Author: "Dixon, Lucas" - Searchworks@Jio Institute Digital Library Search Results

Your search keyword '"Dixon, Lucas"' showing total 198 results

Start Over Author "Dixon, Lucas"

198 results on '"Dixon, Lucas"'

1. Gemma 2: Improving Open Language Models at a Practical Size

Author: Gemma Team, Riviere, Morgane, Pathak, Shreya, Sessa, Pier Giuseppe, Hardin, Cassidy, Bhupatiraju, Surya, Hussenot, Léonard, Mesnard, Thomas, Shahriari, Bobak, Ramé, Alexandre, Ferret, Johan, Liu, Peter, Tafti, Pouya, Friesen, Abe, Casbon, Michelle, Ramos, Sabela, Kumar, Ravin, Lan, Charline Le, Jerome, Sammy, Tsitsulin, Anton, Vieillard, Nino, Stanczyk, Piotr, Girgin, Sertan, Momchev, Nikola, Hoffman, Matt, Thakoor, Shantanu, Grill, Jean-Bastien, Neyshabur, Behnam, Bachem, Olivier, Walton, Alanna, Severyn, Aliaksei, Parrish, Alicia, Ahmad, Aliya, Hutchison, Allen, Abdagic, Alvin, Carl, Amanda, Shen, Amy, Brock, Andy, Coenen, Andy, Laforge, Anthony, Paterson, Antonia, Bastian, Ben, Piot, Bilal, Wu, Bo, Royal, Brandon, Chen, Charlie, Kumar, Chintu, Perry, Chris, Welty, Chris, Choquette-Choo, Christopher A., Sinopalnikov, Danila, Weinberger, David, Vijaykumar, Dimple, Rogozińska, Dominika, Herbison, Dustin, Bandy, Elisa, Wang, Emma, Noland, Eric, Moreira, Erica, Senter, Evan, Eltyshev, Evgenii, Visin, Francesco, Rasskin, Gabriel, Wei, Gary, Cameron, Glenn, Martins, Gus, Hashemi, Hadi, Klimczak-Plucińska, Hanna, Batra, Harleen, Dhand, Harsh, Nardini, Ivan, Mein, Jacinda, Zhou, Jack, Svensson, James, Stanway, Jeff, Chan, Jetha, Zhou, Jin Peng, Carrasqueira, Joana, Iljazi, Joana, Becker, Jocelyn, Fernandez, Joe, van Amersfoort, Joost, Gordon, Josh, Lipschultz, Josh, Newlan, Josh, Ji, Ju-yeong, Mohamed, Kareem, Badola, Kartikeya, Black, Kat, Millican, Katie, McDonell, Keelin, Nguyen, Kelvin, Sodhia, Kiranbir, Greene, Kish, Sjoesund, Lars Lowe, Usui, Lauren, Sifre, Laurent, Heuermann, Lena, Lago, Leticia, McNealus, Lilly, Soares, Livio Baldini, Kilpatrick, Logan, Dixon, Lucas, Martins, Luciano, Reid, Machel, Singh, Manvinder, Iverson, Mark, Görner, Martin, Velloso, Mat, Wirth, Mateo, Davidow, Matt, Miller, Matt, Rahtz, Matthew, Watson, Matthew, Risdal, Meg, Kazemi, Mehran, Moynihan, Michael, Zhang, Ming, Kahng, Minsuk, Park, Minwoo, Rahman, Mofi, Khatwani, Mohit, Dao, Natalie, Bardoliwalla, Nenshad, Devanathan, Nesh, Dumai, Neta, Chauhan, Nilay, Wahltinez, Oscar, Botarda, Pankil, Barnes, Parker, Barham, Paul, Michel, Paul, Jin, Pengchong, Georgiev, Petko, Culliton, Phil, Kuppala, Pradeep, Comanescu, Ramona, Merhej, Ramona, Jana, Reena, Rokni, Reza Ardeshir, Agarwal, Rishabh, Mullins, Ryan, Saadat, Samaneh, Carthy, Sara Mc, Perrin, Sarah, Arnold, Sébastien M. R., Krause, Sebastian, Dai, Shengyang, Garg, Shruti, Sheth, Shruti, Ronstrom, Sue, Chan, Susan, Jordan, Timothy, Yu, Ting, Eccles, Tom, Hennigan, Tom, Kocisky, Tomas, Doshi, Tulsee, Jain, Vihan, Yadav, Vikas, Meshram, Vilobh, Dharmadhikari, Vishal, Barkley, Warren, Wei, Wei, Ye, Wenming, Han, Woohyun, Kwon, Woosuk, Xu, Xiang, Shen, Zhe, Gong, Zhitao, Wei, Zichuan, Cotruta, Victor, Kirk, Phoebe, Rao, Anand, Giang, Minh, Peran, Ludovic, Warkentin, Tris, Collins, Eli, Barral, Joelle, Ghahramani, Zoubin, Hadsell, Raia, Sculley, D., Banks, Jeanine, Dragan, Anca, Petrov, Slav, Vinyals, Oriol, Dean, Jeff, Hassabis, Demis, Kavukcuoglu, Koray, Farabet, Clement, Buchatskaya, Elena, Borgeaud, Sebastian, Fiedel, Noah, Joulin, Armand, Kenealy, Kathleen, Dadashi, Robert, and Andreev, Alek
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence
Abstract: In this work, we introduce Gemma 2, a new addition to the Gemma family of lightweight, state-of-the-art open models, ranging in scale from 2 billion to 27 billion parameters. In this new version, we apply several known technical modifications to the Transformer architecture, such as interleaving local-global attentions (Beltagy et al., 2020a) and group-query attention (Ainslie et al., 2023). We also train the 2B and 9B models with knowledge distillation (Hinton et al., 2015) instead of next token prediction. The resulting models deliver the best performance for their size, and even offer competitive alternatives to models that are 2-3 times bigger. We release all our models to the community.
Published: 2024

2. Who's asking? User personas and the mechanics of latent misalignment

Author: Ghandeharioun, Asma, Yuan, Ann, Guerard, Marius, Reif, Emily, Lepori, Michael A., and Dixon, Lucas
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence
Abstract: Despite investments in improving model safety, studies show that misaligned capabilities remain latent in safety-tuned models. In this work, we shed light on the mechanics of this phenomenon. First, we show that even when model generations are safe, harmful content can persist in hidden representations and can be extracted by decoding from earlier layers. Then, we show that whether the model divulges such content depends significantly on its perception of who it is talking to, which we refer to as user persona. In fact, we find manipulating user persona to be even more effective for eliciting harmful content than direct attempts to control model refusal. We study both natural language prompting and activation steering as control methods and show that activation steering is significantly more effective at bypassing safety filters. We investigate why certain personas break model safeguards and find that they enable the model to form more charitable interpretations of otherwise dangerous queries. Finally, we show we can predict a persona's effect on refusal given only the geometry of its steering vector.
Published: 2024

3. Interactive Prompt Debugging with Sequence Salience

Author: Tenney, Ian, Mullins, Ryan, Du, Bin, Pandya, Shree, Kahng, Minsuk, and Dixon, Lucas
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Human-Computer Interaction, Computer Science - Machine Learning
Abstract: We present Sequence Salience, a visual tool for interactive prompt debugging with input salience methods. Sequence Salience builds on widely used salience methods for text classification and single-token prediction, and extends this to a system tailored for debugging complex LLM prompts. Our system is well-suited for long texts, and expands on previous work by 1) providing controllable aggregation of token-level salience to the word, sentence, or paragraph level, making salience over long inputs tractable; and 2) supporting rapid iteration where practitioners can act on salience results, refine prompts, and run salience on the new output. We include case studies showing how Sequence Salience can help practitioners work with several complex prompting strategies, including few-shot, chain-of-thought, and constitutional principles. Sequence Salience is built on the Learning Interpretability Tool, an open-source platform for ML model visualizations, and code, notebooks, and tutorials are available at http://goo.gle/sequence-salience.
Published: 2024

4. 'We Need Structured Output': Towards User-centered Constraints on Large Language Model Output

Author: Liu, Michael Xieyang, Liu, Frederick, Fiannaca, Alexander J., Koo, Terry, Dixon, Lucas, Terry, Michael, and Cai, Carrie J.
Subjects: Computer Science - Human-Computer Interaction
Abstract: Large language models can produce creative and diverse responses. However, to integrate them into current developer workflows, it is essential to constrain their outputs to follow specific formats or standards. In this work, we surveyed 51 experienced industry professionals to understand the range of scenarios and motivations driving the need for output constraints from a user-centered perspective. We identified 134 concrete use cases for constraints at two levels: low-level, which ensures the output adhere to a structured format and an appropriate length, and high-level, which requires the output to follow semantic and stylistic guidelines without hallucination. Critically, applying output constraints could not only streamline the currently repetitive process of developing, testing, and integrating LLM prompts for developers, but also enhance the user experience of LLM-powered features and applications. We conclude with a discussion on user preferences and needs towards articulating intended constraints for LLMs, alongside an initial design for a constraint prototyping tool.
Published: 2024
Full Text: View/download PDF

5. Parameter Efficient Reinforcement Learning from Human Feedback

Author: Sidahmed, Hakim, Phatale, Samrat, Hutcheson, Alex, Lin, Zhuonan, Chen, Zhang, Yu, Zac, Jin, Jarvis, Chaudhary, Simral, Komarytsia, Roman, Ahlheim, Christiane, Zhu, Yonghao, Li, Bowen, Ganesh, Saravanan, Byrne, Bill, Hoffmann, Jessica, Mansoor, Hassan, Li, Wei, Rastogi, Abhinav, and Dixon, Lucas
Subjects: Computer Science - Machine Learning, Computer Science - Artificial Intelligence, Computer Science - Computation and Language
Abstract: While Reinforcement Learning from Human Feedback (RLHF) effectively aligns pretrained Large Language and Vision-Language Models (LLMs, and VLMs) with human preferences, its computational cost and complexity hamper its wider adoption. To alleviate some of the computational burden of fine-tuning, parameter efficient methods, like LoRA were introduced. In this work, we empirically evaluate the setup of Parameter Efficient Reinforcement Learning from Human Feedback (PE-RLHF) that leverages LoRA fine-tuning for Reward Modeling, and Reinforcement Learning. We benchmark the PE-RLHF setup on six diverse datasets spanning summarization, harmless/helpful response generation, UI automation, and visual question answering in terms of effectiveness of the trained models, and the training resources required. Our findings show, for the first time, that PE-RLHF achieves comparable performance to RLHF, while significantly reducing training time (up to 90% faster for reward models, and 30% faster for RL), and memory footprint (up to 50% reduction for reward models, and 27% for RL). We provide comprehensive ablations across LoRA ranks, and model sizes for both reward modeling and reinforcement learning. By mitigating the computational burden associated with RLHF, we push for a broader adoption of PE-RLHF as an alignment technique for LLMs and VLMs.
Published: 2024

6. Gemma: Open Models Based on Gemini Research and Technology

Author: Gemma Team, Mesnard, Thomas, Hardin, Cassidy, Dadashi, Robert, Bhupatiraju, Surya, Pathak, Shreya, Sifre, Laurent, Rivière, Morgane, Kale, Mihir Sanjay, Love, Juliette, Tafti, Pouya, Hussenot, Léonard, Sessa, Pier Giuseppe, Chowdhery, Aakanksha, Roberts, Adam, Barua, Aditya, Botev, Alex, Castro-Ros, Alex, Slone, Ambrose, Héliou, Amélie, Tacchetti, Andrea, Bulanova, Anna, Paterson, Antonia, Tsai, Beth, Shahriari, Bobak, Lan, Charline Le, Choquette-Choo, Christopher A., Crepy, Clément, Cer, Daniel, Ippolito, Daphne, Reid, David, Buchatskaya, Elena, Ni, Eric, Noland, Eric, Yan, Geng, Tucker, George, Muraru, George-Christian, Rozhdestvenskiy, Grigory, Michalewski, Henryk, Tenney, Ian, Grishchenko, Ivan, Austin, Jacob, Keeling, James, Labanowski, Jane, Lespiau, Jean-Baptiste, Stanway, Jeff, Brennan, Jenny, Chen, Jeremy, Ferret, Johan, Chiu, Justin, Mao-Jones, Justin, Lee, Katherine, Yu, Kathy, Millican, Katie, Sjoesund, Lars Lowe, Lee, Lisa, Dixon, Lucas, Reid, Machel, Mikuła, Maciej, Wirth, Mateo, Sharman, Michael, Chinaev, Nikolai, Thain, Nithum, Bachem, Olivier, Chang, Oscar, Wahltinez, Oscar, Bailey, Paige, Michel, Paul, Yotov, Petko, Chaabouni, Rahma, Comanescu, Ramona, Jana, Reena, Anil, Rohan, McIlroy, Ross, Liu, Ruibo, Mullins, Ryan, Smith, Samuel L, Borgeaud, Sebastian, Girgin, Sertan, Douglas, Sholto, Pandya, Shree, Shakeri, Siamak, De, Soham, Klimenko, Ted, Hennigan, Tom, Feinberg, Vlad, Stokowiec, Wojciech, Chen, Yu-hui, Ahmed, Zafarali, Gong, Zhitao, Warkentin, Tris, Peran, Ludovic, Giang, Minh, Farabet, Clément, Vinyals, Oriol, Dean, Jeff, Kavukcuoglu, Koray, Hassabis, Demis, Ghahramani, Zoubin, Eck, Douglas, Barral, Joelle, Pereira, Fernando, Collins, Eli, Joulin, Armand, Fiedel, Noah, Senter, Evan, Andreev, Alek, and Kenealy, Kathleen
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence
Abstract: This work introduces Gemma, a family of lightweight, state-of-the art open models built from the research and technology used to create Gemini models. Gemma models demonstrate strong performance across academic benchmarks for language understanding, reasoning, and safety. We release two sizes of models (2 billion and 7 billion parameters), and provide both pretrained and fine-tuned checkpoints. Gemma outperforms similarly sized open models on 11 out of 18 text-based tasks, and we present comprehensive evaluations of safety and responsibility aspects of the models, alongside a detailed description of model development. We believe the responsible release of LLMs is critical for improving the safety of frontier models, and for enabling the next wave of LLM innovations.
Published: 2024

7. Detecting Hallucination and Coverage Errors in Retrieval Augmented Generation for Controversial Topics

Author: Chang, Tyler A., Tomanek, Katrin, Hoffmann, Jessica, Thain, Nithum, van Liemt, Erin, Meier-Hellstern, Kathleen, and Dixon, Lucas
Subjects: Computer Science - Computation and Language
Abstract: We explore a strategy to handle controversial topics in LLM-based chatbots based on Wikipedia's Neutral Point of View (NPOV) principle: acknowledge the absence of a single true answer and surface multiple perspectives. We frame this as retrieval augmented generation, where perspectives are retrieved from a knowledge base and the LLM is tasked with generating a fluent and faithful response from the given perspectives. As a starting point, we use a deterministic retrieval system and then focus on common LLM failure modes that arise during this approach to text generation, namely hallucination and coverage errors. We propose and evaluate three methods to detect such errors based on (1) word-overlap, (2) salience, and (3) LLM-based classifiers. Our results demonstrate that LLM-based classifiers, even when trained only on synthetic errors, achieve high error detection performance, with ROC AUC scores of 95.3% for hallucination and 90.5% for coverage error detection on unambiguous error cases. We show that when no training data is available, our other methods still yield good results on hallucination (84.0%) and coverage error (85.2%) detection., Comment: Accepted at LREC-COLING 2024
Published: 2024

8. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context

Author: Gemini Team, Georgiev, Petko, Lei, Ving Ian, Burnell, Ryan, Bai, Libin, Gulati, Anmol, Tanzer, Garrett, Vincent, Damien, Pan, Zhufeng, Wang, Shibo, Mariooryad, Soroosh, Ding, Yifan, Geng, Xinyang, Alcober, Fred, Frostig, Roy, Omernick, Mark, Walker, Lexi, Paduraru, Cosmin, Sorokin, Christina, Tacchetti, Andrea, Gaffney, Colin, Daruki, Samira, Sercinoglu, Olcan, Gleicher, Zach, Love, Juliette, Voigtlaender, Paul, Jain, Rohan, Surita, Gabriela, Mohamed, Kareem, Blevins, Rory, Ahn, Junwhan, Zhu, Tao, Kawintiranon, Kornraphop, Firat, Orhan, Gu, Yiming, Zhang, Yujing, Rahtz, Matthew, Faruqui, Manaal, Clay, Natalie, Gilmer, Justin, Co-Reyes, JD, Penchev, Ivo, Zhu, Rui, Morioka, Nobuyuki, Hui, Kevin, Haridasan, Krishna, Campos, Victor, Mahdieh, Mahdis, Guo, Mandy, Hassan, Samer, Kilgour, Kevin, Vezer, Arpi, Cheng, Heng-Tze, de Liedekerke, Raoul, Goyal, Siddharth, Barham, Paul, Strouse, DJ, Noury, Seb, Adler, Jonas, Sundararajan, Mukund, Vikram, Sharad, Lepikhin, Dmitry, Paganini, Michela, Garcia, Xavier, Yang, Fan, Valter, Dasha, Trebacz, Maja, Vodrahalli, Kiran, Asawaroengchai, Chulayuth, Ring, Roman, Kalb, Norbert, Soares, Livio Baldini, Brahma, Siddhartha, Steiner, David, Yu, Tianhe, Mentzer, Fabian, He, Antoine, Gonzalez, Lucas, Xu, Bibo, Kaufman, Raphael Lopez, Shafey, Laurent El, Oh, Junhyuk, Hennigan, Tom, Driessche, George van den, Odoom, Seth, Lucic, Mario, Roelofs, Becca, Lall, Sid, Marathe, Amit, Chan, Betty, Ontanon, Santiago, He, Luheng, Teplyashin, Denis, Lai, Jonathan, Crone, Phil, Damoc, Bogdan, Ho, Lewis, Riedel, Sebastian, Lenc, Karel, Yeh, Chih-Kuan, Chowdhery, Aakanksha, Xu, Yang, Kazemi, Mehran, Amid, Ehsan, Petrushkina, Anastasia, Swersky, Kevin, Khodaei, Ali, Chen, Gowoon, Larkin, Chris, Pinto, Mario, Yan, Geng, Badia, Adria Puigdomenech, Patil, Piyush, Hansen, Steven, Orr, Dave, Arnold, Sebastien M. R., Grimstad, Jordan, Dai, Andrew, Douglas, Sholto, Sinha, Rishika, Yadav, Vikas, Chen, Xi, Gribovskaya, Elena, Austin, Jacob, Zhao, Jeffrey, Patel, Kaushal, Komarek, Paul, Austin, Sophia, Borgeaud, Sebastian, Friso, Linda, Goyal, Abhimanyu, Caine, Ben, Cao, Kris, Chung, Da-Woon, Lamm, Matthew, Barth-Maron, Gabe, Kagohara, Thais, Olszewska, Kate, Chen, Mia, Shivakumar, Kaushik, Agarwal, Rishabh, Godhia, Harshal, Rajwar, Ravi, Snaider, Javier, Dotiwalla, Xerxes, Liu, Yuan, Barua, Aditya, Ungureanu, Victor, Zhang, Yuan, Batsaikhan, Bat-Orgil, Wirth, Mateo, Qin, James, Danihelka, Ivo, Doshi, Tulsee, Chadwick, Martin, Chen, Jilin, Jain, Sanil, Le, Quoc, Kar, Arjun, Gurumurthy, Madhu, Li, Cheng, Sang, Ruoxin, Liu, Fangyu, Lamprou, Lampros, Munoz, Rich, Lintz, Nathan, Mehta, Harsh, Howard, Heidi, Reynolds, Malcolm, Aroyo, Lora, Wang, Quan, Blanco, Lorenzo, Cassirer, Albin, Griffith, Jordan, Das, Dipanjan, Lee, Stephan, Sygnowski, Jakub, Fisher, Zach, Besley, James, Powell, Richard, Ahmed, Zafarali, Paulus, Dominik, Reitter, David, Borsos, Zalan, Joshi, Rishabh, Pope, Aedan, Hand, Steven, Selo, Vittorio, Jain, Vihan, Sethi, Nikhil, Goel, Megha, Makino, Takaki, May, Rhys, Yang, Zhen, Schalkwyk, Johan, Butterfield, Christina, Hauth, Anja, Goldin, Alex, Hawkins, Will, Senter, Evan, Brin, Sergey, Woodman, Oliver, Ritter, Marvin, Noland, Eric, Giang, Minh, Bolina, Vijay, Lee, Lisa, Blyth, Tim, Mackinnon, Ian, Reid, Machel, Sarvana, Obaid, Silver, David, Chen, Alexander, Wang, Lily, Maggiore, Loren, Chang, Oscar, Attaluri, Nithya, Thornton, Gregory, Chiu, Chung-Cheng, Bunyan, Oskar, Levine, Nir, Chung, Timothy, Eltyshev, Evgenii, Si, Xiance, Lillicrap, Timothy, Brady, Demetra, Aggarwal, Vaibhav, Wu, Boxi, Xu, Yuanzhong, McIlroy, Ross, Badola, Kartikeya, Sandhu, Paramjit, Moreira, Erica, Stokowiec, Wojciech, Hemsley, Ross, Li, Dong, Tudor, Alex, Shyam, Pranav, Rahimtoroghi, Elahe, Haykal, Salem, Sprechmann, Pablo, Zhou, Xiang, Mincu, Diana, Li, Yujia, Addanki, Ravi, Krishna, Kalpesh, Wu, Xiao, Frechette, Alexandre, Eyal, Matan, Dafoe, Allan, Lacey, Dave, Whang, Jay, Avrahami, Thi, Zhang, Ye, Taropa, Emanuel, Lin, Hanzhao, Toyama, Daniel, Rutherford, Eliza, Sano, Motoki, Choe, HyunJeong, Tomala, Alex, Safranek-Shrader, Chalence, Kassner, Nora, Pajarskas, Mantas, Harvey, Matt, Sechrist, Sean, Fortunato, Meire, Lyu, Christina, Elsayed, Gamaleldin, Kuang, Chenkai, Lottes, James, Chu, Eric, Jia, Chao, Chen, Chih-Wei, Humphreys, Peter, Baumli, Kate, Tao, Connie, Samuel, Rajkumar, Santos, Cicero Nogueira dos, Andreassen, Anders, Rakićević, Nemanja, Grewe, Dominik, Kumar, Aviral, Winkler, Stephanie, Caton, Jonathan, Brock, Andrew, Dalmia, Sid, Sheahan, Hannah, Barr, Iain, Miao, Yingjie, Natsev, Paul, Devlin, Jacob, Behbahani, Feryal, Prost, Flavien, Sun, Yanhua, Myaskovsky, Artiom, Pillai, Thanumalayan Sankaranarayana, Hurt, Dan, Lazaridou, Angeliki, Xiong, Xi, Zheng, Ce, Pardo, Fabio, Li, Xiaowei, Horgan, Dan, Stanton, Joe, Ambar, Moran, Xia, Fei, Lince, Alejandro, Wang, Mingqiu, Mustafa, Basil, Webson, Albert, Lee, Hyo, Anil, Rohan, Wicke, Martin, Dozat, Timothy, Sinha, Abhishek, Piqueras, Enrique, Dabir, Elahe, Upadhyay, Shyam, Boral, Anudhyan, Hendricks, Lisa Anne, Fry, Corey, Djolonga, Josip, Su, Yi, Walker, Jake, Labanowski, Jane, Huang, Ronny, Misra, Vedant, Chen, Jeremy, Skerry-Ryan, RJ, Singh, Avi, Rijhwani, Shruti, Yu, Dian, Castro-Ros, Alex, Changpinyo, Beer, Datta, Romina, Bagri, Sumit, Hrafnkelsson, Arnar Mar, Maggioni, Marcello, Zheng, Daniel, Sulsky, Yury, Hou, Shaobo, Paine, Tom Le, Yang, Antoine, Riesa, Jason, Rogozinska, Dominika, Marcus, Dror, Badawy, Dalia El, Zhang, Qiao, Wang, Luyu, Miller, Helen, Greer, Jeremy, Sjos, Lars Lowe, Nova, Azade, Zen, Heiga, Chaabouni, Rahma, Rosca, Mihaela, Jiang, Jiepu, Chen, Charlie, Liu, Ruibo, Sainath, Tara, Krikun, Maxim, Polozov, Alex, Lespiau, Jean-Baptiste, Newlan, Josh, Cankara, Zeyncep, Kwak, Soo, Xu, Yunhan, Chen, Phil, Coenen, Andy, Meyer, Clemens, Tsihlas, Katerina, Ma, Ada, Gottweis, Juraj, Xing, Jinwei, Gu, Chenjie, Miao, Jin, Frank, Christian, Cankara, Zeynep, Ganapathy, Sanjay, Dasgupta, Ishita, Hughes-Fitt, Steph, Chen, Heng, Reid, David, Rong, Keran, Fan, Hongmin, van Amersfoort, Joost, Zhuang, Vincent, Cohen, Aaron, Gu, Shixiang Shane, Mohananey, Anhad, Ilic, Anastasija, Tobin, Taylor, Wieting, John, Bortsova, Anna, Thacker, Phoebe, Wang, Emma, Caveness, Emily, Chiu, Justin, Sezener, Eren, Kaskasoli, Alex, Baker, Steven, Millican, Katie, Elhawaty, Mohamed, Aisopos, Kostas, Lebsack, Carl, Byrd, Nathan, Dai, Hanjun, Jia, Wenhao, Wiethoff, Matthew, Davoodi, Elnaz, Weston, Albert, Yagati, Lakshman, Ahuja, Arun, Gao, Isabel, Pundak, Golan, Zhang, Susan, Azzam, Michael, Sim, Khe Chai, Caelles, Sergi, Keeling, James, Sharma, Abhanshu, Swing, Andy, Li, YaGuang, Liu, Chenxi, Bostock, Carrie Grimes, Bansal, Yamini, Nado, Zachary, Anand, Ankesh, Lipschultz, Josh, Karmarkar, Abhijit, Proleev, Lev, Ittycheriah, Abe, Yeganeh, Soheil Hassas, Polovets, George, Faust, Aleksandra, Sun, Jiao, Rrustemi, Alban, Li, Pen, Shivanna, Rakesh, Liu, Jeremiah, Welty, Chris, Lebron, Federico, Baddepudi, Anirudh, Krause, Sebastian, Parisotto, Emilio, Soricut, Radu, Xu, Zheng, Bloxwich, Dawn, Johnson, Melvin, Neyshabur, Behnam, Mao-Jones, Justin, Wang, Renshen, Ramasesh, Vinay, Abbas, Zaheer, Guez, Arthur, Segal, Constant, Nguyen, Duc Dung, Svensson, James, Hou, Le, York, Sarah, Milan, Kieran, Bridgers, Sophie, Gworek, Wiktor, Tagliasacchi, Marco, Lee-Thorp, James, Chang, Michael, Guseynov, Alexey, Hartman, Ale Jakse, Kwong, Michael, Zhao, Ruizhe, Kashem, Sheleem, Cole, Elizabeth, Miech, Antoine, Tanburn, Richard, Phuong, Mary, Pavetic, Filip, Cevey, Sebastien, Comanescu, Ramona, Ives, Richard, Yang, Sherry, Du, Cosmo, Li, Bo, Zhang, Zizhao, Iinuma, Mariko, Hu, Clara Huiyi, Roy, Aurko, Bijwadia, Shaan, Zhu, Zhenkai, Martins, Danilo, Saputro, Rachel, Gergely, Anita, Zheng, Steven, Jia, Dawei, Antonoglou, Ioannis, Sadovsky, Adam, Gu, Shane, Bi, Yingying, Andreev, Alek, Samangooei, Sina, Khan, Mina, Kocisky, Tomas, Filos, Angelos, Kumar, Chintu, Bishop, Colton, Yu, Adams, Hodkinson, Sarah, Mittal, Sid, Shah, Premal, Moufarek, Alexandre, Cheng, Yong, Bloniarz, Adam, Lee, Jaehoon, Pejman, Pedram, Michel, Paul, Spencer, Stephen, Feinberg, Vladimir, Xiong, Xuehan, Savinov, Nikolay, Smith, Charlotte, Shakeri, Siamak, Tran, Dustin, Chesus, Mary, Bohnet, Bernd, Tucker, George, von Glehn, Tamara, Muir, Carrie, Mao, Yiran, Kazawa, Hideto, Slone, Ambrose, Soparkar, Kedar, Shrivastava, Disha, Cobon-Kerr, James, Sharman, Michael, Pavagadhi, Jay, Araya, Carlos, Misiunas, Karolis, Ghelani, Nimesh, Laskin, Michael, Barker, David, Li, Qiujia, Briukhov, Anton, Houlsby, Neil, Glaese, Mia, Lakshminarayanan, Balaji, Schucher, Nathan, Tang, Yunhao, Collins, Eli, Lim, Hyeontaek, Feng, Fangxiaoyu, Recasens, Adria, Lai, Guangda, Magni, Alberto, De Cao, Nicola, Siddhant, Aditya, Ashwood, Zoe, Orbay, Jordi, Dehghani, Mostafa, Brennan, Jenny, He, Yifan, Xu, Kelvin, Gao, Yang, Saroufim, Carl, Molloy, James, Wu, Xinyi, Arnold, Seb, Chang, Solomon, Schrittwieser, Julian, Buchatskaya, Elena, Radpour, Soroush, Polacek, Martin, Giordano, Skye, Bapna, Ankur, Tokumine, Simon, Hellendoorn, Vincent, Sottiaux, Thibault, Cogan, Sarah, Severyn, Aliaksei, Saleh, Mohammad, Thakoor, Shantanu, Shefey, Laurent, Qiao, Siyuan, Gaba, Meenu, Chang, Shuo-yiin, Swanson, Craig, Zhang, Biao, Lee, Benjamin, Rubenstein, Paul Kishan, Song, Gan, Kwiatkowski, Tom, Koop, Anna, Kannan, Ajay, Kao, David, Schuh, Parker, Stjerngren, Axel, Ghiasi, Golnaz, Gibson, Gena, Vilnis, Luke, Yuan, Ye, Ferreira, Felipe Tiengo, Kamath, Aishwarya, Klimenko, Ted, Franko, Ken, Xiao, Kefan, Bhattacharya, Indro, Patel, Miteyan, Wang, Rui, Morris, Alex, Strudel, Robin, Sharma, Vivek, Choy, Peter, Hashemi, Sayed Hadi, Landon, Jessica, Finkelstein, Mara, Jhakra, Priya, Frye, Justin, Barnes, Megan, Mauger, Matthew, Daun, Dennis, Baatarsukh, Khuslen, Tung, Matthew, Farhan, Wael, Michalewski, Henryk, Viola, Fabio, Quitry, Felix de Chaumont, Lan, Charline Le, Hudson, Tom, Wang, Qingze, Fischer, Felix, Zheng, Ivy, White, Elspeth, Dragan, Anca, Alayrac, Jean-baptiste, Ni, Eric, Pritzel, Alexander, Iwanicki, Adam, Isard, Michael, Bulanova, Anna, Zilka, Lukas, Dyer, Ethan, Sachan, Devendra, Srinivasan, Srivatsan, Muckenhirn, Hannah, Cai, Honglong, Mandhane, Amol, Tariq, Mukarram, Rae, Jack W., Wang, Gary, Ayoub, Kareem, FitzGerald, Nicholas, Zhao, Yao, Han, Woohyun, Alberti, Chris, Garrette, Dan, Krishnakumar, Kashyap, Gimenez, Mai, Levskaya, Anselm, Sohn, Daniel, Matak, Josip, Iturrate, Inaki, Chang, Michael B., Xiang, Jackie, Cao, Yuan, Ranka, Nishant, Brown, Geoff, Hutter, Adrian, Mirrokni, Vahab, Chen, Nanxin, Yao, Kaisheng, Egyed, Zoltan, Galilee, Francois, Liechty, Tyler, Kallakuri, Praveen, Palmer, Evan, Ghemawat, Sanjay, Liu, Jasmine, Tao, David, Thornton, Chloe, Green, Tim, Jasarevic, Mimi, Lin, Sharon, Cotruta, Victor, Tan, Yi-Xuan, Fiedel, Noah, Yu, Hongkun, Chi, Ed, Neitz, Alexander, Heitkaemper, Jens, Sinha, Anu, Zhou, Denny, Sun, Yi, Kaed, Charbel, Hulse, Brice, Mishra, Swaroop, Georgaki, Maria, Kudugunta, Sneha, Farabet, Clement, Shafran, Izhak, Vlasic, Daniel, Tsitsulin, Anton, Ananthanarayanan, Rajagopal, Carin, Alen, Su, Guolong, Sun, Pei, V, Shashank, Carvajal, Gabriel, Broder, Josef, Comsa, Iulia, Repina, Alena, Wong, William, Chen, Warren Weilun, Hawkins, Peter, Filonov, Egor, Loher, Lucia, Hirnschall, Christoph, Wang, Weiyi, Ye, Jingchen, Burns, Andrea, Cate, Hardie, Wright, Diana Gage, Piccinini, Federico, Zhang, Lei, Lin, Chu-Cheng, Gog, Ionel, Kulizhskaya, Yana, Sreevatsa, Ashwin, Song, Shuang, Cobo, Luis C., Iyer, Anand, Tekur, Chetan, Garrido, Guillermo, Xiao, Zhuyun, Kemp, Rupert, Zheng, Huaixiu Steven, Li, Hui, Agarwal, Ananth, Ngani, Christel, Goshvadi, Kati, Santamaria-Fernandez, Rebeca, Fica, Wojciech, Chen, Xinyun, Gorgolewski, Chris, Sun, Sean, Garg, Roopal, Ye, Xinyu, Eslami, S. M. Ali, Hua, Nan, Simon, Jon, Joshi, Pratik, Kim, Yelin, Tenney, Ian, Potluri, Sahitya, Thiet, Lam Nguyen, Yuan, Quan, Luisier, Florian, Chronopoulou, Alexandra, Scellato, Salvatore, Srinivasan, Praveen, Chen, Minmin, Koverkathu, Vinod, Dalibard, Valentin, Xu, Yaming, Saeta, Brennan, Anderson, Keith, Sellam, Thibault, Fernando, Nick, Huot, Fantine, Jung, Junehyuk, Varadarajan, Mani, Quinn, Michael, Raul, Amit, Le, Maigo, Habalov, Ruslan, Clark, Jon, Jalan, Komal, Bullard, Kalesha, Singhal, Achintya, Luong, Thang, Wang, Boyu, Rajayogam, Sujeevan, Eisenschlos, Julian, Jia, Johnson, Finchelstein, Daniel, Yakubovich, Alex, Balle, Daniel, Fink, Michael, Agarwal, Sameer, Li, Jing, Dvijotham, Dj, Pal, Shalini, Kang, Kai, Konzelmann, Jaclyn, Beattie, Jennifer, Dousse, Olivier, Wu, Diane, Crocker, Remi, Elkind, Chen, Jonnalagadda, Siddhartha Reddy, Lee, Jong, Holtmann-Rice, Dan, Kallarackal, Krystal, Liu, Rosanne, Vnukov, Denis, Vats, Neera, Invernizzi, Luca, Jafari, Mohsen, Zhou, Huanjie, Taylor, Lilly, Prendki, Jennifer, Wu, Marcus, Eccles, Tom, Liu, Tianqi, Kopparapu, Kavya, Beaufays, Francoise, Angermueller, Christof, Marzoca, Andreea, Sarcar, Shourya, Dib, Hilal, Stanway, Jeff, Perbet, Frank, Trdin, Nejc, Sterneck, Rachel, Khorlin, Andrey, Li, Dinghua, Wu, Xihui, Goenka, Sonam, Madras, David, Goldshtein, Sasha, Gierke, Willi, Zhou, Tong, Liu, Yaxin, Liang, Yannie, White, Anais, Li, Yunjie, Singh, Shreya, Bahargam, Sanaz, Epstein, Mark, Basu, Sujoy, Lao, Li, Ozturel, Adnan, Crous, Carl, Zhai, Alex, Lu, Han, Tung, Zora, Gaur, Neeraj, Walton, Alanna, Dixon, Lucas, Zhang, Ming, Globerson, Amir, Uy, Grant, Bolt, Andrew, Wiles, Olivia, Nasr, Milad, Shumailov, Ilia, Selvi, Marco, Piccinno, Francesco, Aguilar, Ricardo, McCarthy, Sara, Khalman, Misha, Shukla, Mrinal, Galic, Vlado, Carpenter, John, Villela, Kevin, Zhang, Haibin, Richardson, Harry, Martens, James, Bosnjak, Matko, Belle, Shreyas Rammohan, Seibert, Jeff, Alnahlawi, Mahmoud, McWilliams, Brian, Singh, Sankalp, Louis, Annie, Ding, Wen, Popovici, Dan, Simicich, Lenin, Knight, Laura, Mehta, Pulkit, Gupta, Nishesh, Shi, Chongyang, Fatehi, Saaber, Mitrovic, Jovana, Grills, Alex, Pagadora, Joseph, Petrova, Dessie, Eisenbud, Danielle, Zhang, Zhishuai, Yates, Damion, Mittal, Bhavishya, Tripuraneni, Nilesh, Assael, Yannis, Brovelli, Thomas, Jain, Prateek, Velimirovic, Mihajlo, Akbulut, Canfer, Mu, Jiaqi, Macherey, Wolfgang, Kumar, Ravin, Xu, Jun, Qureshi, Haroon, Comanici, Gheorghe, Wiesner, Jeremy, Gong, Zhitao, Ruddock, Anton, Bauer, Matthias, Felt, Nick, GP, Anirudh, Arnab, Anurag, Zelle, Dustin, Rothfuss, Jonas, Rosgen, Bill, Shenoy, Ashish, Seybold, Bryan, Li, Xinjian, Mudigonda, Jayaram, Erdogan, Goker, Xia, Jiawei, Simsa, Jiri, Michi, Andrea, Yao, Yi, Yew, Christopher, Kan, Steven, Caswell, Isaac, Radebaugh, Carey, Elisseeff, Andre, Valenzuela, Pedro, McKinney, Kay, Paterson, Kim, Cui, Albert, Latorre-Chimoto, Eri, Kim, Solomon, Zeng, William, Durden, Ken, Ponnapalli, Priya, Sosea, Tiberiu, Choquette-Choo, Christopher A., Manyika, James, Robenek, Brona, Vashisht, Harsha, Pereira, Sebastien, Lam, Hoi, Velic, Marko, Owusu-Afriyie, Denese, Lee, Katherine, Bolukbasi, Tolga, Parrish, Alicia, Lu, Shawn, Park, Jane, Venkatraman, Balaji, Talbert, Alice, Rosique, Lambert, Cheng, Yuchung, Sozanschi, Andrei, Paszke, Adam, Kumar, Praveen, Austin, Jessica, Li, Lu, Salama, Khalid, Kim, Wooyeol, Dukkipati, Nandita, Baryshnikov, Anthony, Kaplanis, Christos, Sheng, XiangHai, Chervonyi, Yuri, Unlu, Caglar, Casas, Diego de Las, Askham, Harry, Tunyasuvunakool, Kathryn, Gimeno, Felix, Poder, Siim, Kwak, Chester, Miecnikowski, Matt, Dimitriev, Alek, Parisi, Aaron, Liu, Dangyi, Tsai, Tomy, Shevlane, Toby, Kouridi, Christina, Garmon, Drew, Goedeckemeyer, Adrian, Brown, Adam R., Vijayakumar, Anitha, Elqursh, Ali, Jazayeri, Sadegh, Huang, Jin, Carthy, Sara Mc, Hoover, Jay, Kim, Lucy, Kumar, Sandeep, Chen, Wei, Biles, Courtney, Bingham, Garrett, Rosen, Evan, Wang, Lisa, Tan, Qijun, Engel, David, Pongetti, Francesco, de Cesare, Dario, Hwang, Dongseong, Yu, Lily, Pullman, Jennifer, Narayanan, Srini, Levin, Kyle, Gopal, Siddharth, Li, Megan, Aharoni, Asaf, Trinh, Trieu, Lo, Jessica, Casagrande, Norman, Vij, Roopali, Matthey, Loic, Ramadhana, Bramandia, Matthews, Austin, Carey, CJ, Johnson, Matthew, Goranova, Kremena, Shah, Rohin, Ashraf, Shereen, Dasgupta, Kingshuk, Larsen, Rasmus, Wang, Yicheng, Vuyyuru, Manish Reddy, Jiang, Chong, Ijazi, Joana, Osawa, Kazuki, Smith, Celine, Boppana, Ramya Sree, Bilal, Taylan, Koizumi, Yuma, Xu, Ying, Altun, Yasemin, Shabat, Nir, Bariach, Ben, Korchemniy, Alex, Choo, Kiam, Ronneberger, Olaf, Iwuanyanwu, Chimezie, Zhao, Shubin, Soergel, David, Hsieh, Cho-Jui, Cai, Irene, Iqbal, Shariq, Sundermeyer, Martin, Chen, Zhe, Bursztein, Elie, Malaviya, Chaitanya, Biadsy, Fadi, Shroff, Prakash, Dhillon, Inderjit, Latkar, Tejasi, Dyer, Chris, Forbes, Hannah, Nicosia, Massimo, Nikolaev, Vitaly, Greene, Somer, Georgiev, Marin, Wang, Pidong, Martin, Nina, Sedghi, Hanie, Zhang, John, Banzal, Praseem, Fritz, Doug, Rao, Vikram, Wang, Xuezhi, Zhang, Jiageng, Patraucean, Viorica, Du, Dayou, Mordatch, Igor, Jurin, Ivan, Liu, Lewis, Dubey, Ayush, Mohan, Abhi, Nowakowski, Janek, Ion, Vlad-Doru, Wei, Nan, Tojo, Reiko, Raad, Maria Abi, Hudson, Drew A., Keshava, Vaishakh, Agrawal, Shubham, Ramirez, Kevin, Wu, Zhichun, Nguyen, Hoang, Liu, Ji, Sewak, Madhavi, Petrini, Bryce, Choi, DongHyun, Philips, Ivan, Wang, Ziyue, Bica, Ioana, Garg, Ankush, Wilkiewicz, Jarek, Agrawal, Priyanka, Guo, Danhao, Xue, Emily, Shaik, Naseer, Leach, Andrew, Khan, Sadh MNM, Wiesinger, Julia, Jerome, Sammy, Chakladar, Abhishek, Wang, Alek Wenjiao, Ornduff, Tina, Abu, Folake, Ghaffarkhah, Alireza, Wainwright, Marcus, Cortes, Mario, Liu, Frederick, Maynez, Joshua, Terzis, Andreas, Samangouei, Pouya, Mansour, Riham, Kępa, Tomasz, Aubet, François-Xavier, Algymr, Anton, Banica, Dan, Weisz, Agoston, Orban, Andras, Senges, Alexandre, Andrejczuk, Ewa, Geller, Mark, Santo, Niccolo Dal, Anklin, Valentin, Merey, Majd Al, Baeuml, Martin, Strohman, Trevor, Bai, Junwen, Petrov, Slav, Wu, Yonghui, Hassabis, Demis, Kavukcuoglu, Koray, Dean, Jeffrey, and Vinyals, Oriol
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence
Abstract: In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio. The family includes two new models: (1) an updated Gemini 1.5 Pro, which exceeds the February version on the great majority of capabilities and benchmarks; (2) Gemini 1.5 Flash, a more lightweight variant designed for efficiency with minimal regression in quality. Gemini 1.5 models achieve near-perfect recall on long-context retrieval tasks across modalities, improve the state-of-the-art in long-document QA, long-video QA and long-context ASR, and match or surpass Gemini 1.0 Ultra's state-of-the-art performance across a broad set of benchmarks. Studying the limits of Gemini 1.5's long-context ability, we find continued improvement in next-token prediction and near-perfect retrieval (>99%) up to at least 10M tokens, a generational leap over existing models such as Claude 3.0 (200k) and GPT-4 Turbo (128k). Finally, we highlight real-world use cases, such as Gemini 1.5 collaborating with professionals on completing their tasks achieving 26 to 75% time savings across 10 different job categories, as well as surprising new capabilities of large language models at the frontier; when given a grammar manual for Kalamang, a language with fewer than 200 speakers worldwide, the model learns to translate English to Kalamang at a similar level to a person who learned from the same content.
Published: 2024

