Weaviate Podcast Features MIT Work on Recursive Language Models
Episode covers language model harnesses as compositional generalizers along with related MIT research.
Weaviate researcher Connor Shorten announced the 142nd episode of the Weaviate Podcast with MIT PhD student Alex Zhang. The episode explains recursive language models that treat prompts as program variables and use recursive sub-agent calls for task decomposition instead of naive context stuffing. It covers Prime Agent, speculative programmatic tool calling, running RLMs in the cloud, and their impact on search. Omar Khattab replied to thank Shorten and Weaviate for three years of support.
I'm SUPER EXCITED to publish the 142nd episode of the Weaviate Podcast with Alex Zhang (@a1zhang)! 🔥 Alex is a Ph.D. student at MIT, where he has lead the work behind "Recursive Language Models", as well as "The Mismanaged Genius Hypothesis", "Language Model Harnesses are…
Reactions
2 postsCS 
Connor Shorten@CShorten30Thanks Omar! 💚 And love this! I think the industry moved to function calling loops with "naive context stuffing" because the flexibility of the interface trumped the ideas at the time. But it certainly never sat well with the DSPy community, and I think you and @a1zhang…
- likes: 5
- replies: 1
- bookmarks: 0
- reposts: 0
OK Omar Khattab@lateinteraction@CShorten30 @a1zhang Can’t appreciate your support enough Connor and team @weaviate_io :-) It’s been over three years!
- likes: 5
- replies: 0
- bookmarks: 0
- reposts: 0
