Pure language processing (NLP) stays some of the shortly evolving fields in AI, as new analysis continues to quickly advance massive language fashions (LLMs), methods for speech recognition and era, language brokers, and extra. This know-how is crucial to lots of right now’s AI experiences, together with Apple Intelligence and Siri, and elementary analysis in NLP can be foundational to future AI.
Apple lately hosted the Workshop on Pure Language and Interactive Techniques, bringing collectively Apple and members of the educational analysis group for a two-day occasion targeted on latest advances in NLP. The workshop targeted on three key analysis themes – Spoken Language Interactive Techniques, LLM Coaching and Alignment, and Language Brokers – and explored novel approaches to enabling intuitive interactions with gadgets by way of pure language understanding and era. All through the workshop, talks and discussions underscored the significance of privateness, safety, efficiency, and effectivity on this quickly altering panorama.
On this submit, we share recordings of chosen talks and a recap of the publications mentioned on the workshop.
Apple Workshop on Pure Language and Interactive Techniques 2025 Movies
Workshop Sources
Printed Work Introduced on the Workshop
2 OLMo 2 Livid by Pete Walsh (Allen Institute for AI), Luca Soldaini (Allen Institute for AI), Dirk Groeneveld (Allen Institute for AI), Kyle Lo (Allen Institute for AI), Shane Arora (Allen Institute for AI), Akshita Bhagia (Allen Institute for AI), Yuling Gu (Allen Institute for AI), Shengyi Huang (Allen Institute for AI), Matt Jordan (Allen Institute for AI), Nathan Lambert (Allen Institute for AI), Dustin Schwenk (Allen Institute for AI), Oyvind Tafjord (Allen Institute for AI), Taira Anderson (Allen Institute for AI), David Atkinson (Allen Institute for AI), Faeze Brahman (Allen Institute for AI), Christopher Clark (Allen Institute for AI), Pradeep Dasigi (Allen Institute for AI), Nouha Dziri (Allen Institute for AI), Michal Guerquin (Allen Institute for AI), Hamish Ivison (Allen Institute for AI, College of Washington), Pang Wei Koh (Allen Institute for AI, College of Washington), Jiacheng Liu (Allen Institute for AI, College of Washington), Saumya Malik (Allen Institute for AI), William Merrill (Allen Institute for AI, New York College), Lester James V. Miranda (Allen Institute for AI), Jacob Morrison (Allen Institute for AI), Tyler Murray (Allen Institute for AI), Crystal Nam (Allen Institute for AI), Valentina Pyatkin (Allen Institute for AI, College of Washington), Aman Rangapur (Allen Institute for AI), Michael Schmitz (Allen Institute for AI), Sam Skjonsberg (Allen Institute for AI), David Wadden (Allen Institute for AI), Christopher Wilhelm (Allen Institute for AI), Michael Wilson (Allen Institute for AI), Luke Zettlemoyer (College of Washington), Ali Farhadi (Allen Institute for AI, College of Washington), Noah A. Smith (Allen Institute for AI, College of Washington), and Hannaneh Hajishirzi (Allen Institute for AI, College of Washington)
Adaptable Logical Management for Massive Language Fashions by Honghua Zhang (UCLA), Po-Nien Kung (UCLA), Masahiro Yoshida (UCLA, Sony Group Company), Man Van den Broeck (UCLA), and Nanyun Peng (UCLA)
Agent S: An Open Agentic Framework that Makes use of Computer systems Like a Human by Saaket Agashe (UC Santa Cruz), Jiuzhou Han (Monash College), Shuyu Gan (UC Santa Cruz), Jiachen Yang (Tianjin College), Ang Li, and Xin Eric Wang (UC Santa Cruz)
AI fashions collapse when educated on recursively generated knowledge by Ilia Shumailov (College of Oxford), Zakhar Shumaylov (College of Cambridge), Yiren Zhao (Imperial School London), Nicolas Papernot (College of Toronto, Vector Institute), Ross Anderson (College of Cambridge, College of Edinburgh), Yarin Gal (College of Oxford)
