Upload folder using huggingface_hub
Browse files- config.json +351 -0
- onnx/model.onnx +3 -0
- onnx/model_bnb4.onnx +3 -0
- onnx/model_fp16.onnx +3 -0
- onnx/model_int8.onnx +3 -0
- onnx/model_q4.onnx +3 -0
- onnx/model_q4f16.onnx +3 -0
- onnx/model_quantized.onnx +3 -0
- onnx/model_uint8.onnx +3 -0
- preprocessor_config.json +26 -0
config.json
ADDED
@@ -0,0 +1,351 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_attn_implementation_autoset": true,
|
3 |
+
"activation_dropout": 0.0,
|
4 |
+
"activation_function": "silu",
|
5 |
+
"anchor_image_size": null,
|
6 |
+
"architectures": [
|
7 |
+
"DFineForObjectDetection"
|
8 |
+
],
|
9 |
+
"attention_dropout": 0.0,
|
10 |
+
"auxiliary_loss": true,
|
11 |
+
"backbone": null,
|
12 |
+
"backbone_config": {
|
13 |
+
"depths": [
|
14 |
+
3,
|
15 |
+
4,
|
16 |
+
6,
|
17 |
+
3
|
18 |
+
],
|
19 |
+
"downsample_in_bottleneck": false,
|
20 |
+
"downsample_in_first_stage": false,
|
21 |
+
"embedding_size": 32,
|
22 |
+
"hidden_act": "relu",
|
23 |
+
"hidden_sizes": [
|
24 |
+
128,
|
25 |
+
256,
|
26 |
+
512,
|
27 |
+
1024
|
28 |
+
],
|
29 |
+
"initializer_range": 0.02,
|
30 |
+
"layer_type": "basic",
|
31 |
+
"model_type": "hgnet_v2",
|
32 |
+
"num_channels": 3,
|
33 |
+
"out_features": [
|
34 |
+
"stage2",
|
35 |
+
"stage3",
|
36 |
+
"stage4"
|
37 |
+
],
|
38 |
+
"out_indices": [
|
39 |
+
2,
|
40 |
+
3,
|
41 |
+
4
|
42 |
+
],
|
43 |
+
"stage_downsample": [
|
44 |
+
false,
|
45 |
+
true,
|
46 |
+
true,
|
47 |
+
true
|
48 |
+
],
|
49 |
+
"stage_in_channels": [
|
50 |
+
16,
|
51 |
+
64,
|
52 |
+
256,
|
53 |
+
512
|
54 |
+
],
|
55 |
+
"stage_kernel_size": [
|
56 |
+
3,
|
57 |
+
3,
|
58 |
+
5,
|
59 |
+
5
|
60 |
+
],
|
61 |
+
"stage_light_block": [
|
62 |
+
false,
|
63 |
+
false,
|
64 |
+
true,
|
65 |
+
true
|
66 |
+
],
|
67 |
+
"stage_mid_channels": [
|
68 |
+
16,
|
69 |
+
32,
|
70 |
+
64,
|
71 |
+
128
|
72 |
+
],
|
73 |
+
"stage_names": [
|
74 |
+
"stem",
|
75 |
+
"stage1",
|
76 |
+
"stage2",
|
77 |
+
"stage3",
|
78 |
+
"stage4"
|
79 |
+
],
|
80 |
+
"stage_num_blocks": [
|
81 |
+
1,
|
82 |
+
1,
|
83 |
+
2,
|
84 |
+
1
|
85 |
+
],
|
86 |
+
"stage_numb_of_layers": [
|
87 |
+
3,
|
88 |
+
3,
|
89 |
+
3,
|
90 |
+
3
|
91 |
+
],
|
92 |
+
"stage_out_channels": [
|
93 |
+
64,
|
94 |
+
256,
|
95 |
+
512,
|
96 |
+
1024
|
97 |
+
],
|
98 |
+
"stem_channels": [
|
99 |
+
3,
|
100 |
+
16,
|
101 |
+
16
|
102 |
+
],
|
103 |
+
"use_learnable_affine_block": true
|
104 |
+
},
|
105 |
+
"backbone_kwargs": null,
|
106 |
+
"batch_norm_eps": 1e-05,
|
107 |
+
"box_noise_scale": 1.0,
|
108 |
+
"d_model": 256,
|
109 |
+
"decoder_activation_function": "relu",
|
110 |
+
"decoder_attention_heads": 8,
|
111 |
+
"decoder_ffn_dim": 1024,
|
112 |
+
"decoder_in_channels": [
|
113 |
+
256,
|
114 |
+
256,
|
115 |
+
