Archive
This page contains archived content from previous Stanford ACM activities and workshops. This content is preserved for historical reference but may no longer be current or active.
Stanford AI Clinic (Archived)
Note: This is archived content from the Stanford AI Clinic.
The Stanford AI Clinic is a student group that studies AI policy and offers technical advice and policy recommendations for researchers and developers in AI.
Please join our weekly meetings for technical workshops, policy discussions, and to join our quarterly project. This quarter's project is focused on improving people's access to AI services.
Where: Kimball Hall Lounge | When: Thursdays @ 7PM
(If you are interested in scheduling a pro bono advisory session or partnering with us, please email jk23541@stanford.edu to discuss details.)
AI Clinic Workshop Notes
Note: These are archived workshop notes from previous AI Clinic activities.
Reasoning with the s1 method (Archived)
The s1 paper develops reasoning capabilities in a large language model for just $50. In this workshop, we discussed the paper, how to run a similar experiment for free on Google Colab (or for $5 on Vast.ai, a GPU rental service), the HuggingFace library, and techniques to make the most of a limited GPU.
Beyond the technical notes, the most interesting discussion question was whether future deep learning advances will be driven by better data sources (like s1 curating a high-quality reasoning dataset and DeepSeek) or by even larger datasets. It opens related questions to whether the model providers need to continue scraping so much data, without permission even. We will continue discussing these questions and identifying solutions!