Comment by rurban
7 months ago
Did you see that halluzination in the paper?
It optimized
initializers.normal (0.0
to
initializers.normal (0 + 1j * 0,
I thought the results were being reviewed?
Anyway, impressive results. That's why OpenAI and Elon were so frightened about Hassabi.
I mean that is changing the dtype, perhaps that's relevant.