9. LLM Comparator: Visual Analytics for Side-by-Side Evaluation of Large Language Models

Author: Kahng, Minsuk, Tenney, Ian, Pushkarna, Mahima, Liu, Michael Xieyang, Wexler, James, Reif, Emily, Kallarackal, Krystal, Chang, Minsuk, Terry, Michael, and Dixon, Lucas
Subjects: Computer Science - Human-Computer Interaction, Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Machine Learning
Abstract: Automatic side-by-side evaluation has emerged as a promising approach to evaluating the quality of responses from large language models (LLMs). However, analyzing the results from this evaluation approach raises scalability and interpretability challenges. In this paper, we present LLM Comparator, a novel visual analytics tool for interactively analyzing results from automatic side-by-side evaluation. The tool supports interactive workflows for users to understand when and why a model performs better or worse than a baseline model, and how the responses from two models are qualitatively different. We iteratively designed and developed the tool by closely working with researchers and engineers at a large technology company. This paper details the user challenges we identified, the design and development of the tool, and an observational study with participants who regularly evaluate their models.
Published: 2024

10. Decoding-time Realignment of Language Models

Author: Liu, Tianlin, Guo, Shangmin, Bianco, Leonardo, Calandriello, Daniele, Berthet, Quentin, Llinares, Felipe, Hoffmann, Jessica, Dixon, Lucas, Valko, Michal, and Blondel, Mathieu
Subjects: Computer Science - Machine Learning, Computer Science - Artificial Intelligence, Computer Science - Computation and Language
Abstract: Aligning language models with human preferences is crucial for reducing errors and biases in these models. Alignment techniques, such as reinforcement learning from human feedback (RLHF), are typically cast as optimizing a tradeoff between human preference rewards and a proximity regularization term that encourages staying close to the unaligned model. Selecting an appropriate level of regularization is critical: insufficient regularization can lead to reduced model capabilities due to reward hacking, whereas excessive regularization hinders alignment. Traditional methods for finding the optimal regularization level require retraining multiple models with varying regularization strengths. This process, however, is resource-intensive, especially for large models. To address this challenge, we propose decoding-time realignment (DeRa), a simple method to explore and evaluate different regularization strengths in aligned models without retraining. DeRa enables control over the degree of alignment, allowing users to smoothly transition between unaligned and aligned models. It also enhances the efficiency of hyperparameter tuning by enabling the identification of effective regularization strengths using a validation dataset., Comment: In Proceedings of the 41st International Conference on Machine Learning (ICML 2024)
Published: 2024

11. Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Models

Author: Ghandeharioun, Asma, Caciularu, Avi, Pearce, Adam, Dixon, Lucas, and Geva, Mor
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Machine Learning
Abstract: Understanding the internal representations of large language models (LLMs) can help explain models' behavior and verify their alignment with human values. Given the capabilities of LLMs in generating human-understandable text, we propose leveraging the model itself to explain its internal representations in natural language. We introduce a framework called Patchscopes and show how it can be used to answer a wide range of questions about an LLM's computation. We show that many prior interpretability methods based on projecting representations into the vocabulary space and intervening on the LLM computation can be viewed as instances of this framework. Moreover, several of their shortcomings such as failure in inspecting early layers or lack of expressivity can be mitigated by Patchscopes. Beyond unifying prior inspection techniques, Patchscopes also opens up new possibilities such as using a more capable model to explain the representations of a smaller model, and multihop reasoning error correction., Comment: ICML 2024 (to appear)
Published: 2024

