RT @_xjdr: if i am reading this right, the first 3 layers of DSv3 are — mimi · mimi10v3

Dec 26, 2024, 19:35 EST

RT @_xjdr: if i am reading this right, the first 3 layers of DSv3 are Dense?!?!

A screenshot of Python source code showing a PyTorch `Block` class definition, with line numbers 364-377 visible on the left. The code defines an `__init__` method and a `forward` method for a neural network layer.

A screenshot of a code snippet displaying a JSON configuration object with line numbers on a dark background. The text lists various model parameters such as vocabulary size, dimensions, and layer counts.