Hugging Face's logo

brandyDolly
/
segformer-finetuned-sidewalk-10k-steps

Image Segmentation
Transformers
TensorBoard
Safetensors
upernet
vision
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
brandyDolly commited on
Commit
577c4dd
·
verified ·
1 Parent(s): e796d5d

Training in progress, epoch 0

Browse files
Files changed (6) hide show
  1. config.json +123 -102
  2. model.safetensors +2 -2
  3. runs/Nov28_08-36-44_d2a3b59c97a0/events.out.tfevents.1732783017.d2a3b59c97a0.1155.0 +3 -0
  4. runs/Nov28_08-41-09_d2a3b59c97a0/events.out.tfevents.1732783276.d2a3b59c97a0.2326.0 +3 -0
  5. runs/Nov28_08-42-41_d2a3b59c97a0/events.out.tfevents.1732783365.d2a3b59c97a0.2788.0 +3 -0
  6. training_args.bin +1 -1
config.json CHANGED
@@ -1,116 +1,137 @@
1
  {
2
- "_name_or_path": "nvidia/mit-b0",
3
  "architectures": [
4
- "SegformerForSemanticSegmentation"
5
- ],
6
- "attention_probs_dropout_prob": 0.0,
7
- "classifier_dropout_prob": 0.1,
8
- "decoder_hidden_size": 256,
9
- "depths": [
10
- 2,
11
- 2,
12
- 2,
13
- 2
14
- ],
15
- "downsampling_rates": [
16
- 1,
17
- 4,
18
- 8,
19
- 16
20
- ],
21
- "drop_path_rate": 0.1,
22
- "hidden_act": "gelu",
23
- "hidden_dropout_prob": 0.0,
24
- "hidden_sizes": [
25
- 32,
26
- 64,
27
- 160,
28
- 256
29
  ],
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30
  "id2label": {
31
- "0": "background",
32
- "1": "aeroplane",
33
- "2": "bicycle",
34
- "3": "bird",
35
- "4": "boat",
36
- "5": "bottle",
37
- "6": "bus",
38
- "7": "car",
39
- "8": "cat",
40
- "9": "chair",
41
- "10": "cow",
42
- "11": "diningtable",
43
- "12": "dog",
44
- "13": "horse",
45
- "14": "motorbike",
46
- "15": "person",
47
- "16": "pottedplant",
48
- "17": "sheep",
49
- "18": "sofa",
50
- "19": "train",
51
- "20": "tvmonitor"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52
  },
53
- "image_size": 224,
54
  "initializer_range": 0.02,
55
  "label2id": {
56
- "aeroplane": "1",
57
- "background": "0",
58
- "bicycle": "2",
59
- "bird": "3",
60
- "boat": "4",
61
- "bottle": "5",
62
- "bus": "6",
63
- "car": "7",
64
- "cat": "8",
65
- "chair": "9",
66
- "cow": "10",
67
- "diningtable": "11",
68
- "dog": "12",
69
- "horse": "13",
70
- "motorbike": "14",
71
- "person": "15",
72
- "pottedplant": "16",
73
- "sheep": "17",
74
- "sofa": "18",
75
- "train": "19",
76
- "tvmonitor": "20"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
77
  },
78
- "layer_norm_eps": 1e-06,
79
- "mlp_ratios": [
80
- 4,
81
- 4,
82
- 4,
83
- 4
84
- ],
85
- "model_type": "segformer",
86
- "num_attention_heads": [
87
  1,
88
  2,
89
- 5,
90
- 8
91
- ],
92
- "num_channels": 3,
93
- "num_encoder_blocks": 4,
94
- "patch_sizes": [
95
- 7,
96
  3,
97
- 3,
98
- 3
99
- ],
100
- "reshape_last_stage": true,
101
- "semantic_loss_ignore_index": 255,
102
- "sr_ratios": [
103
- 8,
104
- 4,
105
- 2,
106
- 1
107
- ],
108
- "strides": [
109
- 4,
110
- 2,
111
- 2,
112
- 2
113
  ],
114
  "torch_dtype": "float32",
115
- "transformers_version": "4.47.0.dev0"
 
 
 