256
|
116 |
+
],
|
117 |
+
"decoder_layers": 3,
|
118 |
+
"decoder_method": "default",
|
119 |
+
"decoder_n_points": [
|
120 |
+
3,
|
121 |
+
6,
|
122 |
+
3
|
123 |
+
],
|
124 |
+
"decoder_offset_scale": 0.5,
|
125 |
+
"depth_mult": 0.34,
|
126 |
+
"dropout": 0.0,
|
127 |
+
"encode_proj_layers": [
|
128 |
+
2
|
129 |
+
],
|
130 |
+
"encoder_activation_function": "gelu",
|
131 |
+
"encoder_attention_heads": 8,
|
132 |
+
"encoder_ffn_dim": 1024,
|
133 |
+
"encoder_hidden_dim": 256,
|
134 |
+
"encoder_in_channels": [
|
135 |
+
256,
|
136 |
+
512,
|
137 |
+
1024
|
138 |
+
],
|
139 |
+
"encoder_layers": 1,
|
140 |
+
"eos_coefficient": 0.0001,
|
141 |
+
"eval_idx": -1,
|
142 |
+
"eval_size": null,
|
143 |
+
"feat_strides": [
|
144 |
+
8,
|
145 |
+
16,
|
146 |
+
32
|
147 |
+
],
|
148 |
+
"focal_loss_alpha": 0.75,
|
149 |
+
"focal_loss_gamma": 2.0,
|
150 |
+
"freeze_backbone_batch_norms": true,
|
151 |
+
"hidden_expansion": 0.5,
|
152 |
+
"id2label": {
|
153 |
+
"0": "person",
|
154 |
+
"1": "bicycle",
|
155 |
+
"2": "car",
|
156 |
+
"3": "motorbike",
|
157 |
+
"4": "aeroplane",
|
158 |
+
"5": "bus",
|
159 |
+
"6": "train",
|
160 |
+
"7": "truck",
|
161 |
+
"8": "boat",
|
162 |
+
"9": "traffic light",
|
163 |
+
"10": "fire hydrant",
|
164 |
+
"11": "stop sign",
|
165 |
+
"12": "parking meter",
|
166 |
+
"13": "bench",
|
167 |
+
"14": "bird",
|
168 |
+
"15": "cat",
|
169 |
+
"16": "dog",
|
170 |
+
"17": "horse",
|
171 |
+
"18": "sheep",
|
172 |
+
"19": "cow",
|
173 |
+
"20": "elephant",
|
174 |
+
"21": "bear",
|
175 |
+
"22": "zebra",
|
176 |
+
"23": "giraffe",
|
177 |
+
"24": "backpack",
|
178 |
+
"25": "umbrella",
|
179 |
+
"26": "handbag",
|
180 |
+
"27": "tie",
|
181 |
+
"28": "suitcase",
|
182 |
+
"29": "frisbee",
|
183 |
+
"30": "skis",
|
184 |
+
"31": "snowboard",
|
185 |
+
"32": "sports ball",
|
186 |
+
"33": "kite",
|
187 |
+
"34": "baseball bat",
|
188 |
+
"35": "baseball glove",
|
189 |
+
"36": "skateboard",
|
190 |
+
"37": "surfboard",
|
191 |
+
"38": "tennis racket",
|
192 |
+
"39": "bottle",
|
193 |
+
"40": "wine glass",
|
194 |
+
"41": "cup",
|
195 |
+
"42": "fork",
|
196 |
+
"43": "knife",
|
197 |
+
"44": "spoon",
|
198 |
+
"45": "bowl",
|
199 |
+
"46": "banana",
|
200 |
+
"47": "apple",
|
201 |
+
"48": "sandwich",
|
202 |
+
"49": "orange",
|
203 |
+
"50": "broccoli",
|
204 |
+
"51": "carrot",
|
205 |
+
"52": "hot dog",
|
206 |
+
"53": "pizza",
|
207 |
+
"54": "donut",
|
208 |
+
"55": "cake",
|
209 |
+
"56": "chair",
|
210 |
+
"57": "sofa",
|
211 |
+
"58": "pottedplant",
|
212 |
+
"59": "bed",
|
213 |
+
"60": "diningtable",
|
214 |
+
"61": "toilet",
|
215 |
+
"62": "tvmonitor",
|
216 |
+
"63": "laptop",
|
217 |
+
"64": "mouse",
|
218 |
+
"65": "remote",
|
219 |
+
"66": "keyboard",
|
220 |
+
"67": "cell phone",
|
221 |
+
"68": "microwave",
|
222 |
+
"69": "oven",
|
223 |
+
"70": "toaster",
|
224 |
+
"71": "sink",
|
225 |
+
"72": "refrigerator",
|
226 |
+
"73": "book",
|
227 |
+
"74": "clock",
|
228 |
+
"75": "vase",
|
229 |
+
"76": "scissors",
|
230 |
+
"77": "teddy bear",
|
231 |
+
"78": "hair drier",
|
232 |
+
"79": "toothbrush"
|
233 |
+
},
|
234 |
+
"initializer_bias_prior_prob": null,
|
235 |
+
"initializer_range": 0.01,
|
236 |
+
"is_encoder_decoder": true,
|
237 |
+
"label2id": {
|
238 |
+
"aeroplane": 4,
|
239 |
+
"apple": 47,
|
240 |
+
"backpack": 24,
|
241 |
+
"banana": 46,
|
242 |
+
"baseball bat": 34,
|
243 |
+
"baseball glove": 35,
|
244 |
+
"bear": 21,
|
245 |
+
"bed": 59,
|
246 |
+
"bench": 13,
|
247 |
+
"bicycle": 1,
|
248 |
+
"bird": 14,
|
249 |
+
"boat": 8,
|
250 |
+
"book": 73,
|
251 |
+
"bottle": 39,
|
252 |
+
"bowl": 45,
|
253 |
+
"broccoli": 50,
|
254 |
+
"bus": 5,
|
255 |
+
"cake": 55,
|
256 |
+
"car": 2,
|
257 |
+
"carrot": 51,
|
258 |
+
"cat": 15,
|
259 |
+
"cell phone": 67,
|
260 |
+
"chair": 56,
|
261 |
+
"clock": 74,
|
262 |
+
"cow": 19,
|
263 |
+
"cup": 41,
|
264 |
+
"diningtable": 60,
|
265 |
+
"dog": 16,
|
266 |
+
"donut": 54,
|
267 |
+
"elephant": 20,
|
268 |
+
"fire hydrant": 10,
|
269 |
+
"fork": 42,
|
270 |
+
"frisbee": 29,
|
271 |
+
"giraffe": 23,
|
272 |
+
"hair drier": 78,
|
273 |
+
"handbag": 26,
|
274 |
+
"horse": 17,
|
275 |
+
"hot dog": 52,
|
276 |
+
"keyboard": 66,
|
277 |
+
"kite": 33,
|
278 |
+
"knife": 43,
|
279 |
+
"laptop": 63,
|
280 |
+
"microwave": 68,
|
281 |
+
"motorbike": 3,
|
282 |
+
"mouse": 64,
|
283 |
+
"orange": 49,
|
284 |
+
"oven": 69,
|
285 |
+
"parking meter": 12,
|
286 |
+
"person": 0,
|
287 |
+
"pizza": 53,
|
288 |
+
"pottedplant": 58,
|
289 |
+
"refrigerator": 72,
|
290 |
+
"remote": 65,
|
291 |
+
"sandwich": 48,
|
292 |
+
"scissors": 76,
|
293 |
+
"sheep": 18,
|
294 |
+
"sink": 71,
|
295 |
+
"skateboard": 36,
|
296 |
+
"skis": 30,
|
297 |
+
"snowboard": 31,
|
298 |
+
"sofa": 57,
|
299 |
+
"spoon": 44,
|
300 |
+
"sports ball": 32,
|
301 |
+
"stop sign": 11,
|
302 |
+
"suitcase": 28,
|
303 |
+
"surfboard": 37,
|
304 |
+
"teddy bear": 77,
|
305 |
+
"tennis racket": 38,
|
306 |
+
"tie": 27,
|
307 |
+
"toaster": 70,
|
308 |
+
"toilet": 61,
|
309 |
+
"toothbrush": 79,
|
310 |
+
"traffic light": 9,
|
311 |
+
"train": 6,
|
312 |
+
"truck": 7,
|
313 |
+
"tvmonitor": 62,
|
314 |
+
"umbrella": 25,
|
315 |
+
"vase": 75,
|
316 |
+
"wine glass": 40,
|
317 |
+
"zebra": 22
|
318 |
+
},
|
319 |
+
"label_noise_ratio": 0.5,
|
320 |
+
"layer_norm_eps": 1e-05,
|
321 |
+
"layer_scale": 1,
|
322 |
+
"learn_initial_query": false,
|
323 |
+
"lqe_hidden_dim": 64,
|
324 |
+
"lqe_layers": 2,
|
325 |
+
"matcher_alpha": 0.25,
|
326 |
+
"matcher_bbox_cost": 5.0,
|
327 |
+
"matcher_class_cost": 2.0,
