1-layer-addition / model_config.py
melephant's picture
Publish addition-transformer run s85nnxtf
58223a8 verified
Raw
History Blame Contribute Delete
236 Bytes
from __future__ import annotations
from typing import Protocol
class AdditionModelConfig(Protocol):
d_model: int
n_heads: int
d_mlp: int
n_layers: int
max_seq_len: int
init_mode: str
residual_alpha: float