Upload visual_encoder/config.json with huggingface_hub
Browse files- visual_encoder/config.json +22 -0
visual_encoder/config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "apple/DFN5B-CLIP-ViT-H-14-378",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"PikaVidEncoder"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"hidden_act": "quick_gelu",
|
| 8 |
+
"hidden_size": 1280,
|
| 9 |
+
"image_size": 378,
|
| 10 |
+
"initializer_factor": 1.0,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 5120,
|
| 13 |
+
"layer_norm_eps": 1e-05,
|
| 14 |
+
"model_type": "clip_vision_model",
|
| 15 |
+
"num_attention_heads": 16,
|
| 16 |
+
"num_channels": 3,
|
| 17 |
+
"num_hidden_layers": 32,
|
| 18 |
+
"patch_size": 14,
|
| 19 |
+
"projection_dim": 512,
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.44.2"
|
| 22 |
+
}
|