ACL 2025: Search-based multilingual LLM evaluation that discovers English-correct, target-language-wrong failures; code and 6,713-pair dataset across 16 languages.
-
Updated
Sep 21, 2026 - Python
ACL 2025: Search-based multilingual LLM evaluation that discovers English-correct, target-language-wrong failures; code and 6,713-pair dataset across 16 languages.
This is the official implementation of our ACL 2025 Main paper "Balancing Diversity and Risk in LLM Sampling".
Code and data for the Findings of ACL 2025 paper "Voting or Consensus? Decision-Making in Multi-Agent Debate". It compares seven decision protocols in multi-agent LLM debate: voting does 13.2% better on reasoning tasks and consensus 2.8% better on knowledge tasks.
ACL Findings 2025 | RaPID: retrieval-augmented long text generation with planning and information discovery.
Official implementation of "ORMind: A Cognitive-Inspired End-to-End Reasoning Framework for Operations Research" (ACL 2025 Industry Track).
Official source code for the paper "Leveraging Context for Multimodal Fallacy Classification in Political Debates", accepted at the ArgMining Workshop @ ACL 2025.
Official ACL 2025 implementation of PRISM: interpretable political bias embeddings produced by a political-aware cross-encoder.
Reproduction of the MIND zero-shot harmful meme detection pipeline on FHM using LLaVA-1.5-13B.
ACL 2025 research code for explainable ICD-10 code prediction with clinical entities
Project page for our ACL 2025 GeBNLP paper — Assessing the Reliability of LLMs Annotations in the Context of Demographic Bias and Model Explanation
Resources and author information for our ACL 2025 (GeBNLP) paper on the reliability of LLM annotations
The official repository for our ACL 2025 paper, "Who Writes What: Unveiling the Impact of Author Roles on AI-generated Text Detection"
To associate your repository with the acl-2025 topic, visit your repo's landing page and select "manage topics."