Training in progress, step 100
Browse files- config.json +591 -0
- model.safetensors +3 -0
- preprocessor_config.json +25 -0
- runs/Aug26_06-26-03_c238bf28089b/events.out.tfevents.1724653571.c238bf28089b.370.0 +3 -0
- runs/Aug26_06-28-02_c238bf28089b/events.out.tfevents.1724653708.c238bf28089b.370.1 +3 -0
- special_tokens_map.json +37 -0
- tokenizer.json +0 -0
- tokenizer_config.json +62 -0
- training_args.bin +3 -0
- vocab.txt +0 -0
config.json
ADDED
@@ -0,0 +1,591 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "Salesforce/blip-vqa-base",
|
3 |
+
"architectures": [
|
4 |
+
"ViltForQuestionAnswering"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.0,
|
7 |
+
"hidden_act": "gelu",
|
8 |
+
"hidden_dropout_prob": 0.0,
|
9 |
+
"hidden_size": 768,
|
10 |
+
"id2label": {
|
11 |
+
"0": "at table",
|
12 |
+
"1": "skateboard",
|
13 |
+
"2": "lg",
|
14 |
+
"3": "6",
|
15 |
+
"4": "crossing",
|
16 |
+
"5": "don't know",
|
17 |
+
"6": "solid",
|
18 |
+
"7": "picnic table",
|
19 |
+
"8": "full",
|
20 |
+
"9": "plain",
|
21 |
+
"10": "window",
|
22 |
+
"11": "8:35",
|
23 |
+
"12": "red and yellow",
|
24 |
+
"13": "girl",
|
25 |
+
"14": "tabby",
|
26 |
+
"15": "blue",
|
27 |
+
"16": "7:45",
|
28 |
+
"17": "down",
|
29 |
+
"18": "unknown",
|
30 |
+
"19": "hawaii",
|
31 |
+
"20": "woods",
|
32 |
+
"21": "little girl",
|
33 |
+
"22": "roof",
|
34 |
+
"23": "black and white",
|
35 |
+
"24": "in car",
|
36 |
+
"25": "clock tower",
|
37 |
+
"26": "gray",
|
38 |
+
"27": "curtains",
|
39 |
+
"28": "ball",
|
40 |
+
"29": "dog",
|
41 |
+
"30": "woman",
|
42 |
+
"31": "soccer ball",
|
43 |
+
"32": "windows",
|
44 |
+
"33": "donut",
|
45 |
+
"34": "screen",
|
46 |
+
"35": "bus",
|
47 |
+
"36": "neon",
|
48 |
+
"37": "monitor",
|
49 |
+
"38": "jeep",
|
50 |
+
"39": "snowboard",
|
51 |
+
"40": "wine tasting",
|
52 |
+
"41": "french",
|
53 |
+
"42": "wedding",
|
54 |
+
"43": "orange",
|
55 |
+
"44": "king",
|
56 |
+
"45": "tired",
|
57 |
+
"46": "canopy",
|
58 |
+
"47": "low",
|
59 |
+
"48": "bikes",
|
60 |
+
"49": "snowboarding",
|
61 |
+
"50": "2000",
|
62 |
+
"51": "skateboarding",
|
63 |
+
"52": "style",
|
64 |
+
"53": "tent",
|
65 |
+
"54": "necklace",
|
66 |
+
"55": "bike rack",
|
67 |
+
"56": "lying down",
|
68 |
+
"57": "clock",
|
69 |
+
"58": "name tag",
|
70 |
+
"59": "hat",
|
71 |
+
"60": "backpack",
|
72 |
+
"61": "on street",
|
73 |
+
"62": "air",
|
74 |
+
"63": "leather",
|
75 |
+
"64": "2010",
|
76 |
+
"65": "can't tell",
|
77 |
+
"66": "bicycle",
|
78 |
+
"67": "lady",
|
79 |
+
"68": "clear",
|
80 |
+
"69": "tan",
|
81 |
+
"70": "skier",
|
82 |
+
"71": "car",
|
83 |
+
"72": "hair",