12. Gemini: A Family of Highly Capable Multimodal Models

Author: Gemini Team, Anil, Rohan, Borgeaud, Sebastian, Alayrac, Jean-Baptiste, Yu, Jiahui, Soricut, Radu, Schalkwyk, Johan, Dai, Andrew M., Hauth, Anja, Millican, Katie, Silver, David, Johnson, Melvin, Antonoglou, Ioannis, Schrittwieser, Julian, Glaese, Amelia, Chen, Jilin, Pitler, Emily, Lillicrap, Timothy, Lazaridou, Angeliki, Firat, Orhan, Molloy, James, Isard, Michael, Barham, Paul R., Hennigan, Tom, Lee, Benjamin, Viola, Fabio, Reynolds, Malcolm, Xu, Yuanzhong, Doherty, Ryan, Collins, Eli, Meyer, Clemens, Rutherford, Eliza, Moreira, Erica, Ayoub, Kareem, Goel, Megha, Krawczyk, Jack, Du, Cosmo, Chi, Ed, Cheng, Heng-Tze, Ni, Eric, Shah, Purvi, Kane, Patrick, Chan, Betty, Faruqui, Manaal, Severyn, Aliaksei, Lin, Hanzhao, Li, YaGuang, Cheng, Yong, Ittycheriah, Abe, Mahdieh, Mahdis, Chen, Mia, Sun, Pei, Tran, Dustin, Bagri, Sumit, Lakshminarayanan, Balaji, Liu, Jeremiah, Orban, Andras, Güra, Fabian, Zhou, Hao, Song, Xinying, Boffy, Aurelien, Ganapathy, Harish, Zheng, Steven, Choe, HyunJeong, Weisz, Ágoston, Zhu, Tao, Lu, Yifeng, Gopal, Siddharth, Kahn, Jarrod, Kula, Maciej, Pitman, Jeff, Shah, Rushin, Taropa, Emanuel, Merey, Majd Al, Baeuml, Martin, Chen, Zhifeng, Shafey, Laurent El, Zhang, Yujing, Sercinoglu, Olcan, Tucker, George, Piqueras, Enrique, Krikun, Maxim, Barr, Iain, Savinov, Nikolay, Danihelka, Ivo, Roelofs, Becca, White, Anaïs, Andreassen, Anders, von Glehn, Tamara, Yagati, Lakshman, Kazemi, Mehran, Gonzalez, Lucas, Khalman, Misha, Sygnowski, Jakub, Frechette, Alexandre, Smith, Charlotte, Culp, Laura, Proleev, Lev, Luan, Yi, Chen, Xi, Lottes, James, Schucher, Nathan, Lebron, Federico, Rrustemi, Alban, Clay, Natalie, Crone, Phil, Kocisky, Tomas, Zhao, Jeffrey, Perz, Bartek, Yu, Dian, Howard, Heidi, Bloniarz, Adam, Rae, Jack W., Lu, Han, Sifre, Laurent, Maggioni, Marcello, Alcober, Fred, Garrette, Dan, Barnes, Megan, Thakoor, Shantanu, Austin, Jacob, Barth-Maron, Gabriel, Wong, William, Joshi, Rishabh, Chaabouni, Rahma, Fatiha, Deeni, Ahuja, Arun, Tomar, Gaurav Singh, Senter, Evan, Chadwick, Martin, Kornakov, Ilya, Attaluri, Nithya, Iturrate, Iñaki, Liu, Ruibo, Li, Yunxuan, Cogan, Sarah, Chen, Jeremy, Jia, Chao, Gu, Chenjie, Zhang, Qiao, Grimstad, Jordan, Hartman, Ale Jakse, Garcia, Xavier, Pillai, Thanumalayan Sankaranarayana, Devlin, Jacob, Laskin, Michael, Casas, Diego de Las, Valter, Dasha, Tao, Connie, Blanco, Lorenzo, Badia, Adrià Puigdomènech, Reitter, David, Chen, Mianna, Brennan, Jenny, Rivera, Clara, Brin, Sergey, Iqbal, Shariq, Surita, Gabriela, Labanowski, Jane, Rao, Abhi, Winkler, Stephanie, Parisotto, Emilio, Gu, Yiming, Olszewska, Kate, Addanki, Ravi, Miech, Antoine, Louis, Annie, Teplyashin, Denis, Brown, Geoff, Catt, Elliot, Balaguer, Jan, Xiang, Jackie, Wang, Pidong, Ashwood, Zoe, Briukhov, Anton, Webson, Albert, Ganapathy, Sanjay, Sanghavi, Smit, Kannan, Ajay, Chang, Ming-Wei, Stjerngren, Axel, Djolonga, Josip, Sun, Yuting, Bapna, Ankur, Aitchison, Matthew, Pejman, Pedram, Michalewski, Henryk, Yu, Tianhe, Wang, Cindy, Love, Juliette, Ahn, Junwhan, Bloxwich, Dawn, Han, Kehang, Humphreys, Peter, Sellam, Thibault, Bradbury, James, Godbole, Varun, Samangooei, Sina, Damoc, Bogdan, Kaskasoli, Alex, Arnold, Sébastien M. R., Vasudevan, Vijay, Agrawal, Shubham, Riesa, Jason, Lepikhin, Dmitry, Tanburn, Richard, Srinivasan, Srivatsan, Lim, Hyeontaek, Hodkinson, Sarah, Shyam, Pranav, Ferret, Johan, Hand, Steven, Garg, Ankush, Paine, Tom Le, Li, Jian, Li, Yujia, Giang, Minh, Neitz, Alexander, Abbas, Zaheer, York, Sarah, Reid, Machel, Cole, Elizabeth, Chowdhery, Aakanksha, Das, Dipanjan, Rogozińska, Dominika, Nikolaev, Vitaliy, Sprechmann, Pablo, Nado, Zachary, Zilka, Lukas, Prost, Flavien, He, Luheng, Monteiro, Marianne, Mishra, Gaurav, Welty, Chris, Newlan, Josh, Jia, Dawei, Allamanis, Miltiadis, Hu, Clara Huiyi, de Liedekerke, Raoul, Gilmer, Justin, Saroufim, Carl, Rijhwani, Shruti, Hou, Shaobo, Shrivastava, Disha, Baddepudi, Anirudh, Goldin, Alex, Ozturel, Adnan, Cassirer, Albin, Xu, Yunhan, Sohn, Daniel, Sachan, Devendra, Amplayo, Reinald Kim, Swanson, Craig, Petrova, Dessie, Narayan, Shashi, Guez, Arthur, Brahma, Siddhartha, Landon, Jessica, Patel, Miteyan, Zhao, Ruizhe, Villela, Kevin, Wang, Luyu, Jia, Wenhao, Rahtz, Matthew, Giménez, Mai, Yeung, Legg, Keeling, James, Georgiev, Petko, Mincu, Diana, Wu, Boxi, Haykal, Salem, Saputro, Rachel, Vodrahalli, Kiran, Qin, James, Cankara, Zeynep, Sharma, Abhanshu, Fernando, Nick, Hawkins, Will, Neyshabur, Behnam, Kim, Solomon, Hutter, Adrian, Agrawal, Priyanka, Castro-Ros, Alex, Driessche, George van den, Wang, Tao, Yang, Fan, Chang, Shuo-yiin, Komarek, Paul, McIlroy, Ross, Lučić, Mario, Zhang, Guodong, Farhan, Wael, Sharman, Michael, Natsev, Paul, Michel, Paul, Bansal, Yamini, Qiao, Siyuan, Cao, Kris, Shakeri, Siamak, Butterfield, Christina, Chung, Justin, Rubenstein, Paul Kishan, Agrawal, Shivani, Mensch, Arthur, Soparkar, Kedar, Lenc, Karel, Chung, Timothy, Pope, Aedan, Maggiore, Loren, Kay, Jackie, Jhakra, Priya, Wang, Shibo, Maynez, Joshua, Phuong, Mary, Tobin, Taylor, Tacchetti, Andrea, Trebacz, Maja, Robinson, Kevin, Katariya, Yash, Riedel, Sebastian, Bailey, Paige, Xiao, Kefan, Ghelani, Nimesh, Aroyo, Lora, Slone, Ambrose, Houlsby, Neil, Xiong, Xuehan, Yang, Zhen, Gribovskaya, Elena, Adler, Jonas, Wirth, Mateo, Lee, Lisa, Li, Music, Kagohara, Thais, Pavagadhi, Jay, Bridgers, Sophie, Bortsova, Anna, Ghemawat, Sanjay, Ahmed, Zafarali, Liu, Tianqi, Powell, Richard, Bolina, Vijay, Iinuma, Mariko, Zablotskaia, Polina, Besley, James, Chung, Da-Woon, Dozat, Timothy, Comanescu, Ramona, Si, Xiance, Greer, Jeremy, Su, Guolong, Polacek, Martin, Kaufman, Raphaël Lopez, Tokumine, Simon, Hu, Hexiang, Buchatskaya, Elena, Miao, Yingjie, Elhawaty, Mohamed, Siddhant, Aditya, Tomasev, Nenad, Xing, Jinwei, Greer, Christina, Miller, Helen, Ashraf, Shereen, Roy, Aurko, Zhang, Zizhao, Ma, Ada, Filos, Angelos, Besta, Milos, Blevins, Rory, Klimenko, Ted, Yeh, Chih-Kuan, Changpinyo, Soravit, Mu, Jiaqi, Chang, Oscar, Pajarskas, Mantas, Muir, Carrie, Cohen, Vered, Lan, Charline Le, Haridasan, Krishna, Marathe, Amit, Hansen, Steven, Douglas, Sholto, Samuel, Rajkumar, Wang, Mingqiu, Austin, Sophia, Lan, Chang, Jiang, Jiepu, Chiu, Justin, Lorenzo, Jaime Alonso, Sjösund, Lars Lowe, Cevey, Sébastien, Gleicher, Zach, Avrahami, Thi, Boral, Anudhyan, Srinivasan, Hansa, Selo, Vittorio, May, Rhys, Aisopos, Konstantinos, Hussenot, Léonard, Soares, Livio Baldini, Baumli, Kate, Chang, Michael B., Recasens, Adrià, Caine, Ben, Pritzel, Alexander, Pavetic, Filip, Pardo, Fabio, Gergely, Anita, Frye, Justin, Ramasesh, Vinay, Horgan, Dan, Badola, Kartikeya, Kassner, Nora, Roy, Subhrajit, Dyer, Ethan, Campos, Víctor Campos, Tomala, Alex, Tang, Yunhao, Badawy, Dalia El, White, Elspeth, Mustafa, Basil, Lang, Oran, Jindal, Abhishek, Vikram, Sharad, Gong, Zhitao, Caelles, Sergi, Hemsley, Ross, Thornton, Gregory, Feng, Fangxiaoyu, Stokowiec, Wojciech, Zheng, Ce, Thacker, Phoebe, Ünlü, Çağlar, Zhang, Zhishuai, Saleh, Mohammad, Svensson, James, Bileschi, Max, Patil, Piyush, Anand, Ankesh, Ring, Roman, Tsihlas, Katerina, Vezer, Arpi, Selvi, Marco, Shevlane, Toby, Rodriguez, Mikel, Kwiatkowski, Tom, Daruki, Samira, Rong, Keran, Dafoe, Allan, FitzGerald, Nicholas, Gu-Lemberg, Keren, Khan, Mina, Hendricks, Lisa Anne, Pellat, Marie, Feinberg, Vladimir, Cobon-Kerr, James, Sainath, Tara, Rauh, Maribeth, Hashemi, Sayed Hadi, Ives, Richard, Hasson, Yana, Noland, Eric, Cao, Yuan, Byrd, Nathan, Hou, Le, Wang, Qingze, Sottiaux, Thibault, Paganini, Michela, Lespiau, Jean-Baptiste, Moufarek, Alexandre, Hassan, Samer, Shivakumar, Kaushik, van Amersfoort, Joost, Mandhane, Amol, Joshi, Pratik, Goyal, Anirudh, Tung, Matthew, Brock, Andrew, Sheahan, Hannah, Misra, Vedant, Li, Cheng, Rakićević, Nemanja, Dehghani, Mostafa, Liu, Fangyu, Mittal, Sid, Oh, Junhyuk, Noury, Seb, Sezener, Eren, Huot, Fantine, Lamm, Matthew, De Cao, Nicola, Chen, Charlie, Mudgal, Sidharth, Stella, Romina, Brooks, Kevin, Vasudevan, Gautam, Liu, Chenxi, Chain, Mainak, Melinkeri, Nivedita, Cohen, Aaron, Wang, Venus, Seymore, Kristie, Zubkov, Sergey, Goel, Rahul, Yue, Summer, Krishnakumaran, Sai, Albert, Brian, Hurley, Nate, Sano, Motoki, Mohananey, Anhad, Joughin, Jonah, Filonov, Egor, Kępa, Tomasz, Eldawy, Yomna, Lim, Jiawern, Rishi, Rahul, Badiezadegan, Shirin, Bos, Taylor, Chang, Jerry, Jain, Sanil, Padmanabhan, Sri Gayatri Sundara, Puttagunta, Subha, Krishna, Kalpesh, Baker, Leslie, Kalb, Norbert, Bedapudi, Vamsi, Kurzrok, Adam, Lei, Shuntong, Yu, Anthony, Litvin, Oren, Zhou, Xiang, Wu, Zhichun, Sobell, Sam, Siciliano, Andrea, Papir, Alan, Neale, Robby, Bragagnolo, Jonas, Toor, Tej, Chen, Tina, Anklin, Valentin, Wang, Feiran, Feng, Richie, Gholami, Milad, Ling, Kevin, Liu, Lijuan, Walter, Jules, Moghaddam, Hamid, Kishore, Arun, Adamek, Jakub, Mercado, Tyler, Mallinson, Jonathan, Wandekar, Siddhinita, Cagle, Stephen, Ofek, Eran, Garrido, Guillermo, Lombriser, Clemens, Mukha, Maksim, Sun, Botu, Mohammad, Hafeezul Rahman, Matak, Josip, Qian, Yadi, Peswani, Vikas, Janus, Pawel, Yuan, Quan, Schelin, Leif, David, Oana, Garg, Ankur, He, Yifan, Duzhyi, Oleksii, Älgmyr, Anton, Lottaz, Timothée, Li, Qi, Yadav, Vikas, Xu, Luyao, Chinien, Alex, Shivanna, Rakesh, Chuklin, Aleksandr, Li, Josie, Spadine, Carrie, Wolfe, Travis, Mohamed, Kareem, Das, Subhabrata, Dai, Zihang, He, Kyle, von Dincklage, Daniel, Upadhyay, Shyam, Maurya, Akanksha, Chi, Luyan, Krause, Sebastian, Salama, Khalid, Rabinovitch, Pam G, M, Pavan Kumar Reddy, Selvan, Aarush, Dektiarev, Mikhail, Ghiasi, Golnaz, Guven, Erdem, Gupta, Himanshu, Liu, Boyi, Sharma, Deepak, Shtacher, Idan Heimlich, Paul, Shachi, Akerlund, Oscar, Aubet, François-Xavier, Huang, Terry, Zhu, Chen, Zhu, Eric, Teixeira, Elico, Fritze, Matthew, Bertolini, Francesco, Marinescu, Liana-Eleonora, Bölle, Martin, Paulus, Dominik, Gupta, Khyatti, Latkar, Tejasi, Chang, Max, Sanders, Jason, Wilson, Roopa, Wu, Xuewei, Tan, Yi-Xuan, Thiet, Lam Nguyen, Doshi, Tulsee, Lall, Sid, Mishra, Swaroop, Chen, Wanming, Luong, Thang, Benjamin, Seth, Lee, Jasmine, Andrejczuk, Ewa, Rabiej, Dominik, Ranjan, Vipul, Styrc, Krzysztof, Yin, Pengcheng, Simon, Jon, Harriott, Malcolm Rose, Bansal, Mudit, Robsky, Alexei, Bacon, Geoff, Greene, David, Mirylenka, Daniil, Zhou, Chen, Sarvana, Obaid, Goyal, Abhimanyu, Andermatt, Samuel, Siegler, Patrick, Horn, Ben, Israel, Assaf, Pongetti, Francesco, Chen, Chih-Wei "Louis", Selvatici, Marco, Silva, Pedro, Wang, Kathie, Tolins, Jackson, Guu, Kelvin, Yogev, Roey, Cai, Xiaochen, Agostini, Alessandro, Shah, Maulik, Nguyen, Hung, Donnaile, Noah Ó, Pereira, Sébastien, Friso, Linda, Stambler, Adam, Kuang, Chenkai, Romanikhin, Yan, Geller, Mark, Yan, ZJ, Jang, Kane, Lee, Cheng-Chun, Fica, Wojciech, Malmi, Eric, Tan, Qijun, Banica, Dan, Balle, Daniel, Pham, Ryan, Huang, Yanping, Avram, Diana, Shi, Hongzhi, Singh, Jasjot, Hidey, Chris, Ahuja, Niharika, Saxena, Pranab, Dooley, Dan, Potharaju, Srividya Pranavi, O'Neill, Eileen, Gokulchandran, Anand, Foley, Ryan, Zhao, Kai, Dusenberry, Mike, Liu, Yuan, Mehta, Pulkit, Kotikalapudi, Ragha, Safranek-Shrader, Chalence, Goodman, Andrew, Kessinger, Joshua, Globen, Eran, Kolhar, Prateek, Gorgolewski, Chris, Ibrahim, Ali, Song, Yang, Eichenbaum, Ali, Brovelli, Thomas, Potluri, Sahitya, Lahoti, Preethi, Baetu, Cip, Ghorbani, Ali, Chen, Charles, Crawford, Andy, Pal, Shalini, Sridhar, Mukund, Gurita, Petru, Mujika, Asier, Petrovski, Igor, Cedoz, Pierre-Louis, Li, Chenmei, Chen, Shiyuan, Santo, Niccolò