Push model using huggingface_hub.
Browse files- config.json +5 -5
- model.safetensors +2 -2
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"d_model":
|
| 3 |
"debug": true,
|
| 4 |
"layer_norm_eps": 1e-05,
|
| 5 |
"d_vocab": 32001,
|
|
@@ -8,15 +8,15 @@
|
|
| 8 |
"d_head": 64,
|
| 9 |
"dt_head": 64,
|
| 10 |
"d_mlp": 3072,
|
| 11 |
-
"n_heads":
|
| 12 |
-
"nt_heads":
|
| 13 |
-
"n_layers":
|
| 14 |
"dropout": 0.1,
|
| 15 |
"mlp_type": "all",
|
| 16 |
"with_ln": true,
|
| 17 |
"order_attn": true,
|
| 18 |
"attn_eq": false,
|
| 19 |
-
"window_size":
|
| 20 |
"look_backward": 1,
|
| 21 |
"pad_value": 0,
|
| 22 |
"autopad": true
|
|
|
|
| 1 |
{
|
| 2 |
+
"d_model": 512,
|
| 3 |
"debug": true,
|
| 4 |
"layer_norm_eps": 1e-05,
|
| 5 |
"d_vocab": 32001,
|
|
|
|
| 8 |
"d_head": 64,
|
| 9 |
"dt_head": 64,
|
| 10 |
"d_mlp": 3072,
|
| 11 |
+
"n_heads": 6,
|
| 12 |
+
"nt_heads": 2,
|
| 13 |
+
"n_layers": 8,
|
| 14 |
"dropout": 0.1,
|
| 15 |
"mlp_type": "all",
|
| 16 |
"with_ln": true,
|
| 17 |
"order_attn": true,
|
| 18 |
"attn_eq": false,
|
| 19 |
+
"window_size": 32,
|
| 20 |
"look_backward": 1,
|
| 21 |
"pad_value": 0,
|
| 22 |
"autopad": true
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a0203bdab16d278b88aa327ca1eef72296e3acbf241623de2ef3bcc8d0519a67
|
| 3 |
+
size 322413644
|