|
84 |
+
"73": "curtain",
|
85 |
+
"74": "10",
|
86 |
+
"75": "exit",
|
87 |
+
"76": "natural",
|
88 |
+
"77": "camera",
|
89 |
+
"78": "forest",
|
90 |
+
"79": "station",
|
91 |
+
"80": "skiing",
|
92 |
+
"81": "tv",
|
93 |
+
"82": "fence",
|
94 |
+
"83": "smiling",
|
95 |
+
"84": "platform",
|
96 |
+
"85": "happy",
|
97 |
+
"86": "bedroom",
|
98 |
+
"87": "blonde",
|
99 |
+
"88": "double",
|
100 |
+
"89": "train",
|
101 |
+
"90": "nothing",
|
102 |
+
"91": "street",
|
103 |
+
"92": "soccer",
|
104 |
+
"93": "table",
|
105 |
+
"94": "5",
|
106 |
+
"95": "trees",
|
107 |
+
"96": "women",
|
108 |
+
"97": "giraffes",
|
109 |
+
"98": "right",
|
110 |
+
"99": "7",
|
111 |
+
"100": "shelter",
|
112 |
+
"101": "ground",
|
113 |
+
"102": "plate",
|
114 |
+
"103": "laying down",
|
115 |
+
"104": "chopsticks",
|
116 |
+
"105": "red",
|
117 |
+
"106": "many",
|
118 |
+
"107": "shrimp",
|
119 |
+
"108": "not there",
|
120 |
+
"109": "talking",
|
121 |
+
"110": "cloudy",
|
122 |
+
"111": "green",
|
123 |
+
"112": "bicycles",
|
124 |
+
"113": "bricks",
|
125 |
+
"114": "sun",
|
126 |
+
"115": "2013",
|
127 |
+
"116": "brick",
|
128 |
+
"117": "human",
|
129 |
+
"118": "birthday",
|
130 |
+
"119": "snowboarder",
|
131 |
+
"120": "park",
|
132 |
+
"121": "beagle",
|
133 |
+
"122": "yes",
|
134 |
+
"123": "walking",
|
135 |
+
"124": "rack",
|
136 |
+
"125": "purple",
|
137 |
+
"126": "cat",
|
138 |
+
"127": "giraffe",
|
139 |
+
"128": "8",
|
140 |
+
"129": "pink",
|
141 |
+
"130": "plastic",
|
142 |
+
"131": "red and blue",
|
143 |
+
"132": "stripes",
|
144 |
+
"133": "lanyard",
|
145 |
+
"134": "shade",
|
146 |
+
"135": "dirt",
|
147 |
+
"136": "they aren't",
|
148 |
+
"137": "0",
|
149 |
+
"138": "ice cream",
|
150 |
+
"139": "zoo",
|
151 |
+
"140": "wall",
|
152 |
+
"141": "cup",
|
153 |
+
"142": "queen",
|
154 |
+
"143": "cage",
|
155 |
+
"144": "africa",
|
156 |
+
"145": "beige",
|
157 |
+
"146": "white",
|
158 |
+
"147": "snow",
|
159 |
+
"148": "yellow",
|
160 |
+
"149": "white and blue",
|
161 |
+
"150": "calico",
|
162 |
+
"151": "big ben",
|
163 |
+
"152": "wine",
|
164 |
+
"153": "sky",
|
165 |
+
"154": "security",
|
166 |
+
"155": "2",
|
167 |
+
"156": "sidewalk",
|
168 |
+
"157": "stand",
|
169 |
+
"158": "4",
|
170 |
+
"159": "smile",
|
171 |
+
"160": "gray and black",
|
172 |
+
"161": "protection",
|
173 |
+
"162": "3",
|
174 |
+
"163": "watching",
|
175 |
+
"164": "shadow",
|
176 |
+
"165": "shadows",
|
177 |
+
"166": "fashion",
|
178 |
+
"167": "7:35",
|
179 |
+
"168": "crown",
|
180 |
+
"169": "blue and white",
|
181 |
+
"170": "man",
|
182 |
+
"171": "door",
|
183 |
+
