Upload model
Browse files- config.json +47 -4
config.json
CHANGED
|
@@ -4,15 +4,58 @@
|
|
| 4 |
],
|
| 5 |
"initializer_range": 0.02,
|
| 6 |
"mask_decoder_config": {
|
| 7 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
},
|
| 9 |
"model_type": "sam_hq",
|
| 10 |
"prompt_encoder_config": {
|
| 11 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
},
|
| 13 |
"torch_dtype": "float32",
|
| 14 |
-
"transformers_version": "4.
|
| 15 |
"vision_config": {
|
| 16 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
}
|
| 18 |
}
|
|
|
|
| 4 |
],
|
| 5 |
"initializer_range": 0.02,
|
| 6 |
"mask_decoder_config": {
|
| 7 |
+
"attention_downsample_rate": 2,
|
| 8 |
+
"hidden_act": "relu",
|
| 9 |
+
"hidden_size": 256,
|
| 10 |
+
"iou_head_depth": 3,
|
| 11 |
+
"iou_head_hidden_dim": 256,
|
| 12 |
+
"layer_norm_eps": 1e-06,
|
| 13 |
+
"mlp_dim": 2048,
|
| 14 |
+
"model_type": "",
|
| 15 |
+
"num_attention_heads": 8,
|
| 16 |
+
"num_hidden_layers": 2,
|
| 17 |
+
"num_multimask_outputs": 3,
|
| 18 |
+
"vit_dim": 768
|
| 19 |
},
|
| 20 |
"model_type": "sam_hq",
|
| 21 |
"prompt_encoder_config": {
|
| 22 |
+
"hidden_act": "gelu",
|
| 23 |
+
"hidden_size": 256,
|
| 24 |
+
"image_embedding_size": 64,
|
| 25 |
+
"image_size": 1024,
|
| 26 |
+
"layer_norm_eps": 1e-06,
|
| 27 |
+
"mask_input_channels": 16,
|
| 28 |
+
"model_type": "",
|
| 29 |
+
"num_point_embeddings": 4,
|
| 30 |
+
"patch_size": 16
|
| 31 |
},
|
| 32 |
"torch_dtype": "float32",
|
| 33 |
+
"transformers_version": "4.52.0.dev0",
|
| 34 |
"vision_config": {
|
| 35 |
+
"attention_dropout": 0.0,
|
| 36 |
+
"global_attn_indexes": [
|
| 37 |
+
2,
|
| 38 |
+
5,
|
| 39 |
+
8,
|
| 40 |
+
11
|
| 41 |
+
],
|
| 42 |
+
"hidden_act": "gelu",
|
| 43 |
+
"hidden_size": 768,
|
| 44 |
+
"image_size": 1024,
|
| 45 |
+
"initializer_range": 1e-10,
|
| 46 |
+
"layer_norm_eps": 1e-06,
|
| 47 |
+
"mlp_dim": 3072,
|
| 48 |
+
"mlp_ratio": 4.0,
|
| 49 |
+
"model_type": "sam_hq_vision_model",
|
| 50 |
+
"num_attention_heads": 12,
|
| 51 |
+
"num_channels": 3,
|
| 52 |
+
"num_hidden_layers": 12,
|
| 53 |
+
"num_pos_feats": 128,
|
| 54 |
+
"output_channels": 256,
|
| 55 |
+
"patch_size": 16,
|
| 56 |
+
"qkv_bias": true,
|
| 57 |
+
"use_abs_pos": true,
|
| 58 |
+
"use_rel_pos": true,
|
| 59 |
+
"window_size": 14
|
| 60 |
}
|
| 61 |
}
|