Update tiny models for PPDocLayoutV3Model
Browse files- config.json +192 -0
- model.safetensors +3 -0
- preprocessor_config.json +26 -0
config.json
ADDED
|
@@ -0,0 +1,192 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_dropout": 0.0,
|
| 3 |
+
"activation_function": "silu",
|
| 4 |
+
"anchor_image_size": null,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"PPDocLayoutV3Model"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"backbone_config": {
|
| 10 |
+
"arch": "L",
|
| 11 |
+
"depths": [
|
| 12 |
+
3,
|
| 13 |
+
4,
|
| 14 |
+
6,
|
| 15 |
+
3
|
| 16 |
+
],
|
| 17 |
+
"embedding_size": 64,
|
| 18 |
+
"freeze_at": 0,
|
| 19 |
+
"freeze_norm": true,
|
| 20 |
+
"freeze_stem_only": true,
|
| 21 |
+
"hidden_act": "relu",
|
| 22 |
+
"hidden_sizes": [
|
| 23 |
+
32,
|
| 24 |
+
32,
|
| 25 |
+
32,
|
| 26 |
+
32
|
| 27 |
+
],
|
| 28 |
+
"initializer_range": 0.02,
|
| 29 |
+
"lr_mult_list": [
|
| 30 |
+
0,
|
| 31 |
+
0.05,
|
| 32 |
+
0.05,
|
| 33 |
+
0.05,
|
| 34 |
+
0.05
|
| 35 |
+
],
|
| 36 |
+
"model_type": "hgnet_v2",
|
| 37 |
+
"num_channels": 3,
|
| 38 |
+
"out_features": [
|
| 39 |
+
"stage1",
|
| 40 |
+
"stage2",
|
| 41 |
+
"stage3",
|
| 42 |
+
"stage4"
|
| 43 |
+
],
|
| 44 |
+
"out_indices": [
|
| 45 |
+
1,
|
| 46 |
+
2,
|
| 47 |
+
3,
|
| 48 |
+
4
|
| 49 |
+
],
|
| 50 |
+
"return_idx": [
|
| 51 |
+
0,
|
| 52 |
+
1,
|
| 53 |
+
2,
|
| 54 |
+
3
|
| 55 |
+
],
|
| 56 |
+
"stage_downsample": [
|
| 57 |
+
false,
|
| 58 |
+
true,
|
| 59 |
+
true,
|
| 60 |
+
true
|
| 61 |
+
],
|
| 62 |
+
"stage_downsample_strides": [
|
| 63 |
+
2,
|
| 64 |
+
2,
|
| 65 |
+
2,
|
| 66 |
+
2
|
| 67 |
+
],
|
| 68 |
+
"stage_in_channels": [
|
| 69 |
+
32,
|
| 70 |
+
32,
|
| 71 |
+
32,
|
| 72 |
+
32
|
| 73 |
+
],
|
| 74 |
+
"stage_kernel_size": [
|
| 75 |
+
3,
|
| 76 |
+
3,
|
| 77 |
+
5,
|
| 78 |
+
5
|
| 79 |
+
],
|
| 80 |
+
"stage_light_block": [
|
| 81 |
+
false,
|
| 82 |
+
false,
|
| 83 |
+
true,
|
| 84 |
+
true
|
| 85 |
+
],
|
| 86 |
+
"stage_mid_channels": [
|
| 87 |
+
32,
|
| 88 |
+
32,
|
| 89 |
+
32,
|
| 90 |
+
32
|
| 91 |
+
],
|
| 92 |
+
"stage_names": [
|
| 93 |
+
"stem",
|
| 94 |
+
"stage1",
|
| 95 |
+
"stage2",
|
| 96 |
+
"stage3",
|
| 97 |
+
"stage4"
|
| 98 |
+
],
|
| 99 |
+
"stage_num_blocks": [
|
| 100 |
+
1,
|
| 101 |
+
1,
|
| 102 |
+
3,
|
| 103 |
+
1
|
| 104 |
+
],
|
| 105 |
+
"stage_numb_of_layers": [
|
| 106 |
+
6,
|
| 107 |
+
6,
|
| 108 |
+
6,
|
| 109 |
+
6
|
| 110 |
+
],
|
| 111 |
+
"stage_out_channels": [
|
| 112 |
+
32,
|
| 113 |
+
32,
|
| 114 |
+
32,
|
| 115 |
+
32
|
| 116 |
+
],
|
| 117 |
+
"stem_channels": [
|
| 118 |
+
3,
|
| 119 |
+
32,
|
| 120 |
+
32
|
| 121 |
+
],
|
| 122 |
+
"stem_strides": [
|
| 123 |
+
2,
|
| 124 |
+
1,
|
| 125 |
+
1,
|
| 126 |
+
2,
|
| 127 |
+
1
|
| 128 |
+
],