"172": "sleeping",
|
184 |
+
"173": "large",
|
185 |
+
"174": "net",
|
186 |
+
"175": "suv",
|
187 |
+
"176": "brown",
|
188 |
+
"177": "not sure",
|
189 |
+
"178": "arrow",
|
190 |
+
"179": "1",
|
191 |
+
"180": "black",
|
192 |
+
"181": "out",
|
193 |
+
"182": "person",
|
194 |
+
"183": "desert",
|
195 |
+
"184": "boy",
|
196 |
+
"185": "tower",
|
197 |
+
"186": "9:35",
|
198 |
+
"187": "chair",
|
199 |
+
"188": "talking on phone",
|
200 |
+
"189": "small",
|
201 |
+
"190": "resting",
|
202 |
+
"191": "church",
|
203 |
+
"192": "outside",
|
204 |
+
"193": "cross",
|
205 |
+
"194": "white and black",
|
206 |
+
"195": "no",
|
207 |
+
"196": "photographer",
|
208 |
+
"197": "on road",
|
209 |
+
"198": "doughnut"
|
210 |
+
},
|
211 |
+
"image_size": 384,
|
212 |
+
"image_text_hidden_size": 256,
|
213 |
+
"initializer_factor": 1.0,
|
214 |
+
"initializer_range": 0.02,
|
215 |
+
"intermediate_size": 3072,
|
216 |
+
"label2id": {
|
217 |
+
"0": 137,
|
218 |
+
"1": 179,
|
219 |
+
"10": 74,
|
220 |
+
"2": 155,
|
221 |
+
"2000": 50,
|
222 |
+
"2010": 64,
|
223 |
+
"2013": 115,
|
224 |
+
"3": 162,
|
225 |
+
"4": 158,
|
226 |
+
"5": 94,
|
227 |
+
"6": 3,
|
228 |
+
"7": 99,
|
229 |
+
"7:35": 167,
|
230 |
+
"7:45": 16,
|
231 |
+
"8": 128,
|
232 |
+
"8:35": 11,
|
233 |
+
"9:35": 186,
|
234 |
+
"africa": 144,
|
235 |
+
"air": 62,
|
236 |
+
"arrow": 178,
|
237 |
+
"at table": 0,
|
238 |
+
"backpack": 60,
|
239 |
+
"ball": 28,
|
240 |
+
"beagle": 121,
|
241 |
+
"bedroom": 86,
|
242 |
+
"beige": 145,
|
243 |
+
"bicycle": 66,
|
244 |
+
"bicycles": 112,
|
245 |
+
"big ben": 151,
|
246 |
+
"bike rack": 55,
|
247 |
+
"bikes": 48,
|
248 |
+
"birthday": 118,
|
249 |
+
"black": 180,
|
250 |
+
"black and white": 23,
|
251 |
+
"blonde": 87,
|
252 |
+
"blue": 15,
|
253 |
+
"blue and white": 169,
|
254 |
+
"boy": 184,
|
255 |
+
"brick": 116,
|
256 |
+
"bricks": 113,
|
257 |
+
"brown": 176,
|
258 |
+
"bus": 35,
|
259 |
+
"cage": 143,
|
260 |
+
"calico": 150,
|
261 |
+
"camera": 77,
|
262 |
+
"can't tell": 65,
|
263 |
+
"canopy": 46,
|
264 |
+
"car": 71,
|
265 |
+
"cat": 126,
|
266 |
+
"chair": 187,
|
267 |
+
"chopsticks": 104,
|
268 |
+
"church": 191,
|
269 |
+
"clear": 68,
|
270 |
+
"clock": 57,
|
271 |
+
"clock tower": 25,
|
272 |
+
"cloudy": 110,
|
273 |
+
"cross": 193,
|
274 |
+
"crossing": 4,
|
275 |
+
"crown": 168,
|
276 |
+
"cup": 141,
|
277 |
+
"curtain": 73,
|
278 |
+
"curtains": 27,
|
279 |
+
"desert": 183,
|
280 |
+
"dirt": 135,
|
281 |
+
"dog": 29,
|
282 |
+
"don't know": 5,
|
283 |
+
"donut": 33,
|
284 |
+
"door": 171,
|
285 |
+
"double": 88,
|
286 |
+
"doughnut": 198,
|
287 |
+
"down": 17,
|
288 |
