Upload veridian_beta_0_config.yaml with huggingface_hub
Browse files- veridian_beta_0_config.yaml +27 -0
veridian_beta_0_config.yaml
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: veridian-1b-bootstrap
|
| 2 |
+
vocab_size: 32000
|
| 3 |
+
pad_token_id: 0
|
| 4 |
+
bos_token_id: 1
|
| 5 |
+
eos_token_id: 2
|
| 6 |
+
max_position_embeddings: 2048
|
| 7 |
+
target_max_position_embeddings: 2048
|
| 8 |
+
hidden_size: 1536
|
| 9 |
+
num_hidden_layers: 16
|
| 10 |
+
num_attention_heads: 12
|
| 11 |
+
num_key_value_heads: 4
|
| 12 |
+
intermediate_size: 4096
|
| 13 |
+
num_local_experts: 4
|
| 14 |
+
num_shared_experts: 1
|
| 15 |
+
num_experts_per_tok: 2
|
| 16 |
+
attention_dropout: 0.0
|
| 17 |
+
residual_dropout: 0.0
|
| 18 |
+
rms_norm_eps: 1.0e-6
|
| 19 |
+
rope_theta: 100000.0
|
| 20 |
+
router_aux_loss_coef: 0.01
|
| 21 |
+
router_z_loss_coef: 1.0e-4
|
| 22 |
+
capacity_factor_train: 1.25
|
| 23 |
+
capacity_factor_eval: 1.0
|
| 24 |
+
initializer_range: 0.02
|
| 25 |
+
tie_word_embeddings: true
|
| 26 |
+
use_bias: false
|
| 27 |
+
use_cache: true
|