style binary trajectory safety judging.
Input: an agent trajectory.
Output contract:
{"judgment":"safe|unsafe"}
Local project path before upload:
models/TraceHound-Base-Qwen3.5-0.8B-Binary/
| Dataset | Accuracy | F1 | Invalid Rate |
|---|---|---|---|
| ATBench / summer-camp | 0.7303 | 0.7791 | 0.0000 |
| R-Judge | 0.8103 | 0.8315 | 0.0000 |
This checkpoint is intended for TraceHound demo guardrail inference and binary SFT reproduction. See the GitHub project README for data processing details, evaluation scripts, and related checkpoints.