High Accuracy, Less Talk (HALT): Reliable LLMs through Capability-Aligned Finetuning

• URL: http://arxiv.org/abs/2506.04051v1
• Date: Wed, 04 Jun 2025 15:16:21 GMT
• Title: High Accuracy, Less Talk (HALT): Reliable LLMs through Capability-Aligned Finetuning
• Authors: Tim Franzmeyer, Archie Sravankumar, Lijuan Liu, Yuning Mao, Rui Hou, Sinong Wang, Jakob N. Foerster, Luke Zettlemoyer, Madian Khabsa,
• Abstract summary: Large Language Models (LLMs) currently respond to every prompt.<n>LLMs can produce incorrect answers when they lack knowledge or capability.<n>We propose post-training an LLM to generate content only when confident in its correctness and to otherwise abstain.
• Score: 84.52940628494879
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models (LLMs) currently respond to every prompt. However, they can produce incorrect answers when they lack knowledge or capability -- a problem known as hallucination. We instead propose post-training an LLM to generate content only when confident in its correctness and to otherwise (partially) abstain. Specifically, our method, HALT, produces capability-aligned post-training data that encodes what the model can and cannot reliably generate. We generate this data by splitting responses of the pretrained LLM into factual fragments (atomic statements or reasoning steps), and use ground truth information to identify incorrect fragments. We achieve capability-aligned finetuning responses by either removing incorrect fragments or replacing them with "Unsure from Here" -- according to a tunable threshold that allows practitioners to trade off response completeness and mean correctness of the response's fragments. We finetune four open-source models for biography writing, mathematics, coding, and medicine with HALT for three different trade-off thresholds. HALT effectively trades off response completeness for correctness, increasing the mean correctness of response fragments by 15% on average, while resulting in a 4% improvement in the F1 score (mean of completeness and correctness of the response) compared to the relevant baselines. By tuning HALT for highest correctness, we train a single reliable Llama3-70B model with correctness increased from 51% to 87% across all four domains while maintaining 53% of the response completeness achieved with standard finetuning.

Related papers

• Fewer Hallucinations, More Verification: A Three-Stage LLM-Based Framework for ASR Error Correction [4.304383298057423]
  We propose the Reliable Correction Framework (RLLM-CF), which consists of three stages: error pre-detection, chain-of-thought sub-tasks iterative correction, and reasoning process verification.<n>Experiments on AISHELL-1, AISHELL-2, and Librispeech show that the GPT-4o model enhanced by our framework achieves 21%, 11%, 9%, and 11.4% relative reductions in CER/WER.
  arXiv  Detail & Related papers  (2025-05-30T08:40:49Z)
• From Misleading Queries to Accurate Answers: A Three-Stage Fine-Tuning Method for LLMs [5.23164145730825]
  Large language models (LLMs) exhibit excellent performance in natural language processing (NLP)<n>Existing methods focus on correcting the output, but they often overlook the potential of improving the ability of LLMs to detect and correct misleading content in the input itself.<n>We propose a novel three-stage fine-tuning method that enhances the ability of LLMs to detect and correct misleading information in the input.
  arXiv  Detail & Related papers  (2025-04-15T15:16:45Z)
• Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory [52.44029486173232]
  Dynamic Cheatsheet (DC) is a lightweight framework that endows a black-box language model with a persistent, evolving memory.<n>DC enables models to store and reuse accumulated strategies, code snippets, and general problem-solving insights at inference time.<n>This test-time learning enhances performance substantially across a range of tasks without needing explicit ground-truth labels or human feedback.
  arXiv  Detail & Related papers  (2025-04-10T17:57:33Z)
• Verbosity $\neq$ Veracity: Demystify Verbosity Compensation Behavior of Large Language Models [8.846200844870767]
  We discover an understudied type of undesirable behavior of Large Language Models (LLMs)<n>We term Verbosity Compensation (VC) as similar to the hesitation behavior of humans under uncertainty.<n>We propose a simple yet effective cascade algorithm that replaces verbose responses with the other model-generated responses.
  arXiv  Detail & Related papers  (2024-11-12T15:15:20Z)
• LLM Robustness Against Misinformation in Biomedical Question Answering [50.98256373698759]
  The retrieval-augmented generation (RAG) approach is used to reduce the confabulation of large language models (LLMs) for question answering. We evaluate the effectiveness and robustness of four LLMs against misinformation in answering biomedical questions.
  arXiv  Detail & Related papers  (2024-10-27T16:23:26Z)
• LACIE: Listener-Aware Finetuning for Confidence Calibration in Large Language Models [69.68379406317682]
  We introduce a listener-aware finetuning method (LACIE) to calibrate implicit and explicit confidence markers. We show that LACIE models the listener, considering not only whether an answer is right, but whether it will be accepted by a listener. We find that training with LACIE results in 47% fewer incorrect answers being accepted while maintaining the same level of acceptance for correct answers.
  arXiv  Detail & Related papers  (2024-05-31T17:16:38Z)
• GRATH: Gradual Self-Truthifying for Large Language Models [63.502835648056305]
  GRAdual self-truTHifying (GRATH) is a novel post-processing method to enhance truthfulness of large language models (LLMs) GRATH iteratively refines truthfulness data and updates the model, leading to a gradual improvement in model truthfulness in a self-supervised manner. GRATH achieves state-of-the-art performance on TruthfulQA, with MC1 accuracy of 54.71% and MC2 accuracy of 69.10%, which even surpass those on 70B-LLMs.
  arXiv  Detail & Related papers  (2024-01-22T19:00:08Z)
• The Earth is Flat? Unveiling Factual Errors in Large Language Models [89.94270049334479]
  Large Language Models (LLMs) like ChatGPT are in various applications due to their extensive knowledge from pre-training and fine-tuning. Despite this, they are prone to generating factual and commonsense errors, raising concerns in critical areas like healthcare, journalism, and education. We introduce a novel, automatic testing framework, FactChecker, aimed at uncovering factual inaccuracies in LLMs.
  arXiv  Detail & Related papers  (2024-01-01T14:02:27Z)
• Quantifying Uncertainty in Answers from any Language Model and Enhancing their Trustworthiness [16.35655151252159]
  We introduce BSDetector, a method for detecting bad and speculative answers from a pretrained Large Language Model. Our uncertainty quantification technique works for any LLM accessible only via a black-box API.
  arXiv  Detail & Related papers  (2023-08-30T17:53:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.