svetadomoi commited on
Commit
5e23b66
·
verified ·
1 Parent(s): ae04ada

End of training

Browse files
README.md ADDED
@@ -0,0 +1,90 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: apache-2.0
4
+ base_model: PekingU/rtdetr_v2_r34vd
5
+ tags:
6
+ - generated_from_trainer
7
+ model-index:
8
+ - name: rtdetr-v2-r34-cppe5-finetune-2
9
+ results: []
10
+ ---
11
+
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
+
15
+ # rtdetr-v2-r34-cppe5-finetune-2
16
+
17
+ This model is a fine-tuned version of [PekingU/rtdetr_v2_r34vd](https://huggingface.co/PekingU/rtdetr_v2_r34vd) on an unknown dataset.
18
+ It achieves the following results on the evaluation set:
19
+ - Loss: 8.3132
20
+ - Map: 0.3055
21
+ - Map 50: 0.5372
22
+ - Map 75: 0.3008
23
+ - Map Small: 0.1094
24
+ - Map Medium: 0.241
25
+ - Map Large: 0.3888
26
+ - Mar 1: 0.2863
27
+ - Mar 10: 0.4999
28
+ - Mar 100: 0.578
29
+ - Mar Small: 0.3884
30
+ - Mar Medium: 0.4832
31
+ - Mar Large: 0.7099
32
+ - Map Coverall: 0.5596
33
+ - Mar 100 Coverall: 0.7275
34
+ - Map Face Shield: 0.2013
35
+ - Mar 100 Face Shield: 0.6342
36
+ - Map Gloves: 0.2742
37
+ - Mar 100 Gloves: 0.5192
38
+ - Map Goggles: 0.146
39
+ - Mar 100 Goggles: 0.4769
40
+ - Map Mask: 0.3462
41
+ - Mar 100 Mask: 0.5324
42
+
43
+ ## Model description
44
+
45
+ More information needed
46
+
47
+ ## Intended uses & limitations
48
+
49
+ More information needed
50
+
51
+ ## Training and evaluation data
52
+
53
+ More information needed
54
+
55
+ ## Training procedure
56
+
57
+ ### Training hyperparameters
58
+
59
+ The following hyperparameters were used during training:
60
+ - learning_rate: 5e-05
61
+ - train_batch_size: 8
62
+ - eval_batch_size: 8
63
+ - seed: 42
64
+ - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
65
+ - lr_scheduler_type: linear
66
+ - lr_scheduler_warmup_steps: 300
67
+ - num_epochs: 10
68
+
69
+ ### Training results
70
+
71
+ | Training Loss | Epoch | Step | Validation Loss | Map | Map 50 | Map 75 | Map Small | Map Medium | Map Large | Mar 1 | Mar 10 | Mar 100 | Mar Small | Mar Medium | Mar Large | Map Coverall | Mar 100 Coverall | Map Face Shield | Mar 100 Face Shield | Map Gloves | Mar 100 Gloves | Map Goggles | Mar 100 Goggles | Map Mask | Mar 100 Mask |
72
+ |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:---------:|:----------:|:---------:|:------:|:------:|:-------:|:---------:|:----------:|:---------:|:------------:|:----------------:|:---------------:|:-------------------:|:----------:|:--------------:|:-----------:|:---------------:|:--------:|:------------:|
73
+ | No log | 1.0 | 107 | 21.6622 | 0.0574 | 0.1029 | 0.0519 | 0.0001 | 0.0074 | 0.0702 | 0.0808 | 0.1764 | 0.2386 | 0.0255 | 0.1314 | 0.3738 | 0.263 | 0.6131 | 0.0098 | 0.1962 | 0.0056 | 0.1446 | 0.0002 | 0.0692 | 0.0085 | 0.1698 |
74
+ | No log | 2.0 | 214 | 12.3093 | 0.1473 | 0.2769 | 0.1347 | 0.04 | 0.108 | 0.2006 | 0.1941 | 0.3837 | 0.4464 | 0.1571 | 0.3411 | 0.6232 | 0.3946 | 0.6973 | 0.0871 | 0.438 | 0.067 | 0.392 | 0.0198 | 0.3308 | 0.1682 | 0.3742 |
75
