AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
3min snip
Jonathan Frankle: From Lottery Tickets to LLMs
The Gradient: Perspectives on AI
The Lottery Ticket Hypothesis: Reflections and Flaws
The author reflects on their early work, the lottery ticket hypothesis, admitting it is their worst work. They acknowledge its popularity but emphasize it is not their best. The hypothesis suggests the existence of trainable smaller networks, but it lacks falsifiability. They provide existential evidence to support the hypothesis. The author advises PhD students to ask two-sided questions to ensure gaining knowledge regardless of the result. They discuss their follow-up paper on neural network training and the importance of asking measurement questions. They find two-sided questions less stressful and recommend them to students.
00:00
Transcript
Episode notes
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode