CRITICTOOL, a benchmark of 2,740 tool-calling error scenarios built from BFCL and T-Eval with GPT-4o-based error injection, finds most LLMs rarely recover from tool-use errors, with GPT-4o best at 69.01 overall and tool-finetuned ToolLLaMA2 at 0.58.
Varying Shades of Wrong: Aligning LLMs with Wrong Answers Only
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
In the absence of abundant reliable annotations for challenging tasks and contexts, how can we expand the frontier of LLM capabilities with potentially wrong answers? We focus on two research questions: (1) Can LLMs generate reliable preferences among wrong options? And if so, (2) Would alignment with such wrong-over-wrong preferences be helpful? We employ methods based on self-consistency, token probabilities, and LLM-as-a-judge to elicit wrong-over-wrong preferences, and fine-tune language models with preference optimization approaches using these synthesized preferences. Extensive experiments with seven LLMs and eight datasets demonstrate that (1) LLMs do have preliminary capability in distinguishing various shades of wrong, achieving up to 20.9% higher performance than random guess; (2) Alignment with wrong-over-wrong preferences helps LLMs to produce less wrong and sometimes even outright correct answers, while overall improving model calibration.
fields
cs.SE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CRITICTOOL: Evaluating Self-Critique Capabilities of Large Language Models in Tool-Calling Error Scenarios
CRITICTOOL, a benchmark of 2,740 tool-calling error scenarios built from BFCL and T-Eval with GPT-4o-based error injection, finds most LLMs rarely recover from tool-use errors, with GPT-4o best at 69.01 overall and tool-finetuned ToolLLaMA2 at 0.58.