Papers by Bar Alon
Faithful Serum: Mitigating the Faithfulness Gap in Textual Explanations of LLM Decisions via Attribution Guidance (2026.acl-long)
Copied to clipboard
| Challenge: | Prior work has focused on generating convincing rationales that appear to be subjectively faithful, but it remains unclear whether these explanations are epistemic faithful. |
| Approach: | They propose a method that enhances epistemic faithfulness by guiding explanation generation through attention-level interventions, informed by token-level heatmaps. |
| Outcome: | The proposed method significantly improves epistemic faithfulness across multiple models, benchmarks, and prompts. |
Safeguarding Language Models via Self-Destruct Trapdoor (2026.eacl-long)
Copied to clipboard
| Challenge: | Existing mechanisms to restrict behavior of language models (LMs) are vulnerable to misuse and misalignment. |
| Approach: | They propose a mechanism to restrict specific behaviors in language models by exploiting hardware properties. |
| Outcome: | The proposed mechanism can be applied to trigger overflows for specific behaviors or target hardware malfunctions. |
ID10M-JAM: Stress-Testing Idiom Identification Under Challenging Context (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) achieve strong performance on idiom identification benchmarks, yet their robustness to misleading contextual signals remains largely untested. |
| Approach: | They propose an adversarial extension of the ID10M dataset that jams idiom understanding by injecting coherent but conflicting context before each target sentence. |
| Outcome: | The proposed benchmark exposes systematic vulnerabilities in LLMs’ contextual reasoning, pushing idiom identification to its breaking point. |
Beyond Pairwise: Global Zero-shot Temporal Graph Generation (2025.emnlp-main)
Copied to clipboard
| Challenge: | Temporal relation extraction (TRE) is a fundamental task in natural language processing (NLP) that involves identifying the temporal relationships between events in a document. |
| Approach: | They propose a method that generates a document’s complete temporal graph in a single step, followed by temporal constraint optimization to refine predictions and enforce temporal consistency across relations. |
| Outcome: | The proposed method outperforms existing zero-shot approaches and offers a competitive alternative to supervised TRE models. |