+
"exit": 75,
|
289 |
+
"fashion": 166,
|
290 |
+
"fence": 82,
|
291 |
+
"forest": 78,
|
292 |
+
"french": 41,
|
293 |
+
"full": 8,
|
294 |
+
"giraffe": 127,
|
295 |
+
"giraffes": 97,
|
296 |
+
"girl": 13,
|
297 |
+
"gray": 26,
|
298 |
+
"gray and black": 160,
|
299 |
+
"green": 111,
|
300 |
+
"ground": 101,
|
301 |
+
"hair": 72,
|
302 |
+
"happy": 85,
|
303 |
+
"hat": 59,
|
304 |
+
"hawaii": 19,
|
305 |
+
"human": 117,
|
306 |
+
"ice cream": 138,
|
307 |
+
"in car": 24,
|
308 |
+
"jeep": 38,
|
309 |
+
"king": 44,
|
310 |
+
"lady": 67,
|
311 |
+
"lanyard": 133,
|
312 |
+
"large": 173,
|
313 |
+
"laying down": 103,
|
314 |
+
"leather": 63,
|
315 |
+
"lg": 2,
|
316 |
+
"little girl": 21,
|
317 |
+
"low": 47,
|
318 |
+
"lying down": 56,
|
319 |
+
"man": 170,
|
320 |
+
"many": 106,
|
321 |
+
"monitor": 37,
|
322 |
+
"name tag": 58,
|
323 |
+
"natural": 76,
|
324 |
+
"necklace": 54,
|
325 |
+
"neon": 36,
|
326 |
+
"net": 174,
|
327 |
+
"no": 195,
|
328 |
+
"not sure": 177,
|
329 |
+
"not there": 108,
|
330 |
+
"nothing": 90,
|
331 |
+
"on road": 197,
|
332 |
+
"on street": 61,
|
333 |
+
"orange": 43,
|
334 |
+
"out": 181,
|
335 |
+
"outside": 192,
|
336 |
+
"park": 120,
|
337 |
+
"person": 182,
|
338 |
+
"photographer": 196,
|
339 |
+
"picnic table": 7,
|
340 |
+
"pink": 129,
|
341 |
+
"plain": 9,
|
342 |
+
"plastic": 130,
|
343 |
+
"plate": 102,
|
344 |
+
"platform": 84,
|
345 |
+
"protection": 161,
|
346 |
+
"purple": 125,
|
347 |
+
"queen": 142,
|
348 |
+
"rack": 124,
|
349 |
+
"red": 105,
|
350 |
+
"red and blue": 131,
|
351 |
+
"red and yellow": 12,
|
352 |
+
"resting": 190,
|
353 |
+
"right": 98,
|
354 |
+
"roof": 22,
|
355 |
+
"screen": 34,
|
356 |
+
"security": 154,
|
357 |
+
"shade": 134,
|
358 |
+
"shadow": 164,
|
359 |
+
"shadows": 165,
|
360 |
+
"shelter": 100,
|
361 |
+
"shrimp": 107,
|
362 |
+
"sidewalk": 156,
|
363 |
+
"skateboard": 1,
|
364 |
+
"skateboarding": 51,
|
365 |
+
"skier": 70,
|
366 |
+
"skiing": 80,
|
367 |
+
"sky": 153,
|
368 |
+
"sleeping": 172,
|
369 |
+
"small": 189,
|
370 |
+
"smile": 159,
|
371 |
+
"smiling": 83,
|
372 |
+
"snow": 147,
|
373 |
+
"snowboard": 39,
|
374 |
+
"snowboarder": 119,
|
375 |
+
"snowboarding": 49,
|
376 |
+
"soccer": 92,
|
377 |
+
"soccer ball": 31,
|
378 |
+
"solid": 6,
|
379 |
+
"stand": 157,
|
380 |
+
"station": 79,
|
381 |
+
"street": 91,
|
382 |
+
"stripes": 132,
|
383 |
+
"style": 52,
|
384 |
+
"sun": 114,
|
385 |
+
"suv": 175,
|
386 |
+
"tabby": 14,
|
387 |
+
"table": 93,
|
388 |
+
"talking": 109,
|
389 |
+
"talking on phone": 188,
|
390 |
+
"tan": 69,
|
391 |
+
"tent": 53,
|
392 |
+
"they aren't": 136,
|
393 |
+
