Hezar: Upload model and config
Browse files- model_config.yaml +22 -0
model_config.yaml
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
name: t5_text2text
|
| 2 |
+
config_type: model
|
| 3 |
+
vocab_size: 32103
|
| 4 |
+
d_model: 768
|
| 5 |
+
d_kv: 64
|
| 6 |
+
d_ff: 2048
|
| 7 |
+
num_layers: 12
|
| 8 |
+
num_decoder_layers: 12
|
| 9 |
+
num_heads: 12
|
| 10 |
+
relative_attention_num_buckets: 32
|
| 11 |
+
relative_attention_max_distance: 128
|
| 12 |
+
dropout_rate: 0.1
|
| 13 |
+
layer_norm_epsilon: 1.0e-06
|
| 14 |
+
initializer_factor: 1.0
|
| 15 |
+
feed_forward_proj: gated-gelu
|
| 16 |
+
is_encoder_decoder: true
|
| 17 |
+
use_cache: true
|
| 18 |
+
pad_token_id: 0
|
| 19 |
+
decoder_start_token_id: 0
|
| 20 |
+
eos_token_id: 1
|
| 21 |
+
min_length: 0
|
| 22 |
+
max_length: 100
|