Picture for Max Bartolo

Max Bartolo

Improving Reward Models with Synthetic Critiques

Add code
May 31, 2024
Viaarxiv icon

Fishing for Magikarp: Automatically Detecting Under-trained Tokens in Large Language Models

Add code
May 08, 2024
Figure 1 for Fishing for Magikarp: Automatically Detecting Under-trained Tokens in Large Language Models
Figure 2 for Fishing for Magikarp: Automatically Detecting Under-trained Tokens in Large Language Models
Figure 3 for Fishing for Magikarp: Automatically Detecting Under-trained Tokens in Large Language Models
Figure 4 for Fishing for Magikarp: Automatically Detecting Under-trained Tokens in Large Language Models
Viaarxiv icon

The PRISM Alignment Project: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models

Add code
Apr 24, 2024
Viaarxiv icon

Aya Dataset: An Open-Access Collection for Multilingual Instruction Tuning

Add code
Feb 09, 2024
Viaarxiv icon

DMLR: Data-centric Machine Learning Research -- Past, Present and Future

Add code
Nov 21, 2023
Figure 1 for DMLR: Data-centric Machine Learning Research -- Past, Present and Future
Figure 2 for DMLR: Data-centric Machine Learning Research -- Past, Present and Future
Figure 3 for DMLR: Data-centric Machine Learning Research -- Past, Present and Future
Viaarxiv icon

Human Feedback is not Gold Standard

Add code
Sep 28, 2023
Viaarxiv icon

Adversarial Nibbler: A Data-Centric Challenge for Improving the Safety of Text-to-Image Models

Add code
May 22, 2023
Figure 1 for Adversarial Nibbler: A Data-Centric Challenge for Improving the Safety of Text-to-Image Models
Figure 2 for Adversarial Nibbler: A Data-Centric Challenge for Improving the Safety of Text-to-Image Models
Figure 3 for Adversarial Nibbler: A Data-Centric Challenge for Improving the Safety of Text-to-Image Models
Viaarxiv icon

Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality

Apr 07, 2022
Figure 1 for Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality
Figure 2 for Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality
Figure 3 for Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality
Figure 4 for Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality
Viaarxiv icon

Dynatask: A Framework for Creating Dynamic AI Benchmark Tasks

Add code
Apr 05, 2022
Figure 1 for Dynatask: A Framework for Creating Dynamic AI Benchmark Tasks
Figure 2 for Dynatask: A Framework for Creating Dynamic AI Benchmark Tasks
Figure 3 for Dynatask: A Framework for Creating Dynamic AI Benchmark Tasks
Figure 4 for Dynatask: A Framework for Creating Dynamic AI Benchmark Tasks
Viaarxiv icon

Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants

Add code
Dec 16, 2021
Figure 1 for Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants
Figure 2 for Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants
Figure 3 for Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants
Figure 4 for Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants
Viaarxiv icon