Join Us to Discuss GenAI's Latest Research on LLM Interpretability
It's been an exciting couple of weeks for GenAI! Join us tomorrow @ 10:15am PST as we discuss the newest research from Anthropic and OpenAI on LLM Interpretability. Dat Vibhu S. & Sai K. will break down two recent papers on sparse autoencoders (an unsupervised approach for extracting interpretable features from an LLM):
From OpenAI, a paper that proposes using k-sparse autoencoders to directly control sparsity, simplifying tuning and improving the reconstruction-sparsity frontier.
From Anthropic, research showing that (among other things) scaling laws can be used to guide the training of sparse autoencoders.
Link to join the discussion: https://arize.zoom.us/j/89593430181 If you haven't already, register for our community paper readings here: https://arize.com/resource/community-papers-reading/