"tired": 45,
|
394 |
+
"tower": 185,
|
395 |
+
"train": 89,
|
396 |
+
"trees": 95,
|
397 |
+
"tv": 81,
|
398 |
+
"unknown": 18,
|
399 |
+
"walking": 123,
|
400 |
+
"wall": 140,
|
401 |
+
"watching": 163,
|
402 |
+
"wedding": 42,
|
403 |
+
"white": 146,
|
404 |
+
"white and black": 194,
|
405 |
+
"white and blue": 149,
|
406 |
+
"window": 10,
|
407 |
+
"windows": 32,
|
408 |
+
"wine": 152,
|
409 |
+
"wine tasting": 40,
|
410 |
+
"woman": 30,
|
411 |
+
"women": 96,
|
412 |
+
"woods": 20,
|
413 |
+
"yellow": 148,
|
414 |
+
"yes": 122,
|
415 |
+
"zoo": 139
|
416 |
+
},
|
417 |
+
"layer_norm_eps": 1e-12,
|
418 |
+
"logit_scale_init_value": 2.6592,
|
419 |
+
"max_image_length": -1,
|
420 |
+
"max_position_embeddings": 40,
|
421 |
+
"modality_type_vocab_size": 2,
|
422 |
+
"model_type": "vilt",
|
423 |
+
"num_attention_heads": 12,
|
424 |
+
"num_channels": 3,
|
425 |
+
"num_hidden_layers": 12,
|
426 |
+
"num_images": -1,
|
427 |
+
"patch_size": 32,
|
428 |
+
"projection_dim": 512,
|
429 |
+
"qkv_bias": true,
|
430 |
+
"text_config": {
|
431 |
+
"_name_or_path": "",
|
432 |
+
"add_cross_attention": false,
|
433 |
+
"architectures": null,
|
434 |
+
"attention_probs_dropout_prob": 0.0,
|
435 |
+
"bad_words_ids": null,
|
436 |
+
"begin_suppress_tokens": null,
|
437 |
+
"bos_token_id": 30522,
|
438 |
+
"chunk_size_feed_forward": 0,
|
439 |
+
"cross_attention_hidden_size": null,
|
440 |
+
"decoder_start_token_id": null,
|
441 |
+
"diversity_penalty": 0.0,
|
442 |
+
"do_sample": false,
|
443 |
+
"early_stopping": false,
|
444 |
+
"encoder_no_repeat_ngram_size": 0,
|
445 |
+
"eos_token_id": 2,
|
446 |
+
"exponential_decay_length_penalty": null,
|
447 |
+
"finetuning_task": null,
|
448 |
+
"forced_bos_token_id": null,
|
449 |
+
"forced_eos_token_id": null,
|
450 |
+
"hidden_act": "gelu",
|
451 |
+
"hidden_dropout_prob": 0.0,
|
452 |
+
"hidden_size": 768,
|
453 |
+
"id2label": {
|
454 |
+
"0": "LABEL_0",
|
455 |
+
"1": "LABEL_1"
|
456 |
+
},
|
457 |
+
"initializer_factor": 1.0,
|
458 |
+
"initializer_range": 0.02,
|
459 |
+
"intermediate_size": 3072,
|
460 |
+
"is_decoder": true,
|
461 |
+
"is_encoder_decoder": false,
|
462 |
+
"label2id": {
|
463 |
+
"LABEL_0": 0,
|
464 |
+
"LABEL_1": 1
|
465 |
+
},
|
466 |
+
"layer_norm_eps": 1e-12,
|
467 |
+
"length_penalty": 1.0,
|
468 |
+
"max_length": 20,
|
469 |
+
"max_position_embeddings": 512,
|
470 |
+
"min_length": 0,
|
471 |
+
"model_type": "blip_text_model",
|
472 |
+
"no_repeat_ngram_size": 0,
|
473 |
+
"num_attention_heads": 12,
|
474 |
+
"num_beam_groups": 1,
|
475 |
+
"num_beams": 1,
|
476 |
+
"num_hidden_layers": 12,
|
477 |
+
"num_return_sequences": 1,
|
478 |
+
"output_attentions": false,
|
479 |
+
"output_hidden_states": false,
|
480 |
+
"output_scores": false,
|
481 |
+
"pad_token_id": 0,
|
482 |
+
"prefix": null,
|
483 |
+
"problem_type": null,
|
484 |
+
"projection_dim": 768,
|
485 |
+
"pruned_heads": {},
|
486 |
+
"remove_invalid_values": false,
|
487 |
+
"repetition_penalty": 1.0,
|
488 |
+
"return_dict": true,
|
489 |
+
"return_dict_in_generate": false,
|
490 |
+
"sep_token_id": 102,
|
491 |
+
"suppress_tokens": null,
|
492 |
+
"task_specific_params": null,
|
493 |
+
"temperature": 1.0,
|
494 |
+
"tf_legacy_loss": false,
|
495 |
+
"tie_encoder_decoder": false,
|
496 |
+
"tie_word_embeddings": true,
|
497 |
+
"tokenizer_class": null,
|
498 |
+
"top_k": 50,
|
499 |
+
"top_p": 1.0,
|
500 |
+
"torch_dtype": null,
|
501 |
+
"torchscript": false,
|
502 |
+
"transformers_version": "4.26.0.dev0",
|
503 |
+
"typical_p": 1.0,
|
504 |
+
"use_bfloat16": false,
|
505 |
+
"use_cache": true,
|
506 |
+
"vocab_size": 30524
|
507 |
+
},
|
508 |
+
"tie_word_embeddings": false,
|
509 |
+
"torch_dtype": "float32",
|
510 |
+
"transformers_version": "4.42.4",
|
511 |
+
"type_vocab_size": 2,
|
512 |
+
"vision_config": {
|
513 |
+
"_name_or_path": "",
|
514 |
+
"add_cross_attention": false,
|
515 |
+
"architectures": null,
|
516 |
+
"attention_dropout": 0.0,
|
517 |
+
"bad_words_ids": null,
|
518 |
+
"begin_suppress_tokens": null,
|
519 |
+
"bos_token_id": null,
|
520 |
+
"chunk_size_feed_forward": 0,
|
521 |
+
"cross_attention_hidden_size": null,
|
522 |
+
"decoder_start_token_id": null,
|
523 |
+
"diversity_penalty": 0.0,
|
524 |
+
"do_sample": false,
|
525 |
+
"dropout": 0.0,
|
526 |
+
"early_stopping": false,
|
527 |
+
"encoder_no_repeat_ngram_size": 0,
|
528 |
+
"eos_token_id": null,
|
529 |
+
"exponential_decay_length_penalty": null,
|
530 |
+
"finetuning_task": null,
|
531 |
+
"forced_bos_token_id": null,
|
532 |
+
"forced_eos_token_id": null,
|
533 |
+
"hidden_act": "gelu",
|
534 |
+
"hidden_size": 768,
|
535 |
+
"id2label": {
|
536 |
+
"0": "LABEL_0",
|
537 |
+
"1": "LABEL_1"
|
538 |
+
},
|
539 |
+
"image_size": 384,
|
540 |
+
"initializer_factor": 1.0,
|
541 |
+
"initializer_range": 0.02,
|
542 |
+
"intermediate_size": 3072,
|
543 |
+
"is_decoder": false,
|
544 |
+
"is_encoder_decoder": false,
|
545 |
+
"label2id": {
|
546 |
+
"LABEL_0": 0,
|
547 |
+
"LABEL_1": 1
|
548 |
+
},
|
549 |
+
"layer_norm_eps": 1e-05,
|
550 |
+
"length_penalty": 1.0,
|
551 |
+
"max_length": 20,
|
552 |
+
"min_length": 0,
|
553 |
+
"model_type": "blip_vision_model",
|
554 |
+
"no_repeat_ngram_size": 0,
|
555 |
+
"num_attention_heads": 12,
|
556 |
+
