Upload config.yaml
Browse files- config.yaml +33 -0
config.yaml
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
affine_norm: true
|
| 2 |
+
block_type: TransformerBlock
|
| 3 |
+
d_model: 768
|
| 4 |
+
drop_path: 0
|
| 5 |
+
embed_dropout: 0
|
| 6 |
+
gradient_checkpointing: true
|
| 7 |
+
layer_norm_epsilon: 1.0e-05
|
| 8 |
+
learnable_word_embeddings: true
|
| 9 |
+
max_position_embeddings: 4096
|
| 10 |
+
n_layers: 18
|
| 11 |
+
num_heads: 1
|
| 12 |
+
pad_vocab_size_multiple: 1
|
| 13 |
+
pre_norm: true
|
| 14 |
+
resid_dropout: 0
|
| 15 |
+
residuals_in_fp32: True
|
| 16 |
+
sequence_mixer:
|
| 17 |
+
kwargs:
|
| 18 |
+
configs:
|
| 19 |
+
- kwargs:
|
| 20 |
+
kernel_size: 3
|
| 21 |
+
l_max: 4096
|
| 22 |
+
name: rebased.modules.base_conv.BaseConv
|
| 23 |
+
- kwargs:
|
| 24 |
+
feature_dim: 128
|
| 25 |
+
num_heads: 8
|
| 26 |
+
num_key_value_heads: 8
|
| 27 |
+
name: rebased.modules.rebased.ReBased
|
| 28 |
+
name: rebased.modules.hybrid.Hybrid
|
| 29 |
+
state_mixer:
|
| 30 |
+
kwargs:
|
| 31 |
+
hidden_mult: 4
|
| 32 |
+
name: rebased.modules.mlp.MLP
|
| 33 |
+
vocab_size: 50304
|