datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
FabricationGuard-linearprobe-qwen36-27b
🛡️ FabricationGuard — Linear Probe for Qwen3.6-27B
Activation-probe fabrication detection for Qwen3.6-27B. AUROC 0.88 cross-task on SimpleQA, -88% confident-wrong rate reduction in mitigation mode, ~1ms scoring latency.
This is the OpenInterp FabricationGuard production probe — derived from a multi-feature linear probe on the residual stream at layer 31, trained on a multi-benchmark hallucination corpus, validated cross-task on held-out splits.
Value
Base model… See the full description on the dataset page: https://huggingface.co/datasets/caiovicentino1/FabricationGuard-linearprobe-qwen36-27b.ReasoningGuard-linearprobe-qwen36-27b
🧠 ReasonGuard v0.2 — Linear Probe at L55 / mid_think on Qwen3.6-27B
v0.2 update (2026-04-29) — multi-bench training thesis FALSIFIED.
v0.2 trained on combined GSM8K + StrategyQA + MATH (455 samples, 45.8% halu rate) — same methodology that gave FabricationGuard cross-task AUROC 0.882. Within-bench improved on GSM8K (0.888 → 0.908). Cross-domain transfer still fails: StrategyQA 0.612, MATH 0.500 (chance). Position-of-faithfulness in the deep residual stream is more strongly… See the full description on the dataset page: https://huggingface.co/datasets/caiovicentino1/ReasoningGuard-linearprobe-qwen36-27b.LinearEquationsThe linear equations in this dataset are in the form:
zy + ay + b + n = py + dy + c + r
with integer coefficients ranging from -10 to 10.