mirror of
https://github.com/wassname/iris_bigvae.git
synced 2026-09-09 11:24:31 +08:00
11 lines
184 B
YAML
11 lines
184 B
YAML
_target_: models.TransformerConfig
|
|
tokens_per_block: 17
|
|
max_blocks: 20
|
|
attention: 'causal'
|
|
num_layers: 10
|
|
num_heads: 4
|
|
embed_dim: 256
|
|
embed_pdrop: 0.1
|
|
resid_pdrop: 0.1
|
|
attn_pdrop: 0.1
|