DASB – Discrete Audio and Speech Benchmark by Pooneh Mousavi (Concordia College, Mila – Quebec AI Institute), Luca Della Libera (Concordia College, Mila – Quebec AI Institute), Jarod Duret (Avignon Université), Artem Ploujnikov (Université de Montréal, Université Laval, Mila – Quebec AI Institute), Cem Subakan (Université Laval, Mila – Quebec AI Institute, Concordia College), and Mirco Ravanelli (Concordia College, Mila – Quebec AI Institute, Université de Montréal)
Detecting hallucinations in massive language fashions utilizing semantic entropy by Sebastian Farquhar (College of Oxford) ), Jannik Kossen (College of Oxford), Lorenz Kuhn (College of Oxford), and Yarin Gal (College of Oxford)
Direct Massive Language Mannequin Alignment By Self-Rewarding Contrastive Immediate Distillation by Aiwei Liu, Haoping Bai, Zhiyun Lu, Xiang Kong, Simon Wang, Jiulong Shan, Meng Cao, and Lijie Wen
Fleurs-SLU: A Massively Multilingual Benchmark for Spoken Language Understanding by Fabian David Schmidt (College of Würzburg), Ivan Vulic (College of Cambridge), Goran Glavaš (College of Würzburg), and David Ifeoluwa Adelani (Mila, McGill College)
FocalCodec: Low-Bitrate Speech Coding through Focal Modulation Networks by Luca Della Libera (Concordia College, Mila-Quebec AI Institute), Francesco Paissan (Fondazione Bruno Kessler, Mila-Quebec AI Institute), Cem Subakan (Université Laval, Concordia College, Mila-Quebec AI Institute), and Mirco Ravanelli (Concordia College, Mila-Quebec AI Institute)
Speculation-Pushed Idea-of-Thoughts Reasoning for Massive Language Fashions by Hyunwoo Kim, Melanie Sclar (College of Washington), Tan Zhi-Xuan (MIT), Lance Ying (MIT, Harvard College), Sydney Levine (Allen Institute for AI), Yang Liu (Amazon), Joshua B. Tenenbaum (MIT), Yejin Choi (Stanford College)
IrokoBench: A New Benchmark for African Languages within the Age of Massive Language Fashions by David Ifeoluwa Adelani (Mila, McGill College, Canada CIFAR AI Chair), Jessica Ojo (Mila, McGill College, Lelapa AI), Israel Abebe Azime (Saarland College), Jian Yun Zhuang (College of Toronto), Jesujoba O. Alabi (Saarland College), Xuanli He (College School London), Millicent Ochieng (Microsoft Analysis Africa), Sara Hooker (Cohere For AI), Andiswa Bukula (SADiLaR), En-Shiun Annie Lee (Ontario Tech College), Chiamaka Chukwuneke (Lancaster College), Glad Buzaaba (Princeton College), Blessing Sibanda (Masakhane NLP), Godson Kalipe (Masakhane NLP), Jonathan Mukiibi (Makerere College), Salomon Kabongo (Leibniz Universität Hannover), Foutse Yuehgoh (Le CNAM), Mmasibidi Setaka, Lolwethu Ndolela (Masakhane NLP), Nkiruka Odu (Masakhane NLP), Rooweither Mabuya (SADiLaR), Shamsuddeen Hassan Muhammad (Imperial School London), Salomey Osei (Universidad de Deusto), Sokhar Samb (DAUST), Tadesse Kebede Guge (Haramaya College), Tombekai Vangoni Sherman, and Pontus Stenetorp (College School London)
Thoughts the Worth-Motion Hole: Do LLMs Act in Alignment with Their Values? by Hua Shen (College of Washington), Nicholas Clark (College of Washington), and Tanushree Mitra (College of Washington)
Mutual Reinforcement of LLM Dialogue Synthesis and Summarization Capabilities for Few-Shot Dialogue Summarization by Yen-Ju Lu, Ting-Yao Hu, Hema Swetha Koppula, Hadi Pour Ansari, Jen-Hao Rick Chang, Yin Xia, Xiang Kong, Qi Zhu, Simon Wang, Oncel Tuzel, and Raviteja Vemulapalli
Reinforcement Studying for Lengthy-Horizon Interactive LLM Brokers by Kevin Chen, Marco Cusumano-Towner, Brody Huval, Aleksei Petrenko, Jackson Hamburger, Vladlen Koltun, and Philipp Krähenbühl
Retrospective Studying from Interactions by Zizhao Chen (Cornell College), Mustafa Omer Gul (Cornell College), Yiwei Chen (Cornell College), Gloria Geng (Cornell College), Anne Wu (Cornell College), and Yoav Artzi (Cornell College)
s1: Easy Take a look at-Time Scaling by Niklas Muennighoff (Stanford College, Allen Institute for AI, Contextual AI), Zitong Yang (Stanford College), Weijia Shi (College of Washington, Allen Institute for AI), Xiang Lisa Li (Stanford College), Li Fei-Fei (Stanford College), Hannaneh Hajishirzi (College of Washington, Allen Institute for AI), Luke Zettlemoyer (College of Washington), Percy Liang (Stanford College), Emmanuel Candès (Stanford College), and Tatsunori Hashimoto (Stanford College)
