Add "max_position_embeddings": 248
Browse files- config.json +32 -31
config.json
CHANGED
@@ -1,31 +1,32 @@
|
|
1 |
-
{
|
2 |
-
"_name_or_path": "LongCLIP-SAE-ViT-L-14/",
|
3 |
-
"architectures": [
|
4 |
-
"CLIPModel"
|
5 |
-
],
|
6 |
-
"initializer_factor": 1.0,
|
7 |
-
"logit_scale_init_value": 4.6052,
|
8 |
-
"model_type": "clip",
|
9 |
-
"projection_dim": 768,
|
10 |
-
"text_config": {
|
11 |
-
"dropout": 0.0,
|
12 |
-
"hidden_size": 768,
|
13 |
-
"intermediate_size": 3072,
|
14 |
-
"max_position_embeddings": 248,
|
15 |
-
"model_type": "clip_text_model",
|
16 |
-
"num_attention_heads": 12,
|
17 |
-
"projection_dim": 768
|
18 |
-
|
19 |
-
|
20 |
-
"
|
21 |
-
"
|
22 |
-
|
23 |
-
"
|
24 |
-
"
|
25 |
-
"
|
26 |
-
"
|
27 |
-
"
|
28 |
-
"
|
29 |
-
"
|
30 |
-
|
31 |
-
}
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "LongCLIP-SAE-ViT-L-14/",
|
3 |
+
"architectures": [
|
4 |
+
"CLIPModel"
|
5 |
+
],
|
6 |
+
"initializer_factor": 1.0,
|
7 |
+
"logit_scale_init_value": 4.6052,
|
8 |
+
"model_type": "clip",
|
9 |
+
"projection_dim": 768,
|
10 |
+
"text_config": {
|
11 |
+
"dropout": 0.0,
|
12 |
+
"hidden_size": 768,
|
13 |
+
"intermediate_size": 3072,
|
14 |
+
"max_position_embeddings": 248,
|
15 |
+
"model_type": "clip_text_model",
|
16 |
+
"num_attention_heads": 12,
|
17 |
+
"projection_dim": 768,
|
18 |
+
"max_position_embeddings": 248
|
19 |
+
},
|
20 |
+
"torch_dtype": "float32",
|
21 |
+
"transformers_version": "4.46.3",
|
22 |
+
"vision_config": {
|
23 |
+
"dropout": 0.0,
|
24 |
+
"hidden_size": 1024,
|
25 |
+
"intermediate_size": 4096,
|
26 |
+
"model_type": "clip_vision_model",
|
27 |
+
"num_attention_heads": 16,
|
28 |
+
"num_hidden_layers": 24,
|
29 |
+
"patch_size": 14,
|
30 |
+
"projection_dim": 768
|
31 |
+
}
|
32 |
+
}
|