Close Menu
    DevStackTipsDevStackTips
    • Home
    • News & Updates
      1. Tech & Work
      2. View All

      How To Prevent WordPress SQL Injection Attacks

      June 12, 2025

      Java never goes out of style: Celebrating 30 years of the language

      June 12, 2025

      OpenAI o3-pro available in the API, BrowserStack adds Playwright support for real iOS devices, and more – Daily News Digest

      June 12, 2025

      Creating The “Moving Highlight” Navigation Bar With JavaScript And CSS

      June 11, 2025

      Surface Pro 11 with Snapdragon X Elite drops to lowest price ever

      June 12, 2025

      With WH40K Boltgun and Dungeons of Hinterberg, this month’s Humble Choice lineup is stacked for less than $12

      June 12, 2025

      I’ve been loving the upgrade to my favorite mobile controller, and there’s even a version for large tablets

      June 12, 2025

      Copilot Vision just launched — and Microsoft already added new features

      June 12, 2025
    • Development
      1. Algorithms & Data Structures
      2. Artificial Intelligence
      3. Back-End Development
      4. Databases
      5. Front-End Development
      6. Libraries & Frameworks
      7. Machine Learning
      8. Security
      9. Software Engineering
      10. Tools & IDEs
      11. Web Design
      12. Web Development
      13. Web Security
      14. Programming Languages
        • PHP
        • JavaScript
      Featured

      Master Data Management: The Key to Improved Analytics Reporting

      June 12, 2025
      Recent

      Master Data Management: The Key to Improved Analytics Reporting

      June 12, 2025

      Salesforce Lead-to-Revenue Management

      June 12, 2025

      React Native 0.80 – React 19.1, JS API Changes, Freezing Legacy Arch and much more

      June 12, 2025
    • Operating Systems
      1. Windows
      2. Linux
      3. macOS
      Featured

      Surface Pro 11 with Snapdragon X Elite drops to lowest price ever

      June 12, 2025
      Recent

      Surface Pro 11 with Snapdragon X Elite drops to lowest price ever

      June 12, 2025

      With WH40K Boltgun and Dungeons of Hinterberg, this month’s Humble Choice lineup is stacked for less than $12

      June 12, 2025

      I’ve been loving the upgrade to my favorite mobile controller, and there’s even a version for large tablets

      June 12, 2025
    • Learning Resources
      • Books
      • Cheatsheets
      • Tutorials & Guides
    Home»Development»Machine Learning»Teaching AI to Say ‘I Don’t Know’: A New Dataset Mitigates Hallucinations from Reinforcement Finetuning

    Teaching AI to Say ‘I Don’t Know’: A New Dataset Mitigates Hallucinations from Reinforcement Finetuning

    June 6, 2025

    Reinforcement finetuning uses reward signals to guide the large language model toward desirable behavior. This method sharpens the model’s ability to produce logical and structured outputs by reinforcing correct responses. Yet, the challenge persists in ensuring that these models also know when not to respond—particularly when faced with incomplete or misleading questions that don’t have a definite answer.

    The problem arises when language models, after reinforcement finetuning, begin to lose their ability to refuse to answer unclear or ambiguous queries. Instead of signaling uncertainty, the models tend to produce confidently stated but incorrect responses. This phenomenon, identified in the paper as the “hallucination tax,” highlights a growing risk. As models are trained to perform better, they may also become more likely to hallucinate answers in situations where silence would be more appropriate. This is especially hazardous in domains that require high trust and precision.

    Tools currently used in training large language models often overlook the importance of refusal behavior. Reinforcement finetuning frameworks tend to reward only correct answers while penalizing incorrect ones, ignoring cases where a valid response should be no answer at all. The reward systems in use do not sufficiently reinforce refusal, resulting in overconfident models. For instance, the paper shows that refusal rates dropped to near zero across multiple models after standard RFT, demonstrating that current training fails to address hallucination properly.

    Researchers from the University of Southern California developed the Synthetic Unanswerable Math (SUM) dataset. SUM introduces implicitly unanswerable math problems by modifying existing questions through criteria such as missing key information or creating logical inconsistencies. The researchers used DeepScaleR as the base dataset and employed the o3-mini model to generate high-quality unanswerable questions. This synthetic dataset aims to teach models to recognize when a problem lacks sufficient information and respond accordingly.

    SUM’s core technique is to mix answerable and unanswerable problems during training. Questions are modified to become ambiguous or unsolvable while maintaining plausibility. The training prompts instruct models to say “I don’t know” for unanswerable inputs. By introducing only 10% of the SUM data into reinforcement finetuning, models begin to leverage inference-time reasoning to evaluate uncertainty. This structure allows them to refuse answers more appropriately without impairing their performance on solvable problems.

    Performance analysis shows significant improvements. After training with SUM, the Qwen2.5-7B model increased its refusal rate from 0.01 to 0.73 on the SUM benchmark and from 0.01 to 0.81 on the UMWP benchmark. On the SelfAware dataset, refusal accuracy rose dramatically from 0.01 to 0.94. Llama-3.1-8B-Instruct showed a similar trend, with refusal rates improving from 0.00 to 0.75 on SUM and from 0.01 to 0.79 on UMWP. Despite these gains in refusal behavior, accuracy on answerable datasets, such as GSM8K and MATH-500, remained stable, with most changes ranging from 0.00 to -0.05. The minimal drop indicates that refusal training can be introduced without major sacrifices in task performance.

    This study outlines a clear trade-off between improved reasoning and trustworthiness. Reinforcement finetuning, while powerful, tends to suppress cautious behavior. The SUM dataset corrects this by teaching models to recognize what they cannot solve. With only a small addition to training data, language models become better at identifying the boundaries of their knowledge. This approach marks a significant step in making AI systems not just smarter but also more careful and honest.


    Check out the Paper and Dataset on Hugging Face. All credit for this research goes to the researchers of this project.

    🆕 Did you know? Marktechpost is the fastest-growing AI media platform—trusted by over 1 million monthly readers. Book a strategy call to discuss your campaign goals. Also, feel free to follow us on Twitter and don’t forget to join our 95k+ ML SubReddit and Subscribe to our Newsletter.

    The post Teaching AI to Say ‘I Don’t Know’: A New Dataset Mitigates Hallucinations from Reinforcement Finetuning appeared first on MarkTechPost.

    Source: Read More 

    Facebook Twitter Reddit Email Copy Link
    Previous ArticleAlibaba Qwen Team Releases Qwen3-Embedding and Qwen3-Reranker Series – Redefining Multilingual Embedding and Ranking Standards
    Next Article FLUX.1 Kontext — The First AI Image Editor I Can Actually Control

    Related Posts

    Machine Learning

    How to Evaluate Jailbreak Methods: A Case Study with the StrongREJECT Benchmark

    June 12, 2025
    Machine Learning

    How VideoAmp uses Amazon Bedrock to power their media analytics interface

    June 12, 2025
    Leave A Reply Cancel Reply

    For security, use of Google's reCAPTCHA service is required which is subject to the Google Privacy Policy and Terms of Use.

    Continue Reading

    CVE-2025-40574 – “Siemens SCALANCE LPE9403 Privilege Escalation”

    Common Vulnerabilities and Exposures (CVEs)

    Critical Cisco IOS XE Vulnerability Allows Arbitrary File Upload – PoC Released

    Security

    Gemini 2.5: Our most intelligent AI model

    Artificial Intelligence

    CVE-2025-47576 – Bringthepixel Bimber Remote File Inclusion Vulnerability

    Common Vulnerabilities and Exposures (CVEs)

    Highlights

    News & Updates

    Windows 11 is getting its own version of the Mac’s “Handoff” feature — resume apps across Android and PC!

    May 20, 2025

    Microsoft has revealed that it’s bringing “Cross Device Resume” to Windows 11, which will let…

    Mozilla Say Google Search Deal Vital to Firefox’s Survival

    May 5, 2025

    $800 million and 13 years later, here’s how you can play one of the world’s most expensive PC games for free

    May 22, 2025

    New generative media models and tools, built with and for creators

    May 29, 2025
    © DevStackTips 2025. All rights reserved.
    • Contact
    • Privacy Policy

    Type above and press Enter to search. Press Esc to cancel.