News for hackers. Stuff that matters.
Re:Qwen2.5-VL-32B: Smarter and Lighter by kmacdough on Tuesday March 25, 2025 @09:47AM (#43471257)
I am having a hard time finding controlled testing, but the premise is straightforward: different modalities encourage different skills and understandings. Text builds up more formal idea tokenization and strengthens logic/reasoning while images require it learns a more robust geometric intuition. Since these learnings are applied to the same latent space, the strengths can be cross-applied.

The same applies to humans. Imagine a human who's only life involved reading books in a dark room, vs one who could see images vs one who can actually interact.

Re:Qwen2.5-VL-32B: Smarter and Lighter by ItDoBeWimdyTho on Monday March 31, 2025 @03:31PM (#43538865)
That comparison actually makes human reasoning abilities more impressive.

Helen Keller still learned robust generalizations.