hf-transformers-bot commited on
Commit
5bd2715
·
verified ·
1 Parent(s): 649b31c

Add tiny processor fixture for testing

Browse files
chat_template.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "chat_template": "{% for message in messages %}{% if message['role'] != 'system' %}{{ message['role'].upper() + ': '}}{% endif %}{# Render all images first #}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>\n' }}{% endfor %}{# Render all images first #}{% for content in message['content'] | selectattr('type', 'equalto', 'video') %}{{ '<video>\n' }}{% endfor %}{# Render all text next #}{% if message['role'] != 'assistant' %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] + ' '}}{% endfor %}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{% generation %}{{ content['text'] + ' '}}{% endgeneration %}{% endfor %}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ 'ASSISTANT:' }}{% endif %}"
3
+ }
config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlavaNextVideoForConditionalGeneration"
4
+ ],
5
+ "ignore_index": -100,
6
+ "image_grid_pinpoints": [
7
+ [
8
+ 336,
9
+ 672
10
+ ],
11
+ [
12
+ 672,
13
+ 336
14
+ ],
15
+ [
16
+ 672,
17
+ 672
18
+ ],
19
+ [
20
+ 1008,
21
+ 336
22
+ ],
23
+ [
24
+ 336,
25
+ 1008
26
+ ]
27
+ ],
28
+ "image_token_index": 32001,
29
+ "model_type": "llava_next_video",
30
+ "projector_hidden_act": "gelu",
31
+ "spatial_pool_mode": "average",
32
+ "spatial_pool_out_channels": 1024,
33
+ "spatial_pool_stride": 2,
34
+ "text_config": {
35
+ "_name_or_path": "lmsys/vicuna-7b-v1.5",
36
+ "architectures": [
37
+ "LlamaForCausalLM"
38
+ ],
39
+ "rope_scaling": {
40
+ "factor": 2.5,
41
+ "type": "linear"
42
+ },
43
+ "max_position_embeddings": 4096,
44
+ "model_type": "llama",
45
+ "pad_token_id": 0,
46
+ "rms_norm_eps": 1e-05,
47
+ "torch_dtype": "float16",
48
+ "type": "linear",
49
+ "vocab_size": 32064
50
+ },
51
+ "tie_word_embeddings": false,
52
+ "torch_dtype": "bfloat16",
53
+ "transformers_version": "4.42.0.dev0",
54
+ "use_image_newline_parameter": true,
55
+ "video_token_index": 32000,
56
+ "vision_config": {
57
+ "hidden_size": 1024,
58
+ "image_size": 336,
59
+ "intermediate_size": 4096,
60
+ "model_type": "clip_vision_model",
61
+ "num_attention_heads": 16,
62
+ "num_hidden_layers": 24,
63
+ "patch_size": 14,
64
+ "projection_dim": 768,
65
+ "vocab_size": 32000
66
+ },
67
+ "vision_feature_layer": -2,
68
+ "vision_feature_select_strategy": "default",
69
+ "vocab_size": 500
70
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.42.0.dev0"
7
+ }
preprocessor_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": {