+ | No log | 3.0 | 321 | 9.8181 | 0.2151 | 0.3903 | 0.2003 | 0.0862 | 0.1856 | 0.2692 | 0.2513 | 0.4746 | 0.5437 | 0.2899 | 0.4531 | 0.6932 | 0.4504 | 0.7221 | 0.1277 | 0.5734 | 0.1439 | 0.471 | 0.0493 | 0.4523 | 0.304 | 0.4996 |
76
+ | No log | 4.0 | 428 | 9.0262 | 0.2471 | 0.44 | 0.2372 | 0.0808 | 0.2084 | 0.3213 | 0.2685 | 0.4722 | 0.5494 | 0.2294 | 0.4611 | 0.7006 | 0.5037 | 0.7387 | 0.147 | 0.5823 | 0.2081 | 0.4938 | 0.0661 | 0.4338 | 0.3107 | 0.4982 |
77
+ | 27.2158 | 5.0 | 535 | 8.6126 | 0.276 | 0.4857 | 0.2769 | 0.104 | 0.2124 | 0.3508 | 0.2741 | 0.4908 | 0.5667 | 0.3254 | 0.4704 | 0.6999 | 0.5335 | 0.7261 | 0.156 | 0.6139 | 0.2473 | 0.5156 | 0.0958 | 0.4523 | 0.3474 | 0.5253 |
78
+ | 27.2158 | 6.0 | 642 | 8.4669 | 0.2826 | 0.5022 | 0.2795 | 0.095 | 0.2191 | 0.3575 | 0.2766 | 0.4948 | 0.5704 | 0.3249 | 0.461 | 0.7107 | 0.5455 | 0.7356 | 0.1524 | 0.6215 | 0.2571 | 0.5125 | 0.1208 | 0.4646 | 0.3372 | 0.5178 |
79
+ | 27.2158 | 7.0 | 749 | 8.3188 | 0.3003 | 0.5202 | 0.2958 | 0.1107 | 0.2348 | 0.3887 | 0.2879 | 0.5044 | 0.5766 | 0.3707 | 0.4823 | 0.7142 | 0.5545 | 0.7288 | 0.2019 | 0.6329 | 0.2655 | 0.5263 | 0.1341 | 0.4631 | 0.3453 | 0.532 |
80
+ | 27.2158 | 8.0 | 856 | 8.3084 | 0.2972 | 0.5265 | 0.2912 | 0.107 | 0.2409 | 0.3732 | 0.2811 | 0.5029 | 0.5811 | 0.3453 | 0.4772 | 0.7138 | 0.5617 | 0.7351 | 0.1649 | 0.6367 | 0.2773 | 0.5192 | 0.1337 | 0.4769 | 0.3485 | 0.5378 |
81
+ | 27.2158 | 9.0 | 963 | 8.2764 | 0.3064 | 0.5313 | 0.3049 | 0.1068 | 0.2421 | 0.3871 | 0.284 | 0.5073 | 0.5802 | 0.373 | 0.4783 | 0.7108 | 0.5621 | 0.7333 | 0.182 | 0.6291 | 0.2765 | 0.5299 | 0.1631 | 0.4738 | 0.3483 | 0.5347 |
82
+ | 11.8005 | 10.0 | 1070 | 8.3132 | 0.3055 | 0.5372 | 0.3008 | 0.1094 | 0.241 | 0.3888 | 0.2863 | 0.4999 | 0.578 | 0.3884 | 0.4832 | 0.7099 | 0.5596 | 0.7275 | 0.2013 | 0.6342 | 0.2742 | 0.5192 | 0.146 | 0.4769 | 0.3462 | 0.5324 |
83
+
84
+
85
+ ### Framework versions
86
+
87
+ - Transformers 4.51.3
88
+ - Pytorch 2.6.0+cu124
89
+ - Datasets 3.5.1
90
+ - Tokenizers 0.21.1
config.json ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "activation_function": "silu",
4
+ "anchor_image_size": null,
5
+ "architectures": [
6
+ "RTDetrV2ForObjectDetection"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "auxiliary_loss": true,
10
+ "backbone": null,
11
+ "backbone_config": {
12
+ "depths": [
13
+ 3,
14
+ 4,
15
+ 6,
16
+ 3
17
+ ],
18
+ "downsample_in_bottleneck": false,
19
+ "downsample_in_first_stage": false,
20
+ "embedding_size": 64,
21
+ "hidden_act": "relu",
22
+ "hidden_sizes": [
23
+ 64,
24
+ 128,
25
+ 256,
26
+ 512
27
+ ],
28
+ "layer_type": "basic",
29
+ "model_type": "rt_detr_resnet",
30
+ "num_channels": 3,
31
+ "out_features": [
32
+ "stage2",
33
+ "stage3",
34
+ "stage4"
35
+ ],
36
+ "out_indices": [
37
+ 2,
38
+ 3,
39
+ 4
40
+ ],
41
+ "stage_names": [
42
+ "stem",
43
+ "stage1",
44
+ "stage2",
45
+ "stage3",
46
+ "stage4"
47
+ ]
48
+ },
49
+ "backbone_kwargs": null,
50
+ "batch_norm_eps": 1e-05,
51
+ "box_noise_scale": 1.0,
52
+ "d_model": 256,
53
+ "decoder_activation_function": "relu",
54
+ "decoder_attention_heads": 8,
55
+ "decoder_ffn_dim": 1024,
56
+ "decoder_in_channels": [
57
+ 256,
58
+ 256,
59
+ 256
60
+ ],
61
+ "decoder_layers": 4,
62
