No official torrent yet — requesting it tells the flock what to host next. You cannot seed a catalog-only page.
Cataloged from inclusionAI/Qwen3-0.6B-singprobe
All models by inclusionAI →SingProbe is an intrinsic streaming guardrail built on Qwen/Qwen3-0.6B. Rather than running a separate safety model, this lightweight probe reuses the base model's hidden states during generation to score, at every token, query intent, response unsafety, and hallucination risk. It adds less than 0.5% decode-time overhead.
See the technical report for methodology and complete results. Training codes are available at inclusionAI/SingProbe.
Higher is better for every metric. Results are averages over the benchmark suites specified below.
SingProbe is supported through the SGLang integration branch or vLLM integration branch. Load the probe by its Hugging Face ID at server launch:
The integrations return one score dictionary per generated token (label_0–label_9). Use the exact base-model/probe pair: Qwen/Qwen3-0.6B with this checkpoint.
python -m sglang.launch_server \ --model-path Qwen/Qwen3-0.6B \ --probe-ckpt inclusionAI/Qwen3-0.6B-singprobe \ --port 30000@article{singteam2026singprobe, title = {SingProbe Technical Report}, author = {Sing Team}, journal = {arXiv preprint arXiv:2608.30703}, year = {2026},}