3
+ "height": 336,
4
+ "width": 336
5
+ },
6
+ "do_center_crop": true,
7
+ "do_convert_rgb": true,
8
+ "do_normalize": true,
9
+ "do_pad": true,
10
+ "do_rescale": true,
11
+ "do_resize": true,
12
+ "image_grid_pinpoints": [
13
+ [
14
+ 336,
15
+ 672
16
+ ],
17
+ [
18
+ 672,
19
+ 336
20
+ ],
21
+ [
22
+ 672,
23
+ 672
24
+ ],
25
+ [
26
+ 1008,
27
+ 336
28
+ ],
29
+ [
30
+ 336,
31
+ 1008
32
+ ]
33
+ ],
34
+ "image_mean": [
35
+ 0.48145466,
36
+ 0.4578275,
37
+ 0.40821073
38
+ ],
39
+ "image_processor_type": "LlavaNextImageProcessor",
40
+ "image_std": [
41
+ 0.26862954,
42
+ 0.26130258,
43
+ 0.27577711
44
+ ],
45
+ "processor_class": "LlavaNextVideoProcessor",
46
+ "resample": 3,
47
+ "rescale_factor": 0.00392156862745098,
48
+ "size": {
49
+ "shortest_edge": 336
50
+ }
51
+ }
processor_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_token": "<image>",
3
+ "video_token": "<video>",
4
+ "num_additional_image_tokens": 1,
5
+ "patch_size": 14,
6
+ "processor_class": "LlavaNextVideoProcessor",
7
+ "vision_feature_select_strategy": "default"
8
+ }
tokenizer.json ADDED
@@ -0,0 +1,680 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": {
5
+ "strategy": "BatchLongest",
6
+ "direction": "Left",
7
+ "pad_to_multiple_of": null,
8
+ "pad_id": 0,
9
+ "pad_type_id": 0,
10
+ "pad_token": "<unk>"
11
+ },
12
+ "added_tokens": [
13
+ {
14
+ "id": 0,
15
+ "content": "<unk>",
16
+ "single_word": false,
17
+ "lstrip": false,
18
+ "rstrip": false,
19
+ "normalized": false,
20
+ "special": true
21
+ },
22
+ {
23
+ "id": 1,
24
+ "content": "<s>",
25
+ "single_word": false,
26
+ "lstrip": false,
27
+ "rstrip": false,
28
+ "normalized": false,
29
+ "special": true
30
+ },
31
+ {
32
+ "id": 2,
33
+ "content": "</s>",
34
+ "single_word": false,
35
+ "lstrip": false,
36
+ "rstrip": false,
37
+ "normalized": false,
38
+ "special": true
39
+ },
40
+ {
41
+ "id": 500,
42
+ "content": "<video>",
43
+ "single_word": false,
44
+ "lstrip": false,
45
+ "rstrip": false,
46
+ "normalized": false,
47
+ "special": true
48
+ },
49
+ {
50
+ "id": 501,
51
+ "content": "<image>",
52
+ "single_word": false,
53
+ "lstrip": false,
54
+ "rstrip": false,
55
+ "normalized": false,
56
+ "special": true
57
+ }
58
+ ],
59
+ "normalizer": null,
60
+ "pre_tokenizer": {
61
+ "type": "Metaspace",
62
+ "replacement": "▁",
63
+ "prepend_scheme": "first",
64
+ "split": false
65
+ },
66
+ "post_processor": {
67
+ "type": "TemplateProcessing",
68
+ "single": [
69
+ {
70
+ "SpecialToken": {
71
+ "id": "<s>",
72
+ "type_id": 0
73
+ }
74
+ },
75
+ {
76
+ "Sequence": {
77