"num_beam_groups": 1,
|
557 |
+
"num_beams": 1,
|
558 |
+
"num_channels": 3,
|
559 |
+
"num_hidden_layers": 12,
|
560 |
+
"num_return_sequences": 1,
|
561 |
+
"output_attentions": false,
|
562 |
+
"output_hidden_states": false,
|
563 |
+
"output_scores": false,
|
564 |
+
"pad_token_id": null,
|
565 |
+
"patch_size": 16,
|
566 |
+
"prefix": null,
|
567 |
+
"problem_type": null,
|
568 |
+
"projection_dim": 512,
|
569 |
+
"pruned_heads": {},
|
570 |
+
"remove_invalid_values": false,
|
571 |
+
"repetition_penalty": 1.0,
|
572 |
+
"return_dict": true,
|
573 |
+
"return_dict_in_generate": false,
|
574 |
+
"sep_token_id": null,
|
575 |
+
"suppress_tokens": null,
|
576 |
+
"task_specific_params": null,
|
577 |
+
"temperature": 1.0,
|
578 |
+
"tf_legacy_loss": false,
|
579 |
+
"tie_encoder_decoder": false,
|
580 |
+
"tie_word_embeddings": true,
|
581 |
+
"tokenizer_class": null,
|
582 |
+
"top_k": 50,
|
583 |
+
"top_p": 1.0,
|
584 |
+
"torch_dtype": null,
|
585 |
+
"torchscript": false,
|
586 |
+
"transformers_version": "4.26.0.dev0",
|
587 |
+
"typical_p": 1.0,
|
588 |
+
"use_bfloat16": false
|
589 |
+
},
|
590 |
+
"vocab_size": 30522
|
591 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:32caca021a5d9250379fc9ecb7a8897be392124ac079e285d0a98832e6823d4a
|
3 |
+
size 452365324
|
preprocessor_config.json
ADDED
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"do_normalize": true,
|
3 |
+
"do_pad": true,
|
4 |
+
"do_rescale": true,
|
5 |
+
"do_resize": true,
|
6 |
+
"image_mean": [
|
7 |
+
0.48145466,
|
8 |
+
0.4578275,
|
9 |
+
0.40821073
|
10 |
+
],
|
11 |
+
"image_processor_type": "ViltImageProcessor",
|
12 |
+
"image_std": [
|
13 |
+
0.26862954,
|
14 |
+
0.26130258,
|
15 |
+
0.27577711
|
16 |
+
],
|
17 |
+
"processor_class": "ViltProcessor",
|
18 |
+
"resample": 3,
|
19 |
+
"rescale_factor": 0.00392156862745098,
|
20 |
+
"size": {
|
21 |
+
"height": 384,
|
22 |
+
"width": 384
|
23 |
+
},
|
24 |
+
"size_divisor": 32
|
25 |
+
}
|
runs/Aug26_06-26-03_c238bf28089b/events.out.tfevents.1724653571.c238bf28089b.370.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:32358e10538065a4d2e0b400ff2fb054fcdf1ec97ffa860d2f298402e5618e63
|
3 |
+
size 17133
|
runs/Aug26_06-28-02_c238bf28089b/events.out.tfevents.1724653708.c238bf28089b.370.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8892c87e0ea71555e9965ca0cf94f9b6e74ba70e5109c92a99190b645f654afc
|
3 |
+
size 17895
|
special_tokens_map.json
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cls_token": {
|
3 |
+
"content": "[CLS]",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": false,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"mask_token": {
