mittagessen
commited on
Upload MBartForCurveRegression
Browse files- config.json +3 -3
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "mittagessen/
|
3 |
"activation_dropout": 0.0,
|
4 |
"activation_function": "gelu",
|
5 |
"add_cross_attention": true,
|
@@ -8,7 +8,7 @@
|
|
8 |
"MBartForCurveRegression"
|
9 |
],
|
10 |
"attention_dropout": 0.0,
|
11 |
-
"bos_token_id":
|
12 |
"classifier_dropout": 0.0,
|
13 |
"d_model": 1024,
|
14 |
"decoder_attention_heads": 16,
|
@@ -28,7 +28,7 @@
|
|
28 |
"max_position_embeddings": 768,
|
29 |
"model_type": "mbart",
|
30 |
"num_hidden_layers": 12,
|
31 |
-
"pad_token_id":
|
32 |
"scale_embedding": true,
|
33 |
"tie_word_embeddings": false,
|
34 |
"torch_dtype": "float32",
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "mittagessen/reg_transformer_seg_decoder",
|
3 |
"activation_dropout": 0.0,
|
4 |
"activation_function": "gelu",
|
5 |
"add_cross_attention": true,
|
|
|
8 |
"MBartForCurveRegression"
|
9 |
],
|
10 |
"attention_dropout": 0.0,
|
11 |
+
"bos_token_id": 1,
|
12 |
"classifier_dropout": 0.0,
|
13 |
"d_model": 1024,
|
14 |
"decoder_attention_heads": 16,
|
|
|
28 |
"max_position_embeddings": 768,
|
29 |
"model_type": "mbart",
|
30 |
"num_hidden_layers": 12,
|
31 |
+
"pad_token_id": 0,
|
32 |
"scale_embedding": true,
|
33 |
"tie_word_embeddings": false,
|
34 |
"torch_dtype": "float32",
|