Researchers at ACL 2026 introduced a method to reuse LLM hidden states for classification tasks, training lightweight probes on token- and layer-selective representations to eliminate separate guard m…
Researchers at Northwestern University and the University of Chicago developed a computational method to measure open codes in qualitative analysis, using LLM-enriched hierarchical clustering and four…
Researchers Bhavik Chandna and Procheta Sen introduced a counterfactual explanation framework for retrieval models that identifies which terms need to be added to a document to improve its ranking, ad…
Researchers introduced PARAMΔ, a method that upcycles a dense LLM into a Mixture-of-Experts (MoE) architecture to expand language capabilities without costly alignment. By grafting a MoE-expanded para…
Researchers at French institutions found that diversity-driven sampling can reduce pre-training dataset size by up to 94% and training time by 73% while maintaining performance in ModernBERT models. I…
Researchers at the Association for Computational Linguistics introduced a Dual-Phase Self-Evolution (DPSE) framework for large language models that jointly optimizes user preference adaptation and dom…
Researchers at the 64th Annual Meeting of the Association for Computational Linguistics introduced a dual-task paradigm combining arithmetic and sentence comprehension to study how language models all…
Researchers introduced DySECT, a dynamic self-evolving extraction system that uses a closed-loop cycle where an LLM extracts triples to populate a knowledge base, which then improves the LLM via promp…
Researchers at ACL 2026 found that entropy collapse in LLM reasoning, which undermines test-time scaling, is driven by premature overconfidence at a small set of critical tokens. They proposed SCOPE, …