+ "id": "A",
78
+ "type_id": 0
79
+ }
80
+ }
81
+ ],
82
+ "pair": [
83
+ {
84
+ "SpecialToken": {
85
+ "id": "<s>",
86
+ "type_id": 0
87
+ }
88
+ },
89
+ {
90
+ "Sequence": {
91
+ "id": "A",
92
+ "type_id": 0
93
+ }
94
+ },
95
+ {
96
+ "SpecialToken": {
97
+ "id": "<s>",
98
+ "type_id": 1
99
+ }
100
+ },
101
+ {
102
+ "Sequence": {
103
+ "id": "B",
104
+ "type_id": 1
105
+ }
106
+ }
107
+ ],
108
+ "special_tokens": {
109
+ "<s>": {
110
+ "id": "<s>",
111
+ "ids": [
112
+ 1
113
+ ],
114
+ "tokens": [
115
+ "<s>"
116
+ ]
117
+ }
118
+ }
119
+ },
120
+ "decoder": {
121
+ "type": "Sequence",
122
+ "decoders": [
123
+ {
124
+ "type": "Replace",
125
+ "pattern": {
126
+ "String": "▁"
127
+ },
128
+ "content": " "
129
+ },
130
+ {
131
+ "type": "ByteFallback"
132
+ },
133
+ {
134
+ "type": "Fuse"
135
+ },
136
+ {
137
+ "type": "Strip",
138
+ "content": " ",
139
+ "start": 1,
140
+ "stop": 0
141
+ }
142
+ ]
143
+ },
144
+ "model": {
145
+ "type": "BPE",
146
+ "dropout": null,
147
+ "unk_token": "<unk>",
148
+ "continuing_subword_prefix": null,
149
+ "end_of_word_suffix": null,
150
+ "fuse_unk": true,
151
+ "byte_fallback": true,
152
+ "ignore_merges": false,
153
+ "vocab": {
154
+ "<unk>": 0,
155
+ "<s>": 1,
156
+ "</s>": 2,
157
+ "<0x00>": 3,
158
+ "<0x01>": 4,
159
+ "<0x02>": 5,
160
+ "<0x03>": 6,
161
+ "<0x04>": 7,
162
+ "<0x05>": 8,
163
+ "<0x06>": 9,
164
+ "<0x07>": 10,
165
+ "<0x08>": 11,
166
+ "<0x09>": 12,
167
+ "<0x0A>": 13,
168
+ "<0x0B>": 14,
169
+ "<0x0C>": 15,
170
+ "<0x0D>": 16,
171
+ "<0x0E>": 17,
172
+ "<0x0F>": 18,
173
+ "<0x10>": 19,
174
+ "<0x11>": 20,
175
+ "<0x12>": 21,
176
+ "<0x13>": 22,
177
+ "<0x14>": 23,
178
+ "<0x15>": 24,
179
+ "<0x16>": 25,
180
+ "<0x17>": 26,
181
+ "<0x18>": 27,
182
+ "<0x19>": 28,
183
+ "<0x1A>": 29,
184
+ "<0x1B>": 30,
185
+ "<0x1C>": 31,
186
+ "<0x1D>": 32,
187
+ "<0x1E>": 33,
188
+ "<0x1F>": 34,
189
+ "<0x20>": 35,
190
+ "<0x21>": 36,
191
+ "<0x22>": 37,
192
+ "<0x23>": 38,
193
+ "<0x24>": 39,
194
+ "<0x25>": 40,
195
+ "<0x26>": 41,
196
+ "<0x27>": 42,
197
+ "<0x28>": 43,
198
+ "<0x29>": 44,
199
+ "<0x2A>": 45,
200
+ "<0x2B>": 46,
201
+ "<0x2C>": 47,
202
+ "<0x2D>": 48,
203
+ "<0x2E>": 49,
204
+ "<0x2F>": 50,
205
+ "<0x30>": 51,
206
+ "<0x31>": 52,
207
+ "<0x32>": 53,
208
+ "<0x33>": 54,
209
+ "<0x34>": 55,
210
+ "<0x35>": 56,
211
+ "<0x36>": 57,
212
+ "<0x37>": 58,
213
+ "<0x38>": 59,
214
+ "<0x39>": 60,
215
+ "<0x3A>": 61,
216
+ "<0x3B>": 62,
217
+ "<0x3C>": 63,
218
+ "<0x3D>": 64,
219
+ "<0x3E>": 65,
220
+ "<0x3F>": 66,
221
+ "<0x40>": 67,
222
+ "<0x41>": 68,
223
+ "<0x42>": 69,
224
+ "<0x43>": 70,
