MCPcopy
hub / github.com/karpathy/makemore / ModelConfig

Class ModelConfig

makemore.py:33–40  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

31
32@dataclass
33class ModelConfig:
34 block_size: int = None # length of the input sequences of integers
35 vocab_size: int = None # the input integers are in range [0 .. vocab_size -1]
36 # parameters below control the sizes of each model slightly differently
37 n_layer: int = 4
38 n_embd: int = 64
39 n_embd2: int = 64
40 n_head: int = 4
41
42# -----------------------------------------------------------------------------
43# Transformer Language Model (*exactly* as used in GPT-2)

Callers 1

makemore.pyFile · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected