Search papers, labs, and topics across Lattice.
The paper introduces ThinknCheck, a 1B-parameter claim verification model that generates a structured rationale before making a binary verdict. They create a 24.1k reasoning-augmented dataset (LLMAggreFact-Think) and fine-tune a 4-bit Gemma3 model on it. ThinknCheck achieves state-of-the-art performance on LLMAggreFact and SciFact compared to larger models, demonstrating the effectiveness of explicit, supervised reasoning for compact verifiers.
Surprisingly, a 1B-parameter model, ThinknCheck, outperforms a 7B-parameter model in claim verification by explicitly generating and leveraging a structured rationale.
We present ThinknCheck, a 1B-parameter verifier for grounded claim verification that first produces a short, structured rationale and then a binary verdict. We construct LLMAggreFact-Think, a 24.1k reasoning-augmented training set derived from LLMAggreFact, and fine-tune a 4-bit Gemma3 model to follow this format. On LLMAggreFact, ThinknCheck attains 78.1 balanced accuracy (BAcc), surpassing MiniCheck-7B (77.4) with 7x fewer parameters; removing the reasoning step reduces BAcc to 57.5. On SciFact, ThinknCheck reaches 64.7 BAcc, a +14.7 absolute gain over MiniCheck-7B. By contrast, zero-shot chain-of-thought on the base Gemma3-1B harms accuracy relative to direct answers, and preference optimization with a simple format+accuracy reward underperforms supervised reasoning. To probe the latter, we introduce GSMClaims and a domain-specialized variant, ThinknCheck-Science, which improves across benchmarks, including 61.0\% accuracy on GSMClaims. Overall, explicit, supervised reasoning enables compact verifiers that are competitive while remaining resource-efficient and interpretable.