|
| 129 |
+
"use_learnable_affine_block": false
|
| 130 |
+
},
|
| 131 |
+
"batch_norm_eps": 1e-05,
|
| 132 |
+
"box_noise_scale": 1.0,
|
| 133 |
+
"d_model": 32,
|
| 134 |
+
"decoder_activation_function": "relu",
|
| 135 |
+
"decoder_attention_heads": 2,
|
| 136 |
+
"decoder_ffn_dim": 8,
|
| 137 |
+
"decoder_in_channels": [
|
| 138 |
+
32,
|
| 139 |
+
32,
|
| 140 |
+
32
|
| 141 |
+
],
|
| 142 |
+
"decoder_layers": 2,
|
| 143 |
+
"decoder_n_points": 4,
|
| 144 |
+
"disable_custom_kernels": true,
|
| 145 |
+
"dropout": 0.0,
|
| 146 |
+
"dtype": "float32",
|
| 147 |
+
"encode_proj_layers": [
|
| 148 |
+
2
|
| 149 |
+
],
|
| 150 |
+
"encoder_activation_function": "gelu",
|
| 151 |
+
"encoder_attention_heads": 2,
|
| 152 |
+
"encoder_ffn_dim": 64,
|
| 153 |
+
"encoder_hidden_dim": 32,
|
| 154 |
+
"encoder_in_channels": [
|
| 155 |
+
20,
|
| 156 |
+
30,
|
| 157 |
+
40
|
| 158 |
+
],
|
| 159 |
+
"encoder_layers": 1,
|
| 160 |
+
"eval_size": null,
|
| 161 |
+
"feat_strides": [
|
| 162 |
+
8,
|
| 163 |
+
16,
|
| 164 |
+
32
|
| 165 |
+
],
|
| 166 |
+
"freeze_backbone_batch_norms": true,
|
| 167 |
+
"global_pointer_head_size": 64,
|
| 168 |
+
"gp_dropout_value": 0.1,
|
| 169 |
+
"hidden_expansion": 1.0,
|
| 170 |
+
"image_size": 128,
|
| 171 |
+
"initializer_bias_prior_prob": null,
|
| 172 |
+
"initializer_range": 0.01,
|
| 173 |
+
"is_encoder_decoder": true,
|
| 174 |
+
"label_noise_ratio": 0.5,
|
| 175 |
+
"layer_norm_eps": 1e-05,
|
| 176 |
+
"learn_initial_query": false,
|
| 177 |
+
"mask_enhanced": true,
|
| 178 |
+
"mask_feature_channels": [
|
| 179 |
+
32,
|
| 180 |
+
32
|
| 181 |
+
],
|
| 182 |
+
"model_type": "pp_doclayout_v3",
|
| 183 |
+
"normalize_before": false,
|
| 184 |
+
"num_denoising": 0,
|
| 185 |
+
"num_feature_levels": 3,
|
| 186 |
+
"num_prototypes": 32,
|
| 187 |
+
"num_queries": 30,
|
| 188 |
+
"positional_encoding_temperature": 10000,
|
| 189 |
+
"tie_word_embeddings": true,
|
| 190 |
+
"transformers_version": "5.16.0.dev0",
|
| 191 |
+
"x4_feat_dim": 32
|
| 192 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d04d1284dd0549db0bf1fc98407e061c11a393142546e341646c8c7e64d31b48
|
| 3 |
+
size 2078304
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 128,
|
| 4 |
+
"width": 128
|
| 5 |
+
},
|
| 6 |
+
"do_normalize": true,
|
| 7 |
+
"do_rescale": true,
|
| 8 |
+
"do_resize": true,
|
| 9 |
+
"image_mean": [
|
| 10 |
+
0,
|
| 11 |
+
0,
|
| 12 |
+
0
|
| 13 |
+
],
|
| 14 |
+
"image_processor_type": "PPDocLayoutV3ImageProcessor",
|
| 15 |
+
"image_std": [
|
| 16 |
+
1,
|
| 17 |
+
1,
|
| 18 |
+
1
|
| 19 |
+
],
|
| 20 |
+
"resample": 3,
|
| 21 |
+
"rescale_factor": 0.00392156862745098,
|
| 22 |
+
"size": {
|
| 23 |
+
"height": 128,
|
| 24 |
+
"width": 128
|
| 25 |
+
}
|
| 26 |
+
}
|