Dal, Goyal, Siddharth, Punjabi, Jitesh, Kappaganthu, Karthik, Kwak, Chester, LV, Pallavi, Velury, Sarmishta, Choudhury, Himadri, Hall, Jamie, Shah, Premal, Figueira, Ricardo, Thomas, Matt, Lu, Minjie, Zhou, Ting, Kumar, Chintu, Jurdi, Thomas, Chikkerur, Sharat, Ma, Yenai, Yu, Adams, Kwak, Soo, Ähdel, Victor, Rajayogam, Sujeevan, Choma, Travis, Liu, Fei, Barua, Aditya, Ji, Colin, Park, Ji Ho, Hellendoorn, Vincent, Bailey, Alex, Bilal, Taylan, Zhou, Huanjie, Khatir, Mehrdad, Sutton, Charles, Rzadkowski, Wojciech, Macintosh, Fiona, Shagin, Konstantin, Medina, Paul, Liang, Chen, Zhou, Jinjing, Shah, Pararth, Bi, Yingying, Dankovics, Attila, Banga, Shipra, Lehmann, Sabine, Bredesen, Marissa, Lin, Zifan, Hoffmann, John Eric, Lai, Jonathan, Chung, Raynald, Yang, Kai, Balani, Nihal, Bražinskas, Arthur, Sozanschi, Andrei, Hayes, Matthew, Alcalde, Héctor Fernández, Makarov, Peter, Chen, Will, Stella, Antonio, Snijders, Liselotte, Mandl, Michael, Kärrman, Ante, Nowak, Paweł, Wu, Xinyi, Dyck, Alex, Vaidyanathan, Krishnan, R, Raghavender, Mallet, Jessica, Rudominer, Mitch, Johnston, Eric, Mittal, Sushil, Udathu, Akhil, Christensen, Janara, Verma, Vishal, Irving, Zach, Santucci, Andreas, Elsayed, Gamaleldin, Davoodi, Elnaz, Georgiev, Marin, Tenney, Ian, Hua, Nan, Cideron, Geoffrey, Leurent, Edouard, Alnahlawi, Mahmoud, Georgescu, Ionut, Wei, Nan, Zheng, Ivy, Scandinaro, Dylan, Jiang, Heinrich, Snoek, Jasper, Sundararajan, Mukund, Wang, Xuezhi, Ontiveros, Zack, Karo, Itay, Cole, Jeremy, Rajashekhar, Vinu, Tumeh, Lara, Ben-David, Eyal, Jain, Rishub, Uesato, Jonathan, Datta, Romina, Bunyan, Oskar, Wu, Shimu, Zhang, John, Stanczyk, Piotr, Zhang, Ye, Steiner, David, Naskar, Subhajit, Azzam, Michael, Johnson, Matthew, Paszke, Adam, Chiu, Chung-Cheng, Elias, Jaume Sanchez, Mohiuddin, Afroz, Muhammad, Faizan, Miao, Jin, Lee, Andrew, Vieillard, Nino, Park, Jane, Zhang, Jiageng, Stanway, Jeff, Garmon, Drew, Karmarkar, Abhijit, Dong, Zhe, Lee, Jong, Kumar, Aviral, Zhou, Luowei, Evens, Jonathan, Isaac, William, Irving, Geoffrey, Loper, Edward, Fink, Michael, Arkatkar, Isha, Chen, Nanxin, Shafran, Izhak, Petrychenko, Ivan, Chen, Zhe, Jia, Johnson, Levskaya, Anselm, Zhu, Zhenkai, Grabowski, Peter, Mao, Yu, Magni, Alberto, Yao, Kaisheng, Snaider, Javier, Casagrande, Norman, Palmer, Evan, Suganthan, Paul, Castaño, Alfonso, Giannoumis, Irene, Kim, Wooyeol, Rybiński, Mikołaj, Sreevatsa, Ashwin, Prendki, Jennifer, Soergel, David, Goedeckemeyer, Adrian, Gierke, Willi, Jafari, Mohsen, Gaba, Meenu, Wiesner, Jeremy, Wright, Diana Gage, Wei, Yawen, Vashisht, Harsha, Kulizhskaya, Yana, Hoover, Jay, Le, Maigo, Li, Lu, Iwuanyanwu, Chimezie, Liu, Lu, Ramirez, Kevin, Khorlin, Andrey, Cui, Albert, LIN, Tian, Wu, Marcus, Aguilar, Ricardo, Pallo, Keith, Chakladar, Abhishek, Perng, Ginger, Abellan, Elena Allica, Zhang, Mingyang, Dasgupta, Ishita, Kushman, Nate, Penchev, Ivo, Repina, Alena, Wu, Xihui, van der Weide, Tom, Ponnapalli, Priya, Kaplan, Caroline, Simsa, Jiri, Li, Shuangfeng, Dousse, Olivier, Piper, Jeff, Ie, Nathan, Pasumarthi, Rama, Lintz, Nathan, Vijayakumar, Anitha, Andor, Daniel, Valenzuela, Pedro, Lui, Minnie, Paduraru, Cosmin, Peng, Daiyi, Lee, Katherine, Zhang, Shuyuan, Greene, Somer, Nguyen, Duc Dung, Kurylowicz, Paula, Hardin, Cassidy, Dixon, Lucas, Janzer, Lili, Choo, Kiam, Feng, Ziqiang, Zhang, Biao, Singhal, Achintya, Du, Dayou, McKinnon, Dan, Antropova, Natasha, Bolukbasi, Tolga, Keller, Orgad, Reid, David, Finchelstein, Daniel, Raad, Maria Abi, Crocker, Remi, Hawkins, Peter, Dadashi, Robert, Gaffney, Colin, Franko, Ken, Bulanova, Anna, Leblond, Rémi, Chung, Shirley, Askham, Harry, Cobo, Luis C., Xu, Kelvin, Fischer, Felix, Xu, Jun, Sorokin, Christina, Alberti, Chris, Lin, Chu-Cheng, Evans, Colin, Dimitriev, Alek, Forbes, Hannah, Banarse, Dylan, Tung, Zora, Omernick, Mark, Bishop, Colton, Sterneck, Rachel, Jain, Rohan, Xia, Jiawei, Amid, Ehsan, Piccinno, Francesco, Wang, Xingyu, Banzal, Praseem, Mankowitz, Daniel J., Polozov, Alex, Krakovna, Victoria, Brown, Sasha, Bateni, MohammadHossein, Duan, Dennis, Firoiu, Vlad, Thotakuri, Meghana, Natan, Tom, Geist, Matthieu, Girgin, Ser tan, Li, Hui, Ye, Jiayu, Roval, Ofir, Tojo, Reiko, Kwong, Michael, Lee-Thorp, James, Yew, Christopher, Sinopalnikov, Danila, Ramos, Sabela, Mellor, John, Sharma, Abhishek, Wu, Kathy, Miller, David, Sonnerat, Nicolas, Vnukov, Denis, Greig, Rory, Beattie, Jennifer, Caveness, Emily, Bai, Libin, Eisenschlos, Julian, Korchemniy, Alex, Tsai, Tomy, Jasarevic, Mimi, Kong, Weize, Dao, Phuong, Zheng, Zeyu, Liu, Frederick, Zhu, Rui, Teh, Tian Huey, Sanmiya, Jason, Gladchenko, Evgeny, Trdin, Nejc, Toyama, Daniel, Rosen, Evan, Tavakkol, Sasan, Xue, Linting, Elkind, Chen, Woodman, Oliver, Carpenter, John, Papamakarios, George, Kemp, Rupert, Kafle, Sushant, Grunina, Tanya, Sinha, Rishika, Talbert, Alice, Wu, Diane, Owusu-Afriyie, Denese, Thornton, Chloe, Pont-Tuset, Jordi, Narayana, Pradyumna, Li, Jing, Fatehi, Saaber, Wieting, John, Ajmeri, Omar, Uria, Benigno, Ko, Yeongil, Knight, Laura, Héliou, Amélie, Niu, Ning, Gu, Shane, Pang, Chenxi, Li, Yeqing, Levine, Nir, Stolovich, Ariel, Santamaria-Fernandez, Rebeca, Goenka, Sonam, Yustalim, Wenny, Strudel, Robin, Elqursh, Ali, Deck, Charlie, Lee, Hyo, Li, Zonglin, Levin, Kyle, Hoffmann, Raphael, Holtmann-Rice, Dan, Bachem, Olivier, Arora, Sho, Koh, Christy, Yeganeh, Soheil Hassas, Põder, Siim, Tariq, Mukarram, Sun, Yanhua, Ionita, Lucian, Seyedhosseini, Mojtaba, Tafti, Pouya, Liu, Zhiyu, Gulati, Anmol, Liu, Jasmine, Ye, Xinyu, Chrzaszcz, Bart, Wang, Lily, Sethi, Nikhil, Li, Tianrun, Brown, Ben, Singh, Shreya, Fan, Wei, Parisi, Aaron, Stanton, Joe, Koverkathu, Vinod, Choquette-Choo, Christopher A., Li, Yunjie, Lu, TJ, Shroff, Prakash, Varadarajan, Mani, Bahargam, Sanaz, Willoughby, Rob, Gaddy, David, Desjardins, Guillaume, Cornero, Marco, Robenek, Brona, Mittal, Bhavishya, Albrecht, Ben, Shenoy, Ashish, Moiseev, Fedor, Jacobsson, Henrik, Ghaffarkhah, Alireza, Rivière, Morgane, Walton, Alanna, Crepy, Clément, Parrish, Alicia, Zhou, Zongwei, Farabet, Clement, Radebaugh, Carey, Srinivasan, Praveen, van der Salm, Claudia, Fidjeland, Andreas, Scellato, Salvatore, Latorre-Chimoto, Eri, Klimczak-Plucińska, Hanna, Bridson, David, de Cesare, Dario, Hudson, Tom, Mendolicchio, Piermaria, Walker, Lexi, Morris, Alex, Mauger, Matthew, Guseynov, Alexey, Reid, Alison, Odoom, Seth, Loher, Lucia, Cotruta, Victor, Yenugula, Madhavi, Grewe, Dominik, Petrushkina, Anastasia, Duerig, Tom, Sanchez, Antonio, Yadlowsky, Steve, Shen, Amy, Globerson, Amir, Webb, Lynette, Dua, Sahil, Li, Dong, Bhupatiraju, Surya, Hurt, Dan, Qureshi, Haroon, Agarwal, Ananth, Shani, Tomer, Eyal, Matan, Khare, Anuj, Belle, Shreyas Rammohan, Wang, Lei, Tekur, Chetan, Kale, Mihir Sanjay, Wei, Jinliang, Sang, Ruoxin, Saeta, Brennan, Liechty, Tyler, Sun, Yi, Zhao, Yao, Lee, Stephan, Nayak, Pandu, Fritz, Doug, Vuyyuru, Manish Reddy, Aslanides, John, Vyas, Nidhi, Wicke, Martin, Ma, Xiao, Eltyshev, Evgenii, Martin, Nina, Cate, Hardie, Manyika, James, Amiri, Keyvan, Kim, Yelin, Xiong, Xi, Kang, Kai, Luisier, Florian, Tripuraneni, Nilesh, Madras, David, Guo, Mandy, Waters, Austin, Wang, Oliver, Ainslie, Joshua, Baldridge, Jason, Zhang, Han, Pruthi, Garima, Bauer, Jakob, Yang, Feng, Mansour, Riham, Gelman, Jason, Xu, Yang, Polovets, George, Liu, Ji, Cai, Honglong, Chen, Warren, Sheng, XiangHai, Xue, Emily, Ozair, Sherjil, Angermueller, Christof, Li, Xiaowei, Sinha, Anoop, Wang, Weiren, Wiesinger, Julia, Koukoumidis, Emmanouil, Tian, Yuan, Iyer, Anand, Gurumurthy, Madhu, Goldenson, Mark, Shah, Parashar, Blake, MK, Yu, Hongkun, Urbanowicz, Anthony, Palomaki, Jennimaria, Fernando, Chrisantha, Durden, Ken, Mehta, Harsh, Momchev, Nikola, Rahimtoroghi, Elahe, Georgaki, Maria, Raul, Amit, Ruder, Sebastian, Redshaw, Morgan, Lee, Jinhyuk, Zhou, Denny, Jalan, Komal, Li, Dinghua, Hechtman, Blake, Schuh, Parker, Nasr, Milad, Milan, Kieran, Mikulik, Vladimir, Franco, Juliana, Green, Tim, Nguyen, Nam, Kelley, Joe, Mahendru, Aroma, Hu, Andrea, Howland, Joshua, Vargas, Ben, Hui, Jeffrey, Bansal, Kshitij, Rao, Vikram, Ghiya, Rakesh, Wang, Emma, Ye, Ke, Sarr, Jean Michel, Preston, Melanie Moranski, Elish, Madeleine, Li, Steve, Kaku, Aakash, Gupta, Jigar, Pasupat, Ice, Juan, Da-Cheng, Someswar, Milan, M., Tejvi, Chen, Xinyun, Amini, Aida, Fabrikant, Alex, Chu, Eric, Dong, Xuanyi, Muthal, Amruta, Buthpitiya, Senaka, Jauhari, Sarthak, Khandelwal, Urvashi, Hitron, Ayal, Ren, Jie, Rinaldi, Larissa, Drath, Shahar, Dabush, Avigail, Jiang, Nan-Jiang, Godhia, Harshal, Sachs, Uli, Chen, Anthony, Fan, Yicheng, Taitelbaum, Hagai, Noga, Hila, Dai, Zhuyun, Wang, James, Hamer, Jenny, Ferng, Chun-Sung, Elkind, Chenel, Atias, Aviel, Lee, Paulina, Listík, Vít, Carlen, Mathias, van de Kerkhof, Jan, Pikus, Marcin, Zaher, Krunoslav, Müller, Paul, Zykova, Sasha, Stefanec, Richard, Gatsko, Vitaly, Hirnschall, Christoph, Sethi, Ashwin, Xu, Xingyu Federico, Ahuja, Chetan, Tsai, Beth, Stefanoiu, Anca, Feng, Bo, Dhandhania, Keshav, Katyal, Manish, Gupta, Akshay, Parulekar, Atharva, Pitta, Divya, Zhao, Jing, Bhatia, Vivaan, Bhavnani, Yashodha, Alhadlaq, Omar, Li, Xiaolin, Danenberg, Peter, Tu, Dennis, Pine, Alex, Filippova, Vera, Ghosh, Abhipso, Limonchik, Ben, Urala, Bhargava, Lanka, Chaitanya Krishna, Clive, Derik, Li, Edward, Wu, Hao, Hongtongsak, Kevin, Li, Ianna, Thakkar, Kalind, Omarov, Kuanysh, Majmundar, Kushal, Alverson, Michael, Kucharski, Michael, Patel, Mohak, Jain, Mudit, Zabelin, Maksim, Pelagatti, Paolo, Kohli, Rohan, Kumar, Saurabh, Kim, Joseph, Sankar, Swetha, Shah, Vineet, Ramachandruni, Lakshmi, Zeng, Xiangkai, Bariach, Ben, Weidinger, Laura, Vu, Tu, Andreev, Alek, He, Antoine, Hui, Kevin, Kashem, Sheleem, Subramanya, Amar, Hsiao, Sissie, Hassabis, Demis, Kavukcuoglu, Koray, Sadovsky, Adam, Le, Quoc, Strohman, Trevor, Wu, Yonghui, Petrov, Slav, Dean, Jeffrey, and Vinyals, Oriol
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Computer Vision and Pattern Recognition
Abstract: This report introduces a new family of multimodal models, Gemini, that exhibit remarkable capabilities across image, audio, video, and text understanding. The Gemini family consists of Ultra, Pro, and Nano sizes, suitable for applications ranging from complex reasoning tasks to on-device memory-constrained use-cases. Evaluation on a broad range of benchmarks shows that our most-capable Gemini Ultra model advances the state of the art in 30 of 32 of these benchmarks - notably being the first model to achieve human-expert performance on the well-studied exam benchmark MMLU, and improving the state of the art in every one of the 20 multimodal benchmarks we examined. We believe that the new capabilities of the Gemini family in cross-modal reasoning and language understanding will enable a wide variety of use cases. We discuss our approach toward post-training and deploying Gemini models responsibly to users through services including Gemini, Gemini Advanced, Google AI Studio, and Cloud Vertex AI.
Published: 2023

