Empirical Methods in Natural Language Processing (EMNLP) 2025

Schedule

Wednesday, November 5

• POSTER
• Bias after Prompting: Persistent Discrimination in Large Language Models
• 08:00 - 09:00, Gather Session 1 (Virtual)
• Niv Sivakumar, Natalie Mackraz, Samira Khorshidi, Krishna Patel, Barry Theobald, Luca Zappella, Nick Apostoloff

• PRESENTATION, POSTER
• Speculative Streaming: Efficient and Scalable Speculative Decoding with Multi-Stream Attention
• 08:00 - 09:00, Gather Session 1 (Virtual)
• Nikhil Bhendawade, Irina Belousova, Qichen Fu, Henry Mason, Antonie Lin, Mohammad Rastegari, Mahyar Najibikohnehshahri

• DEMO
• MLX
• 09:00 - 18:00, Apple Booth
• We demonstrate large model inference and training on device using MLX. MLX is a flexible array framework that is optimized for Apple silicon and brought to you by Apple machine learning research. It enables training and inference of arbitrarily complex models on Apple silicon powered devices with great brevity and flexibility. In this demo we showcase fine-tuning of a 7B parameter LLM on an iPhone, image generation using a large diffusion model on an iPad and text generation using a number of frontier large language models on a cluster of four Mac Studios.

• POSTER
• PrimeX: A Dataset of Worldview, Opinion, and Explanation
• 11:00 - 12:30, Poster Session 1, Hall C3
• Rik Koncel-Kedziorski, Brihi Joshi (USC), Tim Paek

• POSTER
• Toward Machine Interpreting: Lessons from Human Interpreting Studies
• 11:00 - 12:30, Poster Session 1, Hall C3
• Matthias Sperber, Maureen de Seyssel, Jayson Bao, Matthias Paulik

• PRESENTATION, POSTER
• Evaluating Evaluation Metrics — The Mirage of Hallucination Detection
• 13:00 - 14:00, Findings 1, Hall C3
• Atharva Kulkarni (USC), Yuan Zhang, Joel Ruben Antony Moniz (DoorDash), Hugh Ge, Andy Tseng, Dhivya Piraviperumal, Swabha Swayamdipta (USC), Hong Yu

• POSTER
• Memory-Efficient Backpropagation for Fine-Tuning LLMs on Resource-Constrained Mobile Devices
• 14:30 - 16:00, Industry Poster Session 1, Hall C3
• Congzheng Song, Xinyu Tang

• PRESENTATION
• Discriminating Form and Meaning in Multilingual Models with Minimal-Pair ABX Tasks
• 17:00 - 17:15, Resources and Evaluation, A102-A103
• Maureen de Seyssel, Jie Chi, Skyler Seto, Maartje ter Hoeve, Masha Fedzechkina Donaldson, Natalie Schluter

• POSTER
• CLIP-UP: A Simple and Efficient Mixture-of-Experts CLIP Training Recipe with Sparse Upcycling
• 18:00 - 19:00, Gather Session 2 (Virtual)
• Angie Wang, Chen Chen, Yinfei Yang, Hong-You Chen, Bowen Zhang, Adi Pal, Xiangxin Zhu, Xianzhi Du

Thursday, November 6

• POSTER
• Analyzing Dialectical Biases in LLMs for Knowledge and Reasoning Benchmarks
• 08:00 - 09:00, Gather Session 3 (Virtual)
• Eileen Pan (Cornell), Anna Choi (Cornell), Maartje ter Hoeve, Skyler Seto, Allison Koenecke (Cornell)

• DEMO
• MLX
• 9:00 - 18:00, Apple Booth
• We demonstrate large model inference and training on device using MLX. MLX is a flexible array framework that is optimized for Apple silicon and brought to you by Apple machine learning research. It enables training and inference of arbitrarily complex models on Apple silicon powered devices with great brevity and flexibility. In this demo we showcase fine-tuning of a 7B parameter LLM on an iPhone, image generation using a large diffusion model on an iPad and text generation using a number of frontier large language models on a cluster of four Mac Studios.

• POSTER
• Improving Language Model Personas via Rationalization with Psychological Scaffolds
• 12:30 - 13:30, Findings 2, Hall C3
• Brihi Joshi (USC), Xiang Ren (USC), Swabha Swayamdipta (USC), Rik Koncel-Kedziorski, Tim Paek

Friday, November 7

• WORKSHOP
• PolyNorm: Few-Shot LLM-Based Text Normalization for Text-to-Speech
• 08:00 - 09:00, Gather Session 4 (Virtual)
• Michel Wong, Haotian He, Ali Alshehri, Sophia Kao

• POSTER
• Leveraging the Power of Large Language Models in Entity Linking via Adaptive Routing and Targeted Reasoning
• 10:30 - 12:00, Industry Poster Session 3, Hall C3
• Yajie Li (University of Massachusetts Amherst), Albert Galimov (University of Massachusetts Amherst), Mitra Datta Ganapaneni (University of Massachusetts Amherst), Pujitha Thejaswi (University of Massachusetts Amherst), De Meng, Priyanshu Kumar, Saloni Potdar

Saturday, November 8

• AFFINITY EVENT
• Workshop on Widening Natural Language Processing (WiNLP)
• 09:00 - 17:30, Room A303
• Shu W., Tianjun Ye and Vivien Zhao will represent Apple at the WiNLP mentor luncheon, during the affinity workshop.