Update tiny models for Sam2HieraDetModel
Browse files- config.json +57 -0
- model.safetensors +3 -0
- preprocessor_config.json +31 -0
- video_preprocessor_config.json +28 -0
config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Sam2HieraDetModel"
|
| 4 |
+
],
|
| 5 |
+
"blocks_per_stage": [
|
| 6 |
+
1,
|
| 7 |
+
2,
|
| 8 |
+
7,
|
| 9 |
+
2
|
| 10 |
+
],
|
| 11 |
+
"dtype": "float32",
|
| 12 |
+
"embed_dim_per_stage": [
|
| 13 |
+
12,
|
| 14 |
+
24,
|
| 15 |
+
48,
|
| 16 |
+
96
|
| 17 |
+
],
|
| 18 |
+
"global_attention_blocks": [
|
| 19 |
+
5,
|
| 20 |
+
7,
|
| 21 |
+
9
|
| 22 |
+
],
|
| 23 |
+
"hidden_act": "gelu",
|
| 24 |
+
"hidden_size": 12,
|
| 25 |
+
"image_size": 128,
|
| 26 |
+
"initializer_range": 0.02,
|
| 27 |
+
"layer_norm_eps": 1e-06,
|
| 28 |
+
"mlp_ratio": 4.0,
|
| 29 |
+
"model_type": "sam2_hiera_det_model",
|
| 30 |
+
"num_attention_heads": 1,
|
| 31 |
+
"num_attention_heads_per_stage": [
|
| 32 |
+
1,
|
| 33 |
+
2,
|
| 34 |
+
4,
|
| 35 |
+
8
|
| 36 |
+
],
|
| 37 |
+
"num_channels": 3,
|
| 38 |
+
"num_query_pool_stages": 3,
|
| 39 |
+
"patch_kernel_size": 7,
|
| 40 |
+
"patch_padding": 3,
|
| 41 |
+
"patch_stride": 4,
|
| 42 |
+
"query_stride": [
|
| 43 |
+
2,
|
| 44 |
+
2
|
| 45 |
+
],
|
| 46 |
+
"transformers_version": "5.16.0.dev0",
|
| 47 |
+
"window_positional_embedding_background_size": [
|
| 48 |
+
7,
|
| 49 |
+
7
|
| 50 |
+
],
|
| 51 |
+
"window_size_per_stage": [
|
| 52 |
+
8,
|
| 53 |
+
4,
|
| 54 |
+
14,
|
| 55 |
+
7
|
| 56 |
+
]
|
| 57 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6fe928e0ef6875b0b11d54dd5dabcd14317879e4211ef92a531cd7dcc2da6953
|
| 3 |
+
size 1729696
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 128,
|
| 4 |
+
"width": 128
|
| 5 |
+
},
|
| 6 |
+
"do_convert_rgb": true,
|
| 7 |
+
"do_normalize": true,
|
| 8 |
+
"do_rescale": true,
|
| 9 |
+
"do_resize": true,
|
| 10 |
+
"image_mean": [
|
| 11 |
+
0.485,
|
| 12 |
+
0.456,
|
| 13 |
+
0.406
|
| 14 |
+
],
|
| 15 |
+
"image_processor_type": "Sam2ImageProcessor",
|
| 16 |
+
"image_std": [
|
| 17 |
+
0.229,
|
| 18 |
+
0.224,
|
| 19 |
+
0.225
|
| 20 |
+
],
|
| 21 |
+
"mask_size": {
|
| 22 |
+
"height": 256,
|
| 23 |
+
"width": 256
|
| 24 |
+
},
|
| 25 |
+
"resample": 2,
|
| 26 |
+
"rescale_factor": 0.00392156862745098,
|
| 27 |
+
"size": {
|
| 28 |
+
"height": 128,
|
| 29 |
+
"width": 128
|
| 30 |
+
}
|
| 31 |
+
}
|
video_preprocessor_config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 128,
|
| 4 |
+
"width": 128
|
| 5 |
+
},
|
| 6 |
+
"do_convert_rgb": true,
|
| 7 |
+
"do_normalize": true,
|
| 8 |
+
"do_rescale": true,
|
| 9 |
+
"do_resize": true,
|
| 10 |
+
"image_mean": [
|
| 11 |
+
0.485,
|
| 12 |
+
0.456,
|
| 13 |
+
0.406
|
| 14 |
+
],
|
| 15 |
+
"image_std": [
|
| 16 |
+
0.229,
|
| 17 |
+
0.224,
|
| 18 |
+
0.225
|
| 19 |
+
],
|
| 20 |
+
"resample": 2,
|
| 21 |
+
"rescale_factor": 0.00392156862745098,
|
| 22 |
+
"return_metadata": false,
|
| 23 |
+
"size": {
|
| 24 |
+
"height": 128,
|
| 25 |
+
"width": 128
|
| 26 |
+
},
|
| 27 |
+
"video_processor_type": "Sam2VideoVideoProcessor"
|
| 28 |
+
}
|