sbchoi commited on
Commit
a24f65b
1 Parent(s): 75f88e2

Add config from convert_rt_detr_original_pytorch_checkpoint_to_pytorch.py

Browse files
Files changed (1) hide show
  1. config.json +26 -35
config.json CHANGED
@@ -10,6 +10,32 @@
10
  ],
11
  "attention_dropout": 0.0,
12
  "auxiliary_loss": true,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
  "batch_norm_eps": 1e-05,
14
  "box_noise_scale": 1.0,
15
  "d_model": 256,
@@ -23,17 +49,8 @@
23
  ],
24
  "decoder_layers": 3,
25
  "decoder_n_points": 4,
26
- "depths": [
27
- 2,
28
- 2,
29
- 2,
30
- 2
31
- ],
32
  "disable_custom_kernels": true,
33
- "downsample_in_bottleneck": false,
34
- "downsample_in_first_stage": false,
35
  "dropout": 0.0,
36
- "embedding_size": 64,
37
  "encode_proj_layers": [
38
  2
39
  ],
@@ -56,14 +73,7 @@
56
  ],
57
  "focal_loss_alpha": 0.75,
58
  "focal_loss_gamma": 2.0,
59
- "hidden_act": "relu",
60
  "hidden_expansion": 0.5,
61
- "hidden_sizes": [
62
- 64,
63
- 128,
64
- 256,
65
- 512
66
- ],
67
  "id2label": {
68
  "0": "person",
69
  "1": "bicycle",
@@ -232,7 +242,6 @@
232
  },
233
  "label_noise_ratio": 0.5,
234
  "layer_norm_eps": 1e-05,
235
- "layer_type": "basic",
236
  "learn_initial_query": false,
237
  "matcher_alpha": 0.25,
238
  "matcher_bbox_cost": 5.0,
@@ -241,28 +250,10 @@
241
  "matcher_giou_cost": 2.0,
242
  "model_type": "rt_detr",
243
  "normalize_before": false,
244
- "num_channels": 3,
245
  "num_denoising": 100,
246
  "num_feature_levels": 3,
247
  "num_queries": 300,
248
- "out_features": [
249
- "stage2",
250
- "stage3",
251
- "stage4"
252
- ],
253
- "out_indices": [
254
- 2,
255
- 3,
256
- 4
257
- ],
258
  "positional_encoding_temperature": 10000,
259
- "stage_names": [
260
- "stem",
261
- "stage1",
262
- "stage2",
263
- "stage3",
264
- "stage4"
265
- ],
266
  "torch_dtype": "float32",
267
  "transformers_version": "4.42.0.dev0",
268
  "use_focal_loss": true,
 
10
  ],
11
  "attention_dropout": 0.0,
12
  "auxiliary_loss": true,
13
+ "backbone_config": {
14
+ "depths": [
15
+ 2,
16
+ 2,
17
+ 2,
18
+ 2
19
+ ],
20
+ "hidden_sizes": [
21
+ 64,
22
+ 128,
23
+ 256,
24
+ 512
25
+ ],
26
+ "layer_type": "basic",
27
+ "model_type": "rtdetr-resnet",
28
+ "out_features": [
29
+ "stage2",
30
+ "stage3",
31
+ "stage4"
32
+ ],
33
+ "out_indices": [
34
+ 2,
35
+ 3,
36
+ 4
37
+ ]
38
+ },
39
  "batch_norm_eps": 1e-05,
40
  "box_noise_scale": 1.0,
41
  "d_model": 256,
 
49
  ],
50
  "decoder_layers": 3,
51
  "decoder_n_points": 4,
 
 
 
 
 
 
52
  "disable_custom_kernels": true,
 
 
53
  "dropout": 0.0,
 
54
  "encode_proj_layers": [
55
  2
56
  ],
 
73
  ],
74
  "focal_loss_alpha": 0.75,
75
  "focal_loss_gamma": 2.0,
 
76
  "hidden_expansion": 0.5,
 
 
 
 
 
 
77
  "id2label": {
78
  "0": "person",
79
  "1": "bicycle",
 
242
  },
243
  "label_noise_ratio": 0.5,
244
  "layer_norm_eps": 1e-05,
 
245
  "learn_initial_query": false,
246
  "matcher_alpha": 0.25,
247
  "matcher_bbox_cost": 5.0,
 
250
  "matcher_giou_cost": 2.0,
251
  "model_type": "rt_detr",
252
  "normalize_before": false,
 
253
  "num_denoising": 100,
254
  "num_feature_levels": 3,
255
  "num_queries": 300,
 
 
 
 
 
 
 
 
 
 
256
  "positional_encoding_temperature": 10000,
 
 
 
 
 
 
 
257
  "torch_dtype": "float32",
258
  "transformers_version": "4.42.0.dev0",
259
  "use_focal_loss": true,