juancopi81's picture
Add t5x and mt3 models
b100e1c
# ByT5 XXL model.
include 't5x/examples/t5/byt5/base.gin' # imports vocab, optimizer and model.
# ------------------- Network specification overrides --------------------------
network.Transformer.config = @network.T5Config()
network.T5Config:
emb_dim = 4672
num_heads = 64
num_encoder_layers = 36
num_decoder_layers = 12
head_dim = 64
mlp_dim = 12352