Moreno La Quatra commited on
Commit ·
9afde18
1
Parent(s): cee4f67
training config v1 model
Browse files- v1_config.json +13 -0
v1_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"batch_size": 64,
|
| 3 |
+
"mask_probability": 0.15,
|
| 4 |
+
"num_layers": 6,
|
| 5 |
+
"num_heads": 8,
|
| 6 |
+
"d_model": 128,
|
| 7 |
+
"d_ff": 256,
|
| 8 |
+
"p_dropout": 0.1,
|
| 9 |
+
"max_seq_len": 128,
|
| 10 |
+
"vocab_size": 25000,
|
| 11 |
+
"learning_rate": 3e-4,
|
| 12 |
+
"grad_clip_value": 1,
|
| 13 |
+
}
|