agentic-tool-calls-where-parameter-fabricated-rather-than-grounded CLM-0360For agentic tool calls where a parameter is fabricated rather than grounded in the dialogue, sampling-based uncertainty detectors (token entropy, semantic entropy, SEP) fail at near-chance because the model fabricates plausible parameters with high confidence, while the model's hidden states at the moment the parameter is generated do carry a separable grounding signal.
Capability: Stating false facts confidently
Observed on
Tool-calling agents (Qwen3-4B, Llama-xLAM-2-8B) under underspecified user instructions; hallucination defined as a parameter unsupported by dialogue history, not factually wrong..
Sources
- Validated by: Detection AUROC 0.966 ID / 0.925 OOD on Qwen3-4B vs 0.915/0.884 for an equal-scale fine-tuned external judge, 0.782/0.616 for a linear probe, and ~0.5 for entropy baselines; >80% exact-parameter localization; in a closed ReAct loop, specific feedback gave parameter F1 61.2 vs 51.3 generic and 52.1 base, false blocks 2.9% vs 9.4%, recovery rate 37.0% vs 23.9%; <10ms overhead in vLLM.. [truncated]
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- For long-horizon LLM agents on tool-use and terminal benchmarks, token-level confidence (predictive entropy, sequence probability) barely separates successful from failed trajectories, and the separable signal lives in the dependency structure of the trajectory: scoring a trajectory by propagating per-step uncertainty over a graph of reasoning, tool-call and environment-feedback nodes raises failure-detection AUROC over both local-confidence and sequence-based agent UQ baselines.Using the tools it is given · reviewed by AI
- A linear probe on a tool-calling model's hidden state at the last generated token detects incorrect tool-calls — including wrong-but-well-typed argument values that parsers and logs do not catch — and works better on larger models and at middle-to-late layers than at the final layer.Using the tools it is given · reviewed by AI
- For tool-calling agents that fabricate parameters never given by the user, a zero-initialized LoRA adapter run concurrently with the frozen base model and trained only to emit a localized verdict at an appended trigger token detects these ungrounded parameters better than sampling-based entropy methods, passive linear probes, and an equivalent-scale fine-tuned external judge, and naming the specific hallucinated parameter back to the agent recovers more of the task than a generic block.Stating false facts confidently · reviewed by AI
- For open models (8B-14B) filling structured tool-call parameters, a linear probe on the hidden state at the position just before a parameter value is written predicts whether that value will be correct better than token log-probability (in-domain AUC about 0.986 vs 0.914), and using that score to filter self-generated calls for fine-tuning beats self-training filtered by log-probability or majority vote.Using the tools it is given · reviewed by AI
- A linear probe on a tool-calling model's hidden state at the last generated token detects incorrect tool-calls — including wrong-but-well-typed argument values that parsers and logs do not catch — and works better on larger models and at middle-to-late layers than at the final layer.Using the tools it is given · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: Show that on the same specification-grounding tool-call data, semantic entropy or token entropy separates ungrounded from grounded parameters at AUROC comparable to hidden-state methods, or that a linear probe on raw base states transfers OOD as well as the adapted states.