Graviscalar

4 ms·

1. 1.▲
   by Graviscalar 1y ago

   The geometric intuition is solid, but actually applying topology has been less fruitful in spite of a lot of people trying their best, as Chris Olah himself has said elsewhere in this thread.
2. 2.▲
   by Graviscalar 1y ago

   I would expect model/algorithm improvements from using topological concepts to analyze the manifolds in question or concrete results in model interpretability. Gunnar has studied some toy examples, but they were barely a step up from t
3. 3.▲
   by Graviscalar 1y ago

   I was one of the people that was super excited after reading the Chris Olah blogpost from 2014, and over the past decade I've seen the insight go exactly nowhere. It's neat but it hasn't driven any interesting results, though
4. 4.▲
   by Graviscalar 2y ago

   He's specifically addressing people who have the same goals as him but aren't seeing progress for the reasons he's listed. Nothing in the article suggests that he sees other goals as less valid, and it seems like you're
5. 5.▲
   Ask HN: Who's regularly using LLMs at work?

   9 points by Graviscalar 2y ago|21 comments