37 citations · 93 across the 25 of their papers we have counts for
26 papers
Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs
Gabrielle Kaili-May Liu, Avi Caciularu, Gal Yona +2
Metacognition is a critical component of intelligence that describes the ability to monitor and regulate one's own cognitive processes. Yet LLMs exhibit systemic deficiencies in ke…
Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models
Dehai Min, Giovanni Vaccarino, Huiyi Chen +3
Large Reasoning Models (LRMs) achieve strong performance by generating long chains of thought (CoT), but often overthink, continuing to reason after a solution has already stabiliz…
Hallucinations Undermine Trust; Metacognition is a Way Forward
Gal Yona, Mor Geva, Yossi Matias
Despite significant strides in factual reliability, errors -- often termed hallucinations -- remain a major concern for generative AI, especially as LLMs are increasingly expected…
Empty Shelves or Lost Keys? Recall Is the Bottleneck for Parametric Factuality
Nitay Calderon, Eyal Ben-David, Zorik Gekhman +2
Standard factuality evaluations of LLMs treat all errors alike, obscuring whether failures arise from missing knowledge (empty shelves) or from limited access to encoded facts (los…
SimpleQA Verified: A Reliable Factuality Benchmark to Measure Parametric Knowledge
Lukas Haas, Gal Yona, Giovanni D'Antonio +2
We introduce SimpleQA Verified, a 1,000-prompt benchmark for evaluating Large Language Model (LLM) short-form factuality based on OpenAI's SimpleQA. It addresses critical limitatio…
MetaFaith: Faithful Natural Language Uncertainty Expression in LLMs
Gabrielle Kaili-May Liu, Gal Yona, Avi Caciularu +3
A critical component in the trustworthiness of LLMs is reliable uncertainty communication, yet LLMs often use assertive language when conveying false claims, leading to over-relian…
Neural Descriptors: Self-Supervised Learning of Robust Local Surface Descriptors Using Polynomial Patches
Gal Yona, Roy Velich, Ron Kimmel +1
Classical shape descriptors such as Heat Kernel Signature (HKS), Wave Kernel Signature (WKS), and Signature of Histograms of OrienTations (SHOT), while widely used in shape analysi…
Confidence Improves Self-Consistency in LLMs
Amir Taubenfeld, Tom Sheffer, Eran Ofek +4
Self-consistency decoding enhances LLMs' performance on reasoning tasks by sampling diverse reasoning paths and selecting the most frequent answer. However, it is computationally e…
Pathways on the Image Manifold: Image Editing via Video Generation
Noam Rotstein, Gal Yona, Daniel Silver +3
Recent advances in image editing, driven by image diffusion models, have shown remarkable progress. However, significant challenges remain, as these models often struggle to follow…
Keep Guessing? When Considering Inference Scaling, Mind the Baselines
Gal Yona, Or Honovich, Omer Levy +1
Scaling inference compute in large language models (LLMs) through repeated sampling consistently increases the coverage (fraction of problems solved) as the number of samples incre…
Can Large Language Models Faithfully Express Their Intrinsic Uncertainty in Words?
Gal Yona, Roee Aharoni, Mor Geva
We posit that large language models (LLMs) should be capable of expressing their intrinsic uncertainty in natural language. For example, if the LLM is equally likely to output two…
Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?
Zorik Gekhman, Gal Yona, Roee Aharoni +4
When large language models are aligned via supervised fine-tuning, they may encounter new factual information that was not acquired through pre-training. It is often conjectured th…
Narrowing the Knowledge Evaluation Gap: Open-Domain Question Answering with Multi-Granularity Answers
Gal Yona, Roee Aharoni, Mor Geva
Factual questions typically can be answered correctly at different levels of granularity. For example, both ``August 4, 1961'' and ``1961'' are correct answers to the question ``Wh…
Surfacing Biases in Large Language Models using Contrastive Input Decoding
Gal Yona, Or Honovich, Itay Laish +1
Ensuring that large language models (LMs) are fair, robust and useful requires an understanding of how different modifications to their inputs impact the model's behaviour. In the…
Malign Overfitting: Interpolation Can Provably Preclude Invariance
Yoav Wald, Gal Yona, Uri Shalit +1
Learned classifiers should often possess certain invariance properties meant to encourage fairness, robustness, or out-of-distribution generalization. However, multiple recent work…
Useful Confidence Measures: Beyond the Max Score
Gal Yona, Amir Feder, Itay Laish
An important component in deploying machine learning (ML) in safety-critic applications is having a reliable measure of confidence in the ML model's predictions. For a classifier $…
Active Learning with Label Comparisons
Gal Yona, Shay Moran, Gal Elidan +1
Supervised learning typically relies on manual annotation of the true labels. When there are many potential classes, searching for the best one can be prohibitive for a human annot…
Decision-Making under Miscalibration
Guy N. Rothblum, Gal Yona
ML-based predictions are used to inform consequential decisions about individuals. How should we use predictions (e.g., risk of heart attack) to inform downstream binary classifica…
On Fairness and Stability in Two-Sided Matchings
Gili Karni, Guy N. Rothblum, Gal Yona
There are growing concerns that algorithms, which increasingly make or influence important decisions pertaining to individuals, might produce outcomes that discriminate against pro…
Revisiting Sanity Checks for Saliency Maps
Gal Yona, Daniel Greenfeld
Saliency methods are a popular approach for model debugging and explainability. However, in the absence of ground-truth data for what the correct maps should be, evaluating and com…
Consider the Alternatives: Navigating Fairness-Accuracy Tradeoffs via Disqualification
Guy N. Rothblum, Gal Yona
In many machine learning settings there is an inherent tension between fairness and accuracy desiderata. How should one proceed in light of such trade-offs? In this work we introdu…
Multi-group Agnostic PAC Learnability
Guy N Rothblum, Gal Yona
An agnostic PAC learning algorithm finds a predictor that is competitive with the best predictor in a benchmark hypothesis class, where competitiveness is measured with respect to…
Outcome Indistinguishability
Cynthia Dwork, Michael P. Kim, Omer Reingold +2
Prediction algorithms assign numbers to individuals that are popularly understood as individual "probabilities" -- what is the probability of 5-year survival after cancer diagnosis…
Who's responsible? Jointly quantifying the contribution of the learning algorithm and training data
Gal Yona, Amirata Ghorbani, James Zou
A learning algorithm trained on a dataset is revealed to have poor performance on some subpopulation at test time. Where should the responsibility for this lay? It can be a…