nafyi.

What Comes After Transformers?

The Data Exchange ·

Zuzanna Stamirowska calls RAG, context, and memory bottlenecks in 2023 and 2024 for what she describes as 'AI with no ceiling, if you wish.' She recalls realizing that reasoning should not just be linguistic, and says latent thinking can help solve constraint-satisfaction problems and math with 'tiny, tiny models.'. Lee 3 puntos de vista con sus evidencias y enlaces a las fuentes.

Ben LoricaZuzanna Stamirowska
Leer los puntos claveAbrir Editor (se abre en una pestaña nueva)

De un vistazo

  • RAG, context, and memory were bottlenecks in 2023–2024

    Stamirowska says the team saw RAG, context, and memory as bottlenecks in 2023 and 2024. They were not efficient or sufficient for what she describes as 'AI with no ceiling, if you wish.'

    Ver el momento de apoyo · Publisher transcript paragraph 7

  • Stamirowska says reasoning shouldn't just be linguistic

    Stamirowska says, 'We realized that reasoning shouldn't just be linguistic.' She says verbalizing things in chain-of-thought is not optimal: it involves 'gluing it on top of a transformer,' and transformers are naturally ill-suited to dealing with time and learning from experience.

    Ver el momento de apoyo · Publisher transcript paragraph 8

  • Latent thinking can help solve constraint-satisfaction and math problems

    Stamirowska says latent thinking can help solve constraint-satisfaction problems and math with 'tiny, tiny models.'

    Ver el momento de apoyo · Publisher transcript paragraph 37

Momentos clave3

Fragmentos breves y atribuidos, con el contexto necesario para verificarlos. La conversación completa permanece con su editor.

Publisher transcript paragraph 7 (se abre en una pestaña nueva)
post-transformer architecture motivation

RAG, context, and memory were bottlenecks in 2023–2024

Extracto original

But very concretely, what prompted us to start Pathway and to start working on post-transformer architectures was that we were looking forward to reasoning becoming a thing. We have a very strong background in dealing with the dimension of time in machine learning—any sort of online learning. And we saw RAG, context, and memory, and we’re talking about 2023 and 2024, as bottlenecks. They were not efficient or sufficient to unlock AI as we would like—AI with no ceiling, if you wish.
Zuzanna Stamirowska·
Publisher transcript paragraph 8 (se abre en una pestaña nueva)
reasoning modality

Stamirowska says reasoning shouldn't just be linguistic

Extracto original

Specifically, we’re talking about continual learning, memory, and reasoning first. And as we moved through it, we realized that reasoning shouldn’t just be linguistic. Verbalizing things in chain of thought is not the optimal way to do reasoning. With chain of thought, we’re essentially gluing it on top of a transformer, and transformers are naturally ill-suited to dealing with time and learning from experience.
Zuzanna Stamirowska·
Publisher transcript paragraph 37 (se abre en una pestaña nueva)
latent reasoning benefits

Latent thinking can help solve constraint-satisfaction and math problems

Extracto original

Latent thinking has huge benefits. It can help us solve constraint-satisfaction problems natively, like what we’re showing. We’re solving constraint-satisfaction problems and math with tiny, tiny models. And then there’s cost efficiency. There are just insane cost gains coming from latent reasoning.
Zuzanna Stamirowska·

Fuente y metodología

Estas perspectivas enlazan a sus fuentes originales. Las paráfrasis están identificadas y no son citas textuales.

Abrir transcripción o material de origen (se abre en una pestaña nueva)Reportar un problema