|
10 |
+
"content": "[MASK]",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": false,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"pad_token": {
|
17 |
+
"content": "[PAD]",
|
18 |
+
"lstrip": false,
|
19 |
+
"normalized": false,
|
20 |
+
"rstrip": false,
|
21 |
+
"single_word": false
|
22 |
+
},
|
23 |
+
"sep_token": {
|
24 |
+
"content": "[SEP]",
|
25 |
+
"lstrip": false,
|
26 |
+
"normalized": false,
|
27 |
+
"rstrip": false,
|
28 |
+
"single_word": false
|
29 |
+
},
|
30 |
+
"unk_token": {
|
31 |
+
"content": "[UNK]",
|
32 |
+
"lstrip": false,
|
33 |
+
"normalized": false,
|
34 |
+
"rstrip": false,
|
35 |
+
"single_word": false
|
36 |
+
}
|
37 |
+
}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
ADDED
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"added_tokens_decoder": {
|
3 |
+
"0": {
|
4 |
+
"content": "[PAD]",
|
5 |
+
"lstrip": false,
|
6 |
+
"normalized": false,
|
7 |
+
"rstrip": false,
|
8 |
+
"single_word": false,
|
9 |
+
"special": true
|
10 |
+
},
|
11 |
+
"100": {
|
12 |
+
"content": "[UNK]",
|
13 |
+
"lstrip": false,
|
14 |
+
"normalized": false,
|
15 |
+
"rstrip": false,
|
16 |
+
"single_word": false,
|
17 |
+
"special": true
|
18 |
+
},
|
19 |
+
"101": {
|
20 |
+
"content": "[CLS]",
|
21 |
+
"lstrip": false,
|
22 |
+
"normalized": false,
|
23 |
+
"rstrip": false,
|
24 |
+
"single_word": false,
|
25 |
+
"special": true
|
26 |
+
},
|
27 |
+
"102": {
|
28 |
+
"content": "[SEP]",
|
29 |
+
"lstrip": false,
|
30 |
+
"normalized": false,
|
31 |
+
"rstrip": false,
|
32 |
+
"single_word": false,
|
33 |
+
"special": true
|
34 |
+
},
|
35 |
+
"103": {
|
36 |
+
"content": "[MASK]",
|
37 |
+
"lstrip": false,
|
38 |
+
"normalized": false,
|
39 |
+
"rstrip": false,
|
40 |
+
"single_word": false,
|
41 |
+
"special": true
|
42 |
+
}
|
43 |
+
},
|
44 |
+
"clean_up_tokenization_spaces": true,
|
45 |
+
"cls_token": "[CLS]",
|
46 |
+
"do_basic_tokenize": true,
|
47 |
+
"do_lower_case": true,
|
48 |
+
"mask_token": "[MASK]",
|
49 |
+
"model_input_names": [
|
50 |
+
"input_ids",
|
51 |
+
"attention_mask"
|
52 |
+
],
|
53 |
+
"model_max_length": 512,
|
54 |
+
"never_split": null,
|
55 |
+
"pad_token": "[PAD]",
|
56 |
+
"processor_class": "ViltProcessor",
|
57 |
+
"sep_token": "[SEP]",
|
58 |
+
"strip_accents": null,
|
59 |
+
"tokenize_chinese_chars": true,
|
60 |
+
"tokenizer_class": "BertTokenizer",
|
61 |
+
"unk_token": "[UNK]"
|
62 |
+
}
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0ce5c0a75edabe7d5fab6b5b751796594f3d798fc0564dae7a0a543eae4f5624
|
3 |
+
size 5112
|
vocab.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|