|
328 |
+
"matcher_gamma": 2.0,
|
329 |
+
"matcher_giou_cost": 2.0,
|
330 |
+
"max_num_bins": 32,
|
331 |
+
"model_type": "d_fine",
|
332 |
+
"normalize_before": false,
|
333 |
+
"num_denoising": 100,
|
334 |
+
"num_feature_levels": 3,
|
335 |
+
"num_queries": 300,
|
336 |
+
"positional_encoding_temperature": 10000,
|
337 |
+
"reg_scale": 4.0,
|
338 |
+
"top_prob_values": 4,
|
339 |
+
"torch_dtype": "float32",
|
340 |
+
"transformers_version": "4.52.0.dev0",
|
341 |
+
"up": 0.5,
|
342 |
+
"use_focal_loss": true,
|
343 |
+
"use_pretrained_backbone": false,
|
344 |
+
"use_timm_backbone": false,
|
345 |
+
"weight_loss_bbox": 5.0,
|
346 |
+
"weight_loss_ddf": 1.5,
|
347 |
+
"weight_loss_fgl": 0.15,
|
348 |
+
"weight_loss_giou": 2.0,
|
349 |
+
"weight_loss_vfl": 1.0,
|
350 |
+
"with_box_refine": true
|
351 |
+
}
|
onnx/model.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b9e2e76610053aeeac3b2f1f685d8f9a1182a93a338f624b6c8cb7fb390cb532
|
3 |
+
size 41535197
|
onnx/model_bnb4.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:28e2da4b91dcd3c3b0cf3633424170ef0bb1a931d000d05577108c03c40771c5
|
3 |
+
size 23830540
|
onnx/model_fp16.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:08d9b1e4287f1068777ace892eef354b231c9f8f3a905d9ecf3ef8443a5aa6c0
|
3 |
+
size 21041872
|
onnx/model_int8.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e8fa9fc8215e42a29a135eabd44a40b8c2a7f2e369425fd4ab3cb0a5436ee8a1
|
3 |
+
size 11184985
|
onnx/model_q4.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cab5cb19bafacb36b853ca8bfe02b1d0825fb170cd2ec52123873a1eaf6094a6
|
3 |
+
size 24162815
|
onnx/model_q4f16.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ba08658fb9ca30b7685cc0a4cde3a6d8406a06a60090b7894eef8dba15712ca5
|
3 |
+
size 13652648
|
onnx/model_quantized.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e59db2a51c50561c2df841fbd8961943bfc8745ed92545fd4b5dc652aedc08dc
|
3 |
+
size 11185088
|
onnx/model_uint8.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e59db2a51c50561c2df841fbd8961943bfc8745ed92545fd4b5dc652aedc08dc
|
3 |
+
size 11185088
|
preprocessor_config.json
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"do_convert_annotations": true,
|
3 |
+
"do_normalize": false,
|
4 |
+
"do_pad": false,
|
5 |
+
"do_rescale": true,
|
6 |
+
"do_resize": true,
|
7 |
+
"format": "coco_detection",
|
8 |
+
"image_mean": [
|
9 |
+
0.485,
|
10 |
+
0.456,
|
11 |
+
0.406
|
12 |
+
],
|
13 |
+
"image_processor_type": "RTDetrImageProcessor",
|
14 |
+
"image_std": [
|
15 |
+
0.229,
|
16 |
+
0.224,
|
17 |
+
0.225
|
18 |
+
],
|
19 |
+
"pad_size": null,
|
20 |
+
"resample": 2,
|
21 |
+
"rescale_factor": 0.00392156862745098,
|
22 |
+
"size": {
|
23 |
+
"height": 640,
|
24 |
+
"width": 640
|
25 |
+
}
|
26 |
+
}
|