13. Interpretability Illusions in the Generalization of Simplified Models

Author: Friedman, Dan, Lampinen, Andrew, Dixon, Lucas, Chen, Danqi, and Ghandeharioun, Asma
Subjects: Computer Science - Machine Learning, Computer Science - Computation and Language
Abstract: A common method to study deep learning systems is to use simplified model representations--for example, using singular value decomposition to visualize the model's hidden states in a lower dimensional space. This approach assumes that the results of these simplifications are faithful to the original model. Here, we illustrate an important caveat to this assumption: even if the simplified representations can accurately approximate the full model on the training set, they may fail to accurately capture the model's behavior out of distribution. We illustrate this by training Transformer models on controlled datasets with systematic generalization splits, including the Dyck balanced-parenthesis languages and a code completion task. We simplify these models using tools like dimensionality reduction and clustering, and then explicitly test how these simplified proxies match the behavior of the original model. We find consistent generalization gaps: cases in which the simplified proxies are more faithful to the original model on the in-distribution evaluations and less faithful on various tests of systematic generalization. This includes cases where the original model generalizes systematically but the simplified proxies fail, and cases where the simplified proxies generalize better. Together, our results raise questions about the extent to which mechanistic interpretations derived using tools like SVD can reliably predict what a model will do in novel situations., Comment: ICML 2024
Published: 2023

14. Interactive AI Alignment: Specification, Process, and Evaluation Alignment

Author: Terry, Michael, Kulkarni, Chinmay, Wattenberg, Martin, Dixon, Lucas, and Morris, Meredith Ringel
Subjects: Computer Science - Human-Computer Interaction, Computer Science - Artificial Intelligence
Abstract: Modern AI enables a high-level, declarative form of interaction: Users describe the intended outcome they wish an AI to produce, but do not actually create the outcome themselves. In contrast, in traditional user interfaces, users invoke specific operations to create the desired outcome. This paper revisits the basic input-output interaction cycle in light of this declarative style of interaction, and connects concepts in AI alignment to define three objectives for interactive alignment of AI: specification alignment (aligning on what to do), process alignment (aligning on how to do it), and evaluation alignment (assisting users in verifying and understanding what was produced). Using existing systems as examples, we show how these user-centered views of AI alignment can be used descriptively, prescriptively, and as an evaluative aid.
Published: 2023

15. Large Language Models are Competitive Near Cold-start Recommenders for Language- and Item-based Preferences

Author: Sanner, Scott, Balog, Krisztian, Radlinski, Filip, Wedin, Ben, and Dixon, Lucas
Subjects: Computer Science - Information Retrieval, Computer Science - Machine Learning
Abstract: Traditional recommender systems leverage users' item preference history to recommend novel content that users may like. However, modern dialog interfaces that allow users to express language-based preferences offer a fundamentally different modality for preference input. Inspired by recent successes of prompting paradigms for large language models (LLMs), we study their use for making recommendations from both item-based and language-based preferences in comparison to state-of-the-art item-based collaborative filtering (CF) methods. To support this investigation, we collect a new dataset consisting of both item-based and language-based preferences elicited from users along with their ratings on a variety of (biased) recommended items and (unbiased) random items. Among numerous experimental results, we find that LLMs provide competitive recommendation performance for pure language-based preferences (no item preferences) in the near cold-start case in comparison to item-based CF methods, despite having no supervised training for this specific task (zero-shot) or only a few labels (few-shot). This is particularly promising as language-based preference representations are more explainable and scrutable than item-based or vector-based representations., Comment: To appear at RecSys'23
Published: 2023

16. Large Language Models for User Interest Journeys

Author: Christakopoulou, Konstantina, Lalama, Alberto, Adams, Cj, Qu, Iris, Amir, Yifat, Chucri, Samer, Vollucci, Pierce, Soldo, Fabio, Bseiso, Dina, Scodel, Sarah, Dixon, Lucas, Chi, Ed H., and Chen, Minmin
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Information Retrieval
Abstract: Large language models (LLMs) have shown impressive capabilities in natural language understanding and generation. Their potential for deeper user understanding and improved personalized user experience on recommendation platforms is, however, largely untapped. This paper aims to address this gap. Recommender systems today capture users' interests through encoding their historical activities on the platforms. The generated user representations are hard to examine or interpret. On the other hand, if we were to ask people about interests they pursue in their life, they might talk about their hobbies, like I just started learning the ukulele, or their relaxation routines, e.g., I like to watch Saturday Night Live, or I want to plant a vertical garden. We argue, and demonstrate through extensive experiments, that LLMs as foundation models can reason through user activities, and describe their interests in nuanced and interesting ways, similar to how a human would. We define interest journeys as the persistent and overarching user interests, in other words, the non-transient ones. These are the interests that we believe will benefit most from the nuanced and personalized descriptions. We introduce a framework in which we first perform personalized extraction of interest journeys, and then summarize the extracted journeys via LLMs, using techniques like few-shot prompting, prompt-tuning and fine-tuning. Together, our results in prompting LLMs to name extracted user journeys in a large-scale industrial platform demonstrate great potential of these models in providing deeper, more interpretable, and controllable user understanding. We believe LLM powered user understanding can be a stepping stone to entirely new user experiences on recommendation platforms that are journey-aware, assistive, and enabling frictionless conversation down the line.
Published: 2023

17. Simfluence: Modeling the Influence of Individual Training Examples by Simulating Training Runs

Author: Guu, Kelvin, Webson, Albert, Pavlick, Ellie, Dixon, Lucas, Tenney, Ian, and Bolukbasi, Tolga
Subjects: Computer Science - Machine Learning, Computer Science - Computation and Language
Abstract: Training data attribution (TDA) methods offer to trace a model's prediction on any given example back to specific influential training examples. Existing approaches do so by assigning a scalar influence score to each training example, under a simplifying assumption that influence is additive. But in reality, we observe that training examples interact in highly non-additive ways due to factors such as inter-example redundancy, training order, and curriculum learning effects. To study such interactions, we propose Simfluence, a new paradigm for TDA where the goal is not to produce a single influence score per example, but instead a training run simulator: the user asks, ``If my model had trained on example $z_1$, then $z_2$, ..., then $z_n$, how would it behave on $z_{test}$?''; the simulator should then output a simulated training run, which is a time series predicting the loss on $z_{test}$ at every step of the simulated run. This enables users to answer counterfactual questions about what their model would have learned under different training curricula, and to directly see where in training that learning would occur. We present a simulator, Simfluence-Linear, that captures non-additive interactions and is often able to predict the spiky trajectory of individual example losses with surprising fidelity. Furthermore, we show that existing TDA methods such as TracIn and influence functions can be viewed as special cases of Simfluence-Linear. This enables us to directly compare methods in terms of their simulation accuracy, subsuming several prior TDA approaches to evaluation. In experiments on large language model (LLM) fine-tuning, we show that our method predicts loss trajectories with much higher accuracy than existing TDA methods (doubling Spearman's correlation and reducing mean-squared error by 75%) across several tasks, models, and training methods.
Published: 2023