116
  }
 
1
  {
2
+ "_name_or_path": "openmmlab/upernet-convnext-tiny",
3
  "architectures": [
4
+ "UperNetForSemanticSegmentation"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
  ],
6
+ "auxiliary_channels": 256,
7
+ "auxiliary_concat_input": false,
8
+ "auxiliary_in_channels": 384,
9
+ "auxiliary_loss_weight": 0.4,
10
+ "auxiliary_num_convs": 1,
11
+ "backbone": null,
12
+ "backbone_config": {
13
+ "depths": [
14
+ 3,
15
+ 3,
16
+ 9,
17
+ 3
18
+ ],
19
+ "drop_path_rate": 0.0,
20
+ "hidden_act": "gelu",
21
+ "hidden_sizes": [
22
+ 96,
23
+ 192,
24
+ 384,
25
+ 768
26
+ ],
27
+ "image_size": 224,
28
+ "initializer_range": 0.02,
29
+ "layer_norm_eps": 1e-12,
30
+ "layer_scale_init_value": 1e-06,
31
+ "model_type": "convnext",
32
+ "num_stages": 4,
33
+ "out_features": [
34
+ "stage1",
35
+ "stage2",
36
+ "stage3",
37
+ "stage4"
38
+ ],
39
+ "out_indices": [
40
+ 1,
41
+ 2,
42
+ 3,
43
+ 4
44
+ ],
45
+ "patch_size": 4
46
+ },
47
+ "backbone_kwargs": null,
48
+ "hidden_size": 512,
49
  "id2label": {
50
+ "0": "unlabeled",
51
+ "1": "flat-road",
52
+ "2": "flat-sidewalk",
53
+ "3": "flat-crosswalk",
54
+ "4": "flat-cyclinglane",
55
+ "5": "flat-parkingdriveway",
56
+ "6": "flat-railtrack",
57
+ "7": "flat-curb",
58
+ "8": "human-person",
59
+ "9": "human-rider",
60
+ "10": "vehicle-car",
61
+ "11": "vehicle-truck",
62
+ "12": "vehicle-bus",
63
+ "13": "vehicle-tramtrain",
64
+ "14": "vehicle-motorcycle",
65
+ "15": "vehicle-bicycle",
66
+ "16": "vehicle-caravan",
67
+ "17": "vehicle-cartrailer",
68
+ "18": "construction-building",
69
+ "19": "construction-door",
70
+ "20": "construction-wall",
71
+ "21": "construction-fenceguardrail",
72
+ "22": "construction-bridge",
73
+ "23": "construction-tunnel",
74
+ "24": "construction-stairs",
75
+ "25": "object-pole",
76
+ "26": "object-trafficsign",
77
+ "27": "object-trafficlight",
78
+ "28": "nature-vegetation",
79
+ "29": "nature-terrain",
80
+ "30": "sky",
81
+ "31": "void-ground",
82
+ "32": "void-dynamic",
83
+ "33": "void-static",
84
+ "34": "void-unclear"
85
  },
 
86
  "initializer_range": 0.02,
87
  "label2id": {
88
+ "construction-bridge": "22",
89
+ "construction-building": "18",
90
+ "construction-door": "19",
91
+ "construction-fenceguardrail": "21",
92
+ "construction-stairs": "24",
93
+ "construction-tunnel": "23",
94
+ "construction-wall": "20",
95
+ "flat-crosswalk": "3",
96
+ "flat-curb": "7",
97
+ "flat-cyclinglane": "4",
98
+ "flat-parkingdriveway": "5",
99
+ "flat-railtrack": "6",
100
+ "flat-road": "1",
101
+ "flat-sidewalk": "2",
102
+ "human-person": "8",
103
+ "human-rider": "9",
104
+ "nature-terrain": "29",
105
+ "nature-vegetation": "28",
106
+ "object-pole": "25",
107
+ "object-trafficlight": "27",
108
+ "object-trafficsign": "26",
109
+ "sky": "30",
110
+ "unlabeled": "0",
111
+ "vehicle-bicycle": "15",
112
+ "vehicle-bus": "12",
113
+ "vehicle-car": "10",
114
+ "vehicle-caravan": "16",
115
+ "vehicle-cartrailer": "17",
116
+ "vehicle-motorcycle": "14",
117
+ "vehicle-tramtrain": "13",
118
+ "vehicle-truck": "11",
119
+ "void-dynamic": "32",
120
+ "void-ground": "31",
121
+ "void-static": "33",
122
+ "void-unclear": "34"
123
  },
124
+ "loss_ignore_index": 255,
125
+ "model_type": "upernet",
126
+ "pool_scales": [
 
 
 
 
 
 
127
  1,
128
  2,
 
 
 
 
 
 
 
129
  3,
130
+ 6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
131
  ],
132
  "torch_dtype": "float32",
133
+ "transformers_version": "4.47.0.dev0",
134
+ "use_auxiliary_head": true,
135
+ "use_pretrained_backbone": false,
136
+ "use_timm_backbone": false
137
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c7cfb4d04b84584fed35d05a0b55c94ec48bc743b72417fd62c5ae0b616c9310
3
- size 14904316
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a037bef325aba47da3322d11f276dd2d26691df72b2bc7fa2187a0bdd5d940d2
3
+ size 240698832
runs/Nov28_08-36-44_d2a3b59c97a0/events.out.tfevents.1732783017.d2a3b59c97a0.1155.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4661256e4a3f0f3561347f646296708dffbdfb2e7ac0625da4bf1e7988fae36
3
+ size 7532
runs/Nov28_08-41-09_d2a3b59c97a0/events.out.tfevents.1732783276.d2a3b59c97a0.2326.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6f3b4c4ef7f287d66f9f770cacb4a9a1df969f2d4baba4deb9a77b206785262
3
+ size 7738
runs/Nov28_08-42-41_d2a3b59c97a0/events.out.tfevents.1732783365.d2a3b59c97a0.2788.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7aeeaf6d8ecab1b54d67687190ce6bd8ab245029ba79c9584198258d05c86c8
3
+ size 7735
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0a8bbedadf57d22182fdd99f54b94448e64667e43f665d34ef0bf4dd57edcba0
3
  size 5432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4ba1d4ed3e5bbfa398fb108e84abd1d52e15bcd4e6937d763d5cfa60a77f9171
3
  size 5432