Alert button
Picture for Tomasz Korbak

Tomasz Korbak

Alert button

Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data

Add code
Bookmark button
Alert button
Apr 01, 2024
Matthias Gerstgrasser, Rylan Schaeffer, Apratim Dey, Rafael Rafailov, Henry Sleight, John Hughes, Tomasz Korbak, Rajashree Agrawal, Dhruv Pai, Andrey Gromov, Daniel A. Roberts, Diyi Yang, David L. Donoho, Sanmi Koyejo

Viaarxiv icon

Towards Understanding Sycophancy in Language Models

Add code
Bookmark button
Alert button
Oct 27, 2023
Mrinank Sharma, Meg Tong, Tomasz Korbak, David Duvenaud, Amanda Askell, Samuel R. Bowman, Newton Cheng, Esin Durmus, Zac Hatfield-Dodds, Scott R. Johnston, Shauna Kravec, Timothy Maxwell, Sam McCandlish, Kamal Ndousse, Oliver Rausch, Nicholas Schiefer, Da Yan, Miranda Zhang, Ethan Perez

Figure 1 for Towards Understanding Sycophancy in Language Models
Figure 2 for Towards Understanding Sycophancy in Language Models
Figure 3 for Towards Understanding Sycophancy in Language Models
Figure 4 for Towards Understanding Sycophancy in Language Models
Viaarxiv icon

Compositional preference models for aligning LMs

Add code
Bookmark button
Alert button
Oct 17, 2023
Dongyoung Go, Tomasz Korbak, Germán Kruszewski, Jos Rozen, Marc Dymetman

Viaarxiv icon

The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"

Add code
Bookmark button
Alert button
Sep 22, 2023
Lukas Berglund, Meg Tong, Max Kaufmann, Mikita Balesni, Asa Cooper Stickland, Tomasz Korbak, Owain Evans

Figure 1 for The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
Figure 2 for The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
Figure 3 for The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
Figure 4 for The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
Viaarxiv icon

Taken out of context: On measuring situational awareness in LLMs

Add code
Bookmark button
Alert button
Sep 01, 2023
Lukas Berglund, Asa Cooper Stickland, Mikita Balesni, Max Kaufmann, Meg Tong, Tomasz Korbak, Daniel Kokotajlo, Owain Evans

Figure 1 for Taken out of context: On measuring situational awareness in LLMs
Figure 2 for Taken out of context: On measuring situational awareness in LLMs
Figure 3 for Taken out of context: On measuring situational awareness in LLMs
Figure 4 for Taken out of context: On measuring situational awareness in LLMs
Viaarxiv icon

Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback

Add code
Bookmark button
Alert button
Jul 27, 2023
Stephen Casper, Xander Davies, Claudia Shi, Thomas Krendl Gilbert, Jérémy Scheurer, Javier Rando, Rachel Freedman, Tomasz Korbak, David Lindner, Pedro Freire, Tony Wang, Samuel Marks, Charbel-Raphaël Segerie, Micah Carroll, Andi Peng, Phillip Christoffersen, Mehul Damani, Stewart Slocum, Usman Anwar, Anand Siththaranjan, Max Nadeau, Eric J. Michaud, Jacob Pfau, Dmitrii Krasheninnikov, Xin Chen, Lauro Langosco, Peter Hase, Erdem Bıyık, Anca Dragan, David Krueger, Dorsa Sadigh, Dylan Hadfield-Menell

Figure 1 for Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
Figure 2 for Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
Figure 3 for Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
Figure 4 for Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
Viaarxiv icon

Inverse Scaling: When Bigger Isn't Better

Add code
Bookmark button
Alert button
Jun 15, 2023
Ian R. McKenzie, Alexander Lyzhov, Michael Pieler, Alicia Parrish, Aaron Mueller, Ameya Prabhu, Euan McLean, Aaron Kirtland, Alexis Ross, Alisa Liu, Andrew Gritsevskiy, Daniel Wurgaft, Derik Kauffman, Gabriel Recchia, Jiacheng Liu, Joe Cavanagh, Max Weiss, Sicong Huang, The Floating Droid, Tom Tseng, Tomasz Korbak, Xudong Shen, Yuhui Zhang, Zhengping Zhou, Najoung Kim, Samuel R. Bowman, Ethan Perez

Figure 1 for Inverse Scaling: When Bigger Isn't Better
Figure 2 for Inverse Scaling: When Bigger Isn't Better
Figure 3 for Inverse Scaling: When Bigger Isn't Better
Figure 4 for Inverse Scaling: When Bigger Isn't Better
Viaarxiv icon