+ "decoder_method": "default",
63
+ "decoder_n_levels": 3,
64
+ "decoder_n_points": 4,
65
+ "decoder_offset_scale": 0.5,
66
+ "disable_custom_kernels": true,
67
+ "dropout": 0.0,
68
+ "encode_proj_layers": [
69
+ 2
70
+ ],
71
+ "encoder_activation_function": "gelu",
72
+ "encoder_attention_heads": 8,
73
+ "encoder_ffn_dim": 1024,
74
+ "encoder_hidden_dim": 256,
75
+ "encoder_in_channels": [
76
+ 128,
77
+ 256,
78
+ 512
79
+ ],
80
+ "encoder_layers": 1,
81
+ "eos_coefficient": 0.0001,
82
+ "eval_size": null,
83
+ "feat_strides": [
84
+ 8,
85
+ 16,
86
+ 32
87
+ ],
88
+ "focal_loss_alpha": 0.75,
89
+ "focal_loss_gamma": 2.0,
90
+ "freeze_backbone_batch_norms": true,
91
+ "hidden_expansion": 0.5,
92
+ "id2label": {
93
+ "0": "Coverall",
94
+ "1": "Face_Shield",
95
+ "2": "Gloves",
96
+ "3": "Goggles",
97
+ "4": "Mask"
98
+ },
99
+ "initializer_bias_prior_prob": null,
100
+ "initializer_range": 0.01,
101
+ "is_encoder_decoder": true,
102
+ "label2id": {
103
+ "Coverall": 0,
104
+ "Face_Shield": 1,
105
+ "Gloves": 2,
106
+ "Goggles": 3,
107
+ "Mask": 4
108
+ },
109
+ "label_noise_ratio": 0.5,
110
+ "layer_norm_eps": 1e-05,
111
+ "learn_initial_query": false,
112
+ "matcher_alpha": 0.25,
113
+ "matcher_bbox_cost": 5.0,
114
+ "matcher_class_cost": 2.0,
115
+ "matcher_gamma": 2.0,
116
+ "matcher_giou_cost": 2.0,
117
+ "model_type": "rt_detr_v2",
118
+ "normalize_before": false,
119
+ "num_denoising": 100,
120
+ "num_feature_levels": 3,
121
+ "num_queries": 300,
122
+ "positional_encoding_temperature": 10000,
123
+ "torch_dtype": "float32",
124
+ "transformers_version": "4.51.3",
125
+ "use_focal_loss": true,
126
+ "use_pretrained_backbone": false,
127
+ "use_timm_backbone": false,
128
+ "weight_loss_bbox": 5.0,
129
+ "weight_loss_giou": 2.0,
130
+ "weight_loss_vfl": 1.0,
131
+ "with_box_refine": true
132
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c81bae4c43865cffc76a0b86420f42ae8ccb0219aff7679eee871d045e62384
3
+ size 125517084
preprocessor_config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": null,
3
+ "data_format": "channels_first",
4
+ "default_to_square": false,
5
+ "device": null,
6
+ "do_center_crop": null,
7
+ "do_convert_annotations": true,
8
+ "do_convert_rgb": null,
9
+ "do_normalize": false,
10
+ "do_pad": false,
11
+ "do_rescale": true,
12
+ "do_resize": true,
13
+ "format": "coco_detection",
14
+ "image_mean": [
15
+ 0.485,
16
+ 0.456,
17
+ 0.406
18
+ ],
19
+ "image_processor_type": "RTDetrImageProcessorFast",
20
+ "image_std": [
21
+ 0.229,
22
+ 0.224,
23
+ 0.225
24
+ ],
25
+ "input_data_format": null,
26
+ "pad_size": null,
27
+ "resample": 2,
28
+ "rescale_factor": 0.00392156862745098,
29
+ "return_segmentation_masks": null,
30
+ "return_tensors": null,
31
+ "size": {
32
+ "height": 640,
33
+ "width": 640
34
+ }
35
+ }
runs/May05_12-14-14_b46bab9aa7d4/events.out.tfevents.1746447256.b46bab9aa7d4.1949.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:560e5219e3f336c010b41eabb496587b536b9d120a946a9e02404166120cce0e
3
+ size 22458
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1ccf21fc61a97787944e0fe3ef274a35163b5458e35d1280af6c148d679b38b
3
+ size 5368