All Work Published on Foundation Models
Five teams will use the funding to advance their work in biology, generative AI and creativity, policing, and more.
Five teams will use the funding to advance their work in biology, generative AI and creativity, policing, and more.
Interventions on model-internal states are fundamental operations in many areas of AI, including model editing, steering, robustness, and interpretability. To facilitate such research, we introduce pyvene, an open-source Python library that supports customizable interventions on a range of different PyTorch modules. pyvene supports complex intervention schemes with an intuitive configuration format, and its interventions can be static or include trainable parameters. We show how pyvene provides a unified and extensible framework for performing interventions on neural models and sharing the intervened upon models with others. We illustrate the power of the library via interpretability analyses using causal abstraction and knowledge localization. We publish our library through Python Package Index (PyPI) and provide code, documentation, and tutorials at ‘https://github.com/stanfordnlp/pyvene‘.
Interventions on model-internal states are fundamental operations in many areas of AI, including model editing, steering, robustness, and interpretability. To facilitate such research, we introduce pyvene, an open-source Python library that supports customizable interventions on a range of different PyTorch modules. pyvene supports complex intervention schemes with an intuitive configuration format, and its interventions can be static or include trainable parameters. We show how pyvene provides a unified and extensible framework for performing interventions on neural models and sharing the intervened upon models with others. We illustrate the power of the library via interpretability analyses using causal abstraction and knowledge localization. We publish our library through Python Package Index (PyPI) and provide code, documentation, and tutorials at ‘https://github.com/stanfordnlp/pyvene‘.
What Makes a Good AI Benchmark?
This brief presents a novel assessment framework for evaluating the quality of AI benchmarks and scores 24 benchmarks against the framework.
This brief presents a novel assessment framework for evaluating the quality of AI benchmarks and scores 24 benchmarks against the framework.
In evaluating thousands of benchmarks that AI developers use to assess the quality of their new models, a team of Stanford researchers says 5% could have serious flaws that can lead to major ramifications.
In evaluating thousands of benchmarks that AI developers use to assess the quality of their new models, a team of Stanford researchers says 5% could have serious flaws that can lead to major ramifications.
In this paper, we evaluate the most effective error message types through a large-scale randomized controlled trial conducted in an open-access, online introductory computer science course with 8,762 students from 146 countries. We assess existing error message enhancement strategies, as well as two novel approaches of our own: (1) generating error messages using OpenAI's GPT in real time and (2) constructing error messages that incorporate the course discussion forum. By examining students' direct responses to error messages, and their behavior throughout the course, we quantitatively evaluate the immediate and longer term efficacy of different error message types. We find that students using GPT generated error messages repeat an error 23.1% less often in the subsequent attempt, and resolve an error in 34.8% fewer additional attempts, compared to students using standard error messages. We also perform an analysis across various demographics to understand any disparities in the impact of different error message types. Our results find no significant difference in the effectiveness of GPT generated error messages for students from varying socioeconomic and demographic backgrounds. Our findings underscore GPT generated error messages as the most helpful error message type, especially as a universally effective intervention across demographics.
In this paper, we evaluate the most effective error message types through a large-scale randomized controlled trial conducted in an open-access, online introductory computer science course with 8,762 students from 146 countries. We assess existing error message enhancement strategies, as well as two novel approaches of our own: (1) generating error messages using OpenAI's GPT in real time and (2) constructing error messages that incorporate the course discussion forum. By examining students' direct responses to error messages, and their behavior throughout the course, we quantitatively evaluate the immediate and longer term efficacy of different error message types. We find that students using GPT generated error messages repeat an error 23.1% less often in the subsequent attempt, and resolve an error in 34.8% fewer additional attempts, compared to students using standard error messages. We also perform an analysis across various demographics to understand any disparities in the impact of different error message types. Our results find no significant difference in the effectiveness of GPT generated error messages for students from varying socioeconomic and demographic backgrounds. Our findings underscore GPT generated error messages as the most helpful error message type, especially as a universally effective intervention across demographics.
Stanford scholars respond to a federal RFC on the U.S. AI Safety Institute’s draft guidelines for managing the misuse risk for dual-use foundation models.
Stanford scholars respond to a federal RFC on the U.S. AI Safety Institute’s draft guidelines for managing the misuse risk for dual-use foundation models.