18. Gradient-Based Automated Iterative Recovery for Parameter-Efficient Tuning

Author: Mozes, Maximilian, Bolukbasi, Tolga, Yuan, Ann, Liu, Frederick, Thain, Nithum, and Dixon, Lucas
Subjects: Computer Science - Computation and Language
Abstract: Pretrained large language models (LLMs) are able to solve a wide variety of tasks through transfer learning. Various explainability methods have been developed to investigate their decision making process. TracIn (Pruthi et al., 2020) is one such gradient-based method which explains model inferences based on the influence of training examples. In this paper, we explore the use of TracIn to improve model performance in the parameter-efficient tuning (PET) setting. We develop conversational safety classifiers via the prompt-tuning PET method and show how the unique characteristics of the PET regime enable TracIn to identify the cause for certain misclassifications by LLMs. We develop a new methodology for using gradient-based explainability techniques to improve model performance, G-BAIR: gradient-based automated iterative recovery. We show that G-BAIR can recover LLM performance on benchmarks after manually corrupting training labels. This suggests that influence methods like TracIn can be used to automatically perform data cleaning, and introduces the potential for interactive debugging and relabeling for PET-based transfer learning methods., Comment: Pre-print
Published: 2023

19. Towards Agile Text Classifiers for Everyone

Author: Mozes, Maximilian, Hoffmann, Jessica, Tomanek, Katrin, Kouate, Muhamed, Thain, Nithum, Yuan, Ann, Bolukbasi, Tolga, and Dixon, Lucas
Subjects: Computer Science - Computation and Language
Abstract: Text-based safety classifiers are widely used for content moderation and increasingly to tune generative language model behavior - a topic of growing concern for the safety of digital assistants and chatbots. However, different policies require different classifiers, and safety policies themselves improve from iteration and adaptation. This paper introduces and evaluates methods for agile text classification, whereby classifiers are trained using small, targeted datasets that can be quickly developed for a particular policy. Experimenting with 7 datasets from three safety-related domains, comprising 15 annotation schemes, led to our key finding: prompt-tuning large language models, like PaLM 62B, with a labeled dataset of as few as 80 examples can achieve state-of-the-art performance. We argue that this enables a paradigm shift for text classification, especially for models supporting safer online discourse. Instead of collecting millions of examples to attempt to create universal safety classifiers over months or years, classifiers could be tuned using small datasets, created by individuals or small organizations, tailored for specific use cases, and iterated on and adapted in the time-span of a day., Comment: Findings of EMNLP 2023
Published: 2023

20. KNNs of Semantic Encodings for Rating Prediction

Author: Laugier, Léo, Vadapalli, Raghuram, Bonald, Thomas, and Dixon, Lucas
Subjects: Computer Science - Computation and Language
Abstract: This paper explores a novel application of textual semantic similarity to user-preference representation for rating prediction. The approach represents a user's preferences as a graph of textual snippets from review text, where the edges are defined by semantic similarity. This textual, memory-based approach to rating prediction enables review-based explanations for recommendations. The method is evaluated quantitatively, highlighting that leveraging text in this way outperforms both strong memory-based and model-based collaborative filtering baselines.
Published: 2023

21. Basic Elements of Logical Graphs

Author: Dixon, Lucas
Subjects: Computer Science - Logic in Computer Science, Computer Science - Symbolic Computation
Abstract: We considers how a particular kind of graph corresponds to multiplicative intuitionistic linear logic formula. The main feature of the graphical notation is that it absorbs certain symmetries between conjunction and implication. We look at the basic definitions and present details of an implementation in the functional programming language Standard ML. This provides a functional approach to graph traversal and demonstrates how graph isomorphism be implemented in just a few lines of readable code. This works takes the initial steps towards a graphical language and toolkit for working with logic formula and derivations., Comment: Published in the CAMCAD-09 workshop (https://www.irit.fr/~Ralph.Matthes/CAMCAD09/)
Published: 2022

22. Beyond Rewards: a Hierarchical Perspective on Offline Multiagent Behavioral Analysis

Author: Omidshafiei, Shayegan, Kapishnikov, Andrei, Assogba, Yannick, Dixon, Lucas, and Kim, Been
Subjects: Computer Science - Machine Learning, Computer Science - Multiagent Systems
Abstract: Each year, expert-level performance is attained in increasingly-complex multiagent domains, where notable examples include Go, Poker, and StarCraft II. This rapid progression is accompanied by a commensurate need to better understand how such agents attain this performance, to enable their safe deployment, identify limitations, and reveal potential means of improving them. In this paper we take a step back from performance-focused multiagent learning, and instead turn our attention towards agent behavior analysis. We introduce a model-agnostic method for discovery of behavior clusters in multiagent domains, using variational inference to learn a hierarchy of behaviors at the joint and local agent levels. Our framework makes no assumption about agents' underlying learning algorithms, does not require access to their latent states or policies, and is trained using only offline observational data. We illustrate the effectiveness of our method for enabling the coupled understanding of behaviors at the joint and local agent level, detection of behavior changepoints throughout training, discovery of core behavioral concepts, demonstrate the approach's scalability to a high-dimensional multiagent MuJoCo control domain, and also illustrate that the approach can disentangle previously-trained policies in OpenAI's hide-and-seek domain.
Published: 2022

23. On Natural Language User Profiles for Transparent and Scrutable Recommendation

Author: Radlinski, Filip, Balog, Krisztian, Diaz, Fernando, Dixon, Lucas, and Wedin, Ben
Subjects: Computer Science - Information Retrieval
Abstract: Natural interaction with recommendation and personalized search systems has received tremendous attention in recent years. We focus on the challenge of supporting people's understanding and control of these systems and explore a fundamentally new way of thinking about representation of knowledge in recommendation and personalization systems. Specifically, we argue that it may be both desirable and possible for algorithms that use natural language representations of users' preferences to be developed. We make the case that this could provide significantly greater transparency, as well as affordances for practical actionable interrogation of, and control over, recommendations. Moreover, we argue that such an approach, if successfully applied, may enable a major step towards systems that rely less on noisy implicit observations while increasing portability of knowledge of one's interests., Comment: Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '22), 2022
Published: 2022
Full Text: View/download PDF

24. GLaM: Efficient Scaling of Language Models with Mixture-of-Experts

Author: Du, Nan, Huang, Yanping, Dai, Andrew M., Tong, Simon, Lepikhin, Dmitry, Xu, Yuanzhong, Krikun, Maxim, Zhou, Yanqi, Yu, Adams Wei, Firat, Orhan, Zoph, Barret, Fedus, Liam, Bosma, Maarten, Zhou, Zongwei, Wang, Tao, Wang, Yu Emma, Webster, Kellie, Pellat, Marie, Robinson, Kevin, Meier-Hellstern, Kathleen, Duke, Toju, Dixon, Lucas, Zhang, Kun, Le, Quoc V, Wu, Yonghui, Chen, Zhifeng, and Cui, Claire
Subjects: Computer Science - Computation and Language
Abstract: Scaling language models with more data, compute and parameters has driven significant progress in natural language processing. For example, thanks to scaling, GPT-3 was able to achieve strong results on in-context learning tasks. However, training these large dense models requires significant amounts of computing resources. In this paper, we propose and develop a family of language models named GLaM (Generalist Language Model), which uses a sparsely activated mixture-of-experts architecture to scale the model capacity while also incurring substantially less training cost compared to dense variants. The largest GLaM has 1.2 trillion parameters, which is approximately 7x larger than GPT-3. It consumes only 1/3 of the energy used to train GPT-3 and requires half of the computation flops for inference, while still achieving better overall zero-shot and one-shot performance across 29 NLP tasks., Comment: Accepted to ICML 2022
Published: 2021

25. Toxicity Detection can be Sensitive to the Conversational Context

Author: Xenos, Alexandros, Pavlopoulos, John, Androutsopoulos, Ion, Dixon, Lucas, Sorensen, Jeffrey, and Laugier, Leo
Subjects: Computer Science - Computation and Language
Abstract: User posts whose perceived toxicity depends on the conversational context are rare in current toxicity detection datasets. Hence, toxicity detectors trained on existing datasets will also tend to disregard context, making the detection of context-sensitive toxicity harder when it does occur. We construct and publicly release a dataset of 10,000 posts with two kinds of toxicity labels: (i) annotators considered each post with the previous one as context; and (ii) annotators had no additional context. Based on this, we introduce a new task, context sensitivity estimation, which aims to identify posts whose perceived toxicity changes if the context (previous post) is also considered. We then evaluate machine learning systems on this task, showing that classifiers of practical quality can be developed, and we show that data augmentation with knowledge distillation can improve the performance further. Such systems could be used to enhance toxicity detection datasets with more context-dependent posts, or to suggest when moderators should consider the parent posts, which often may be unnecessary and may otherwise introduce significant additional cost., Comment: 13 pages, 8 figures
Published: 2021

26. Augmenting the User-Item Graph with Textual Similarity Models

Author: López, Federico, Scholz, Martin, Yung, Jessica, Pellat, Marie, Strube, Michael, and Dixon, Lucas
Subjects: Computer Science - Computation and Language, I.2.7
Abstract: This paper introduces a simple and effective form of data augmentation for recommender systems. A paraphrase similarity model is applied to widely available textual data, such as reviews and product descriptions, yielding new semantic relations that are added to the user-item graph. This increases the density of the graph without needing further labeled data. The data augmentation is evaluated on a variety of recommendation algorithms, using Euclidean, hyperbolic, and complex spaces, and over three categories of Amazon product reviews with differing characteristics. Results show that the data augmentation technique provides significant improvements to all types of models, with the most pronounced gains for knowledge graph-based recommenders, particularly in cold-start settings, leading to state-of-the-art performance., Comment: 12 pages, 2 figures
Published: 2021

27. Civil Rephrases Of Toxic Texts With Self-Supervised Transformers

Author: Laugier, Leo, Pavlopoulos, John, Sorensen, Jeffrey, and Dixon, Lucas
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Machine Learning
Abstract: Platforms that support online commentary, from social networks to news sites, are increasingly leveraging machine learning to assist their moderation efforts. But this process does not typically provide feedback to the author that would help them contribute according to the community guidelines. This is prohibitively time-consuming for human moderators to do, and computational approaches are still nascent. This work focuses on models that can help suggest rephrasings of toxic comments in a more civil manner. Inspired by recent progress in unpaired sequence-to-sequence tasks, a self-supervised learning model is introduced, called CAE-T5. CAE-T5 employs a pre-trained text-to-text transformer, which is fine tuned with a denoising and cyclic auto-encoder loss. Experimenting with the largest toxicity detection dataset to date (Civil Comments) our model generates sentences that are more fluent and better at preserving the initial content compared to earlier text style transfer systems which we compare with using several scoring systems and human evaluation.
Published: 2021

28. Six Attributes of Unhealthy Conversation

Author: Price, Ilan, Gifford-Moore, Jordan, Fleming, Jory, Musker, Saul, Roichman, Maayan, Sylvain, Guillaume, Thain, Nithum, Dixon, Lucas, and Sorensen, Jeffrey
Subjects: Computer Science - Computation and Language, Computer Science - Social and Information Networks
Abstract: We present a new dataset of approximately 44000 comments labeled by crowdworkers. Each comment is labelled as either 'healthy' or 'unhealthy', in addition to binary labels for the presence of six potentially 'unhealthy' sub-attributes: (1) hostile; (2) antagonistic, insulting, provocative or trolling; (3) dismissive; (4) condescending or patronising; (5) sarcastic; and/or (6) an unfair generalisation. Each label also has an associated confidence score. We argue that there is a need for datasets which enable research based on a broad notion of 'unhealthy online conversation'. We build this typology to encompass a substantial proportion of the individual comments which contribute to unhealthy online conversation. For some of these attributes, this is the first publicly available dataset of this scale. We explore the quality of the dataset, present some summary statistics and initial models to illustrate the utility of this data, and highlight limitations and directions for further research., Comment: Appearing in the 4th Workshop on Online Abuse and Harms (2020)
Published: 2020

29. Toxicity Detection: Does Context Really Matter?

Author: Pavlopoulos, John, Sorensen, Jeffrey, Dixon, Lucas, Thain, Nithum, and Androutsopoulos, Ion
Subjects: Computer Science - Computation and Language
Abstract: Moderation is crucial to promoting healthy on-line discussions. Although several `toxicity' detection datasets and models have been published, most of them ignore the context of the posts, implicitly assuming that comments maybe judged independently. We investigate this assumption by focusing on two questions: (a) does context affect the human judgement, and (b) does conditioning on context improve performance of toxicity detection systems? We experiment with Wikipedia conversations, limiting the notion of context to the previous post in the thread and the discussion title. We find that context can both amplify or mitigate the perceived toxicity of posts. Moreover, a small but significant subset of manually labeled posts (5% in one of our experiments) end up having the opposite toxicity labels if the annotators are not provided with context. Surprisingly, we also find no evidence that context actually improves the performance of toxicity classifiers, having tried a range of classifiers and mechanisms to make them context aware. This points to the need for larger datasets of comments annotated in context. We make our code and data publicly available.
Published: 2020

30. Classifying Constructive Comments

Author: Kolhatkar, Varada, Thain, Nithum, Sorensen, Jeffrey, Dixon, Lucas, and Taboada, Maite
Subjects: Computer Science - Computation and Language, Computer Science - Computers and Society, Computer Science - Information Retrieval, Computer Science - Machine Learning
Abstract: We introduce the Constructive Comments Corpus (C3), comprised of 12,000 annotated news comments, intended to help build new tools for online communities to improve the quality of their discussions. We define constructive comments as high-quality comments that make a contribution to the conversation. We explain the crowd worker annotation scheme and define a taxonomy of sub-characteristics of constructiveness. The quality of the annotation scheme and the resulting dataset is evaluated using measurements of inter-annotator agreement, expert assessment of a sample, and by the constructiveness sub-characteristics, which we show provide a proxy for the general constructiveness concept. We provide models for constructiveness trained on C3 using both feature-based and a variety of deep learning approaches and demonstrate that these models capture general rather than topic- or domain-specific characteristics of constructiveness, through domain adaptation experiments. We examine the role that length plays in our models, as comment length could be easily gamed if models depend heavily upon this feature. By examining the errors made by each model and their distribution by length, we show that the best performing models are less correlated with comment length.The constructiveness corpus and our experiments pave the way for a moderation tool focused on promoting comments that make a contribution, rather than only filtering out undesirable content.
Published: 2020

31. Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classification

Author: Borkan, Daniel, Dixon, Lucas, Sorensen, Jeffrey, Thain, Nithum, and Vasserman, Lucy
Subjects: Computer Science - Machine Learning, Computer Science - Computation and Language, Statistics - Machine Learning
Abstract: Unintended bias in Machine Learning can manifest as systemic differences in performance for different demographic groups, potentially compounding existing challenges to fairness in society at large. In this paper, we introduce a suite of threshold-agnostic metrics that provide a nuanced view of this unintended bias, by considering the various ways that a classifier's score distribution can vary across designated groups. We also introduce a large new test set of online comments with crowd-sourced annotations for identity references. We use this to show how our metrics can be used to find new and potentially subtle unintended bias in existing public models., Comment: Updated to fix typo in Equation 4
Published: 2019

32. Limitations of Pinned AUC for Measuring Unintended Bias

Author: Borkan, Daniel, Dixon, Lucas, Li, John, Sorensen, Jeffrey, Thain, Nithum, and Vasserman, Lucy
Subjects: Statistics - Machine Learning, Computer Science - Machine Learning
Abstract: This report examines the Pinned AUC metric introduced and highlights some of its limitations. Pinned AUC provides a threshold-agnostic measure of unintended bias in a classification model, inspired by the ROC-AUC metric. However, as we highlight in this report, there are ways that the metric can obscure different kinds of unintended biases when the underlying class distributions on which bias is being measured are not carefully controlled.
Published: 2019

33. WikiConv: A Corpus of the Complete Conversational History of a Large Online Collaborative Community

Author: Hua, Yiqing, Danescu-Niculescu-Mizil, Cristian, Taraborelli, Dario, Thain, Nithum, Sorensen, Jeffery, and Dixon, Lucas
Subjects: Computer Science - Computation and Language
Abstract: We present a corpus that encompasses the complete history of conversations between contributors to Wikipedia, one of the largest online collaborative communities. By recording the intermediate states of conversations---including not only comments and replies, but also their modifications, deletions and restorations---this data offers an unprecedented view of online conversation. This level of detail supports new research questions pertaining to the process (and challenges) of large-scale online collaboration. We illustrate the corpus' potential with two case studies that highlight new perspectives on earlier work. First, we explore how a person's conversational behavior depends on how they relate to the discussion's venue. Second, we show that community moderation of toxic behavior happens at a higher rate than previously estimated. Finally the reconstruction framework is designed to be language agnostic, and we show that it can extract high quality conversational data in both Chinese and English.
Published: 2018

34. "We Need Structured Output": Towards User-centered Constraints on Large Language Model Output

Author: Liu, Michael Xieyang, primary, Liu, Frederick, additional, Fiannaca, Alexander J., additional, Koo, Terry, additional, Dixon, Lucas, additional, Terry, Michael, additional, and Cai, Carrie J., additional
Published: 2024
Full Text: View/download PDF

35. LLM Comparator: Visual Analytics for Side-by-Side Evaluation of Large Language Models

Author: Kahng, Minsuk, primary, Tenney, Ian, additional, Pushkarna, Mahima, additional, Liu, Michael Xieyang, additional, Wexler, James, additional, Reif, Emily, additional, Kallarackal, Krystal, additional, Chang, Minsuk, additional, Terry, Michael, additional, and Dixon, Lucas, additional
Published: 2024
Full Text: View/download PDF

36. Conversations Gone Awry: Detecting Early Signs of Conversational Failure

Author: Zhang, Justine, Chang, Jonathan P., Danescu-Niculescu-Mizil, Cristian, Dixon, Lucas, Hua, Yiqing, Thain, Nithum, and Taraborelli, Dario
Subjects: Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Computers and Society, Computer Science - Human-Computer Interaction, Physics - Physics and Society
Abstract: One of the main challenges online social systems face is the prevalence of antisocial behavior, such as harassment and personal attacks. In this work, we introduce the task of predicting from the very start of a conversation whether it will get out of hand. As opposed to detecting undesirable behavior after the fact, this task aims to enable early, actionable prediction at a time when the conversation might still be salvaged. To this end, we develop a framework for capturing pragmatic devices---such as politeness strategies and rhetorical prompts---used to start a conversation, and analyze their relation to its future trajectory. Applying this framework in a controlled setting, we demonstrate the feasibility of detecting early warning signs of antisocial behavior in online discussions., Comment: To appear in the Proceedings of ACL 2018, 15 pages, 1 figure. Data, quiz, code and additional information at http://www.cs.cornell.edu/~cristian/Conversations_gone_awry.html
Published: 2018

37. Ex Machina: Personal Attacks Seen at Scale

Author: Wulczyn, Ellery, Thain, Nithum, and Dixon, Lucas
Subjects: Computer Science - Computation and Language
Abstract: The damage personal attacks cause to online discourse motivates many platforms to try to curb the phenomenon. However, understanding the prevalence and impact of personal attacks in online platforms at scale remains surprisingly difficult. The contribution of this paper is to develop and illustrate a method that combines crowdsourcing and machine learning to analyze personal attacks at scale. We show an evaluation method for a classifier in terms of the aggregated number of crowd-workers it can approximate. We apply our methodology to English Wikipedia, generating a corpus of over 100k high quality human-labeled comments and 63M machine-labeled ones from a classifier that is as good as the aggregate of 3 crowd-workers, as measured by the area under the ROC curve and Spearman correlation. Using this corpus of machine-labeled scores, our methodology allows us to explore some of the open questions about the nature of online personal attacks. This reveals that the majority of personal attacks on Wikipedia are not the result of a few malicious users, nor primarily the consequence of allowing anonymous contributions from unregistered users.
Published: 2016

38. Network Traffic Obfuscation and Automated Internet Censorship

Author: Dixon, Lucas, Ristenpart, Thomas, and Shrimpton, Thomas
Subjects: Computer Science - Networking and Internet Architecture, Computer Science - Cryptography and Security, C.2
Abstract: Internet censors seek ways to identify and block internet access to information they deem objectionable. Increasingly, censors deploy advanced networking tools such as deep-packet inspection (DPI) to identify such connections. In response, activists and academic researchers have developed and deployed network traffic obfuscation mechanisms. These apply specialized cryptographic tools to attempt to hide from DPI the true nature and content of connections. In this survey, we give an overview of network traffic obfuscation and its role in circumventing Internet censorship. We provide historical and technical background that motivates the need for obfuscation tools, and give an overview of approaches to obfuscation used by state of the art tools. We discuss the latest research on how censors might detect these efforts. We also describe the current challenges to censorship circumvention research and identify concrete ways for the community to address these challenges., Comment: 9 pages
Published: 2016

39. KNNs of Semantic Encodings for Rating Prediction

Author: Laugier, Léo, primary, Vadapalli, Raghuram, additional, Bonald, Thomas, additional, and Dixon, Lucas, additional
Published: 2023
Full Text: View/download PDF

40. Open Graphs and Monoidal Theories

Author: Dixon, Lucas and Kissinger, Aleks
Subjects: Mathematics - Category Theory, Computer Science - Logic in Computer Science, F.4, G.4
Abstract: String diagrams are a powerful tool for reasoning about physical processes, logic circuits, tensor networks, and many other compositional structures. The distinguishing feature of these diagrams is that edges need not be connected to vertices at both ends, and these unconnected ends can be interpreted as the inputs and outputs of a diagram. In this paper, we give a concrete construction for string diagrams using a special kind of typed graph called an open-graph. While the category of open-graphs is not itself adhesive, we introduce the notion of a selective adhesive functor, and show that such a functor embeds the category of open-graphs into the ambient adhesive category of typed graphs. Using this functor, the category of open-graphs inherits "enough adhesivity" from the category of typed graphs to perform double-pushout (DPO) graph rewriting. A salient feature of our theory is that it ensures rewrite systems are "type-safe" in the sense that rewriting respects the inputs and outputs. This formalism lets us safely encode the interesting structure of a computational model, such as evaluation dynamics, with succinct, explicit rewrite rules, while the graphical representation absorbs many of the tedious details. Although topological formalisms exist for string diagrams, our construction is discreet, finitary, and enjoys decidable algorithms for composition and rewriting. We also show how open-graphs can be parametrised by graphical signatures, similar to the monoidal signatures of Joyal and Street, which define types for vertices in the diagrammatic language and constraints on how they can be connected. Using typed open-graphs, we can construct free symmetric monoidal categories, PROPs, and more general monoidal theories. Thus open-graphs give us a handle for mechanised reasoning in monoidal categories., Comment: 31 pages, currently technical report, submitted to MSCS, waiting reviews
Published: 2010

41. Open Graphs and Computational Reasoning

Author: Dixon, Lucas, Duncan, Ross, and Kissinger, Aleks
Subjects: Computer Science - Logic in Computer Science, Computer Science - Mathematical Software, Computer Science - Symbolic Computation
Abstract: We present a form of algebraic reasoning for computational objects which are expressed as graphs. Edges describe the flow of data between primitive operations which are represented by vertices. These graphs have an interface made of half-edges (edges which are drawn with an unconnected end) and enjoy rich compositional principles by connecting graphs along these half-edges. In particular, this allows equations and rewrite rules to be specified between graphs. Particular computational models can then be encoded as an axiomatic set of such rules. Further rules can be derived graphically and rewriting can be used to simulate the dynamics of a computational system, e.g. evaluating a program on an input. Examples of models which can be formalised in this way include traditional electronic circuits as well as recent categorical accounts of quantum information., Comment: In Proceedings DCM 2010, arXiv:1006.1937
Published: 2010
Full Text: View/download PDF

42. Graphical Reasoning in Compact Closed Categories for Quantum Computation

Author: Dixon, Lucas and Duncan, Ross
Subjects: Computer Science - Symbolic Computation, Computer Science - Artificial Intelligence
Abstract: Compact closed categories provide a foundational formalism for a variety of important domains, including quantum computation. These categories have a natural visualisation as a form of graphs. We present a formalism for equational reasoning about such graphs and develop this into a generic proof system with a fixed logical kernel for equational reasoning about compact closed categories. Automating this reasoning process is motivated by the slow and error prone nature of manual graph manipulation. A salient feature of our system is that it provides a formal and declarative account of derived results that can include `ellipses'-style notation. We illustrate the framework by instantiating it for a graphical language of quantum computation and show how this can be used to perform symbolic computation., Comment: 21 pages, 9 figures. This is the journal version of the paper published at AISC
Published: 2009

43. We Usually Give Like This: Social Norms Describe Typical Charitable Causes Supported by Group Members.

Author: Chapman, Cassandra M., Dixon, Lucas, Wallin, Ann, Young, Tarli, Masser, Barbara M., and Louis, Winnifred R.
Subjects: *SOCIAL norms, *FUNDRAISING, *SOCIAL groups, *SOCIAL influence, *CONSERVATIVES, *CHARITIES, *LIBERALS
Abstract: Millions of nonprofits compete for a share of the billions of dollars donated to charity each year. Yet how donors select which charities to support remains relatively understudied. Social norms influence whether people give to charity at all, but no research has yet considered whether norms also communicate information about which causes group members typically support. To address this important question, we surveyed 1,735 people from 117 countries to understand whether they identified normative causes typically supported by their social groups. We found different normative giving profiles for men, women, older people, younger people, conservatives, progressives, religious, and nonreligious people, with varying degrees of consensus within each social group. Results demonstrate empirically—and for the first time—that social identities contain normative content about which charitable causes group members typically support. Some causes were relatively universally approved of or avoided. Results can inform nonprofit fundraising strategy around segmentation and targeting. [ABSTRACT FROM AUTHOR]
Published: 2024
Full Text: View/download PDF

44. Large Language Models are Competitive Near Cold-start Recommenders for Language- and Item-based Preferences

Author: Sanner, Scott, primary, Balog, Krisztian, additional, Radlinski, Filip, additional, Wedin, Ben, additional, and Dixon, Lucas, additional
Published: 2023
Full Text: View/download PDF

45. A proof planning framework for Isabelle

Author: Dixon, Lucas and Fleuriot, Jacques
Subjects: 006.3, PhD thesis, Proof planning
Abstract: Proof planning is a paradigm for the automation of proof that focuses on encoding intelligence to guide the proof process. The idea is to capture common patterns of reasoning which can be used to derive abstract descriptions of proofs known as proof plans. These can then be executed to provide fully formal proofs. This thesis concerns the development and analysis of a novel approach to proof planning that focuses on an explicit representation of choices during search. We embody our approach as a proof planner for the generic proof assistant Isabelle and use the Isar language, which is human-readable and machine-checkable, to represent proof plans. Within this framework we develop an inductive theorem prover as a case study of our approach to proof planning. Our prover uses the difference reduction heuristic known as rippling to automate the step cases of the inductive proofs. The development of a flexible approach to rippling that supports its various modifications and extensions is the second major focus of this thesis. Here, our inductive theorem prover provides a context in which to evaluate rippling experimentally. This work results in an efficient and powerful inductive theorem prover for Isabelle as well as proposals for further improving the efficiency of rippling. We also draw observations in order to direct further work on proof planning. Overall, we aim to make it easier for mathematical techniques, and those specific to mechanical theorem proving, to be encoded and applied to problems.
Published: 2006

46. Classifying constructive comments

Author: Kolhatkar, Varada, primary, Thain, Nithum, additional, Sorensen, Jeffrey, additional, Dixon, Lucas, additional, and Taboada, Maite, additional
Published: 2023
Full Text: View/download PDF

47. The psychology of attraction to multi‐level marketing

Author: Dixon, Lucas J., primary, Hornsey, Matthew J., additional, Hartley, Nicole, additional, Chapman, Cassandra M., additional, and Brienza, Justin P., additional
Published: 2023
Full Text: View/download PDF

48. We Usually Give Like This: Social Norms Describe Typical Charitable Causes Supported by Group Members

Author: Chapman, Cassandra M., primary, Dixon, Lucas, additional, Wallin, Ann, additional, Young, Tarli, additional, Masser, Barbara M., additional, and Louis, Winnifred R., additional
Published: 2023
Full Text: View/download PDF

49. Towards Agile Text Classifiers for Everyone

Author: Mozes, Maximilian, primary, Hoffmann, Jessica, additional, Tomanek, Katrin, additional, Kouate, Muhamed, additional, Thain, Nithum, additional, Yuan, Ann, additional, Bolukbasi, Tolga, additional, and Dixon, Lucas, additional
Published: 2023
Full Text: View/download PDF

50. Harmful Language Datasets: An Assessment of Robustness

Author: Korre, Katerina, primary, Pavlopoulos, John, additional, Sorensen, Jeffrey, additional, Laugier, Léo, additional, Androutsopoulos, Ion, additional, Dixon, Lucas, additional, and Barrón-cedeño, Alberto, additional
Published: 2023
Full Text: View/download PDF

Catalog

Books, media, physical & digital resources

See catalog results

Searchworks

Select search scope, currently: Articles Catalog books, media & more in Jio Institute collections Articles journal articles & other e-resources

Search

Search Constraints

Refine your results

Search Limiters

Topic

Publication Year Range

Language

Publication Type

Journal

Database

Publisher

198 results on '"Dixon, Lucas"'

Search Results

Catalog

Select search scope, currently: Articles

Catalog

books, media & more in Jio Institute collections

Articles

journal articles & other e-resources