Scaling Diffusion Language Fashions through Adaptation from Autoregressive Fashions by Shansan Gong (The College of Hong Kong), Shivam Agarwal (College of Illinois at Urbana-Champaign), Yizhe Zhang, Jiacheng Ye (The College of Hong Kong), Lin Zheng (The College of Hong Kong), Mukai Li (The College of Hong Kong), Chenxin An (The College of Hong Kong), Peilin Zhao (Tencent AI Lab), Wei Bi (Tencent AI Lab), Jiawei Han, Hao Peng (College of Illinois at Urbana-Champaign), and Lingpeng Kong (The College of Hong Kong)
SIB-200: A Easy, Inclusive, and Massive Analysis Dataset for Subject Classification in 200+ Languages and Dialects by David Ifeoluwa Adelani (College School London, Masakhane), Hannah Liu (College of Toronto), Xiaoyu Shen (Amazon Alexa), Nikita Vassilyev (College of Toronto), Jesujoba O. Alabi (Masakhane, Saarland College), Yanke Mao (College of Toronto), Haonan Gao (College of Toronto), and En-Shiun Annie Lee (College of Toronto, College of Ontario Institute of Expertise)
SILO Language Fashions: Isolating Authorized Threat In a Nonparametric Datastore by Sewon Min (College of Washington), Suchin Gururangan (College of Washington), Eric Wallace (UC Berkeley), Weijia Shi (College of Washington), Hannaneh Hajishirzi (College of Washington, Allen Institute for AI), Noah A. Smith (College of Washington, Allen Institute for AI), and Luke Zettlemoyer (College of Washington)
Speculative Streaming: Quick LLM Inference With out Auxiliary Fashions by Nikhil Bhendawade, Irina Belousova, Qichen Fu, Henry Mason, Mohammad Rastegari, and Mahyar Najibi
The Function of Prosody in Spoken Query Answering by Jie Chi, Maureen de Seyssel, and Natalie Schluter
TiC-LM: A Multi-12 months Benchmark for Continuous Pretraining of Language Fashions by Jeffrey Li, Mohammadreza Armandpour, Iman Mirzadeh, Sachin Mehta, Vaishaal Shankar, Raviteja Vemulapalli, Samy Bengio, Oncel Tuzel, Mehrdad Farajtabar, Hadi Pouransari, and Fartash Faghri
Tulu 3: Pushing Frontiers in Open Language Mannequin Submit-Coaching by Nathan Lambert (Allen Institute for AI), Jacob Morrison (Allen Institute for AI), Valentina Pyatkin (Allen Institute for AI, College of Washington), Shengyi Huang (Allen Institute for AI), Hamish Ivison (Allen Institute for AI, College of Washington), Faeze Brahman (Allen Institute for AI), Lester James V. Miranda (Allen Institute for AI), Alisa Liu (College of Washington), Nouha Dziri (Allen Institute for AI), Xinxi Lyu (Allen Institute for AI), Yuling Gu (Allen Institute for AI), Saumya Malik (Allen Institute for AI), Victoria Graf (College of Washington), Jena D. Hwang (Allen Institute for AI), Jiangjiang Yang (Allen Institute for AI), Ronan Le Bras (Allen Institute for AI), Oyvind Tafjord (Allen Institute for AI), Chris Wilhelm (Allen Institute for AI), Luca Soldaini (Allen Institute for AI), Noah A. Smith (Allen Institute for AI, College of Washington), Yizhong Wang (Allen Institute for AI, College of Washington), Pradeep Dasigi (Allen Institute for AI), and Hannaneh Hajishirzi (Allen Institute for AI, College of Washington)
ValueCompass: A Framework for Measuring Contextual Worth Alignment Between Human and LLMs by Hua Shen (College of Washington), Tiffany Knearem (Google), Reshmi Ghosh (Microsoft), Yu-Ju Yang (College of Illinois at Urbana-Champaign), Tanushree Mitra (College of Washington), Nicholas Clark (College of Washington), and Yun Huang (College of Illinois at Urbana-Champaign)
Different Sources
Agent S: an open agentic framework that makes use of computer systems like a human, Xin Eric Wang (UC Santa Cruz)
DASB Benchmark, Mirco Ravanelli (Concordia College, Mila – Quebec AI Institute, Université de Montréal)
Acknowledgments
Many individuals contributed to this workshop together with Irina Belousova, Samy Bengio, Pete Boothroyd, Meng Cao, Kevin Chen, Jie Chi, Fartash Faghri, Tatiana Likhomanenko, Rin Metcalf, Stephen Pulman, Rita Ramos, Natalie Schluter, David Q. Solar, Raviteja Vemulapalli, David Winarsky, and Yizhe Zhang.
