Публикация

Предыдущая публикация

Wenfeng Liang, CEO of DeepSeek, is admirably focused: - coding agent first - no distractions (image, video, world models) - we need continual learning, not “taste” elsewhere.news/en/elsewhere...

rdsf138

23 июл., 19:13

I think it is a misjudgment to take multimodality lightly, but I do agree that current approaches don't meaningfully increase a model's intelligence by adding more modalities, given that those modalities are not actually native to these models.

Обсуждение

0 прямых ответов

Ответов пока нет

Обсуждение ещё не началось.