225
+ "<0x44>": 71,
226
+ "<0x45>": 72,
227
+ "<0x46>": 73,
228
+ "<0x47>": 74,
229
+ "<0x48>": 75,
230
+ "<0x49>": 76,
231
+ "<0x4A>": 77,
232
+ "<0x4B>": 78,
233
+ "<0x4C>": 79,
234
+ "<0x4D>": 80,
235
+ "<0x4E>": 81,
236
+ "<0x4F>": 82,
237
+ "<0x50>": 83,
238
+ "<0x51>": 84,
239
+ "<0x52>": 85,
240
+ "<0x53>": 86,
241
+ "<0x54>": 87,
242
+ "<0x55>": 88,
243
+ "<0x56>": 89,
244
+ "<0x57>": 90,
245
+ "<0x58>": 91,
246
+ "<0x59>": 92,
247
+ "<0x5A>": 93,
248
+ "<0x5B>": 94,
249
+ "<0x5C>": 95,
250
+ "<0x5D>": 96,
251
+ "<0x5E>": 97,
252
+ "<0x5F>": 98,
253
+ "<0x60>": 99,
254
+ "<0x61>": 100,
255
+ "<0x62>": 101,
256
+ "<0x63>": 102,
257
+ "<0x64>": 103,
258
+ "<0x65>": 104,
259
+ "<0x66>": 105,
260
+ "<0x67>": 106,
261
+ "<0x68>": 107,
262
+ "<0x69>": 108,
263
+ "<0x6A>": 109,
264
+ "<0x6B>": 110,
265
+ "<0x6C>": 111,
266
+ "<0x6D>": 112,
267
+ "<0x6E>": 113,
268
+ "<0x6F>": 114,
269
+ "<0x70>": 115,
270
+ "<0x71>": 116,
271
+ "<0x72>": 117,
272
+ "<0x73>": 118,
273
+ "<0x74>": 119,
274
+ "<0x75>": 120,
275
+ "<0x76>": 121,
276
+ "<0x77>": 122,
277
+ "<0x78>": 123,
278
+ "<0x79>": 124,
279
+ "<0x7A>": 125,
280
+ "<0x7B>": 126,
281
+ "<0x7C>": 127,
282
+ "<0x7D>": 128,
283
+ "<0x7E>": 129,
284
+ "<0x7F>": 130,
285
+ "<0x80>": 131,
286
+ "<0x81>": 132,
287
+ "<0x82>": 133,
288
+ "<0x83>": 134,
289
+ "<0x84>": 135,
290
+ "<0x85>": 136,
291
+ "<0x86>": 137,
292
+ "<0x87>": 138,
293
+ "<0x88>": 139,
294
+ "<0x89>": 140,
295
+ "<0x8A>": 141,
296
+ "<0x8B>": 142,
297
+ "<0x8C>": 143,
298
+ "<0x8D>": 144,
299
+ "<0x8E>": 145,
300
+ "<0x8F>": 146,
301
+ "<0x90>": 147,
302
+ "<0x91>": 148,
303
+ "<0x92>": 149,
304
+ "<0x93>": 150,
305
+ "<0x94>": 151,
306
+ "<0x95>": 152,
307
+ "<0x96>": 153,
308
+ "<0x97>": 154,
309
+ "<0x98>": 155,
310
+ "<0x99>": 156,
311
+ "<0x9A>": 157,
312
+ "<0x9B>": 158,
313
+ "<0x9C>": 159,
314
+ "<0x9D>": 160,
315
+ "<0x9E>": 161,
316
+ "<0x9F>": 162,
317
+ "<0xA0>": 163,
318
+ "<0xA1>": 164,
319
+ "<0xA2>": 165,
320
+ "<0xA3>": 166,
321
+ "<0xA4>": 167,
322
+ "<0xA5>": 168,
323
+ "<0xA6>": 169,
324
+ "<0xA7>": 170,
325
+ "<0xA8>": 171,
326
+ "<0xA9>": 172,
327
+ "<0xAA>": 173,
328
+ "<0xAB>": 174,
329
+ "<0xAC>": 175,
330
+ "<0xAD>": 176,
331
+ "<0xAE>": 177,
332
+ "<0xAF>": 178,
333
+ "<0xB0>": 179,
334
+ "<0xB1>": 180,
335
+ "<0xB2>": 181,
336
+ "<0xB3>": 182,
337
+ "<0xB4>": 183,
338
+ "<0xB5>": 184,
339
+ "<0xB6>": 185,
340
+ "<0xB7>": 186,
341
+ "<0xB8>": 187,
342
+ "<0xB9>": 188,
343
+ "<0xBA>": 189,
344
+ "<0xBB>": 190,
345
+ "<0xBC>": 191,
346
+ "<0xBD>": 192,
347
+ "<0xBE>": 193,
348
+ "<0xBF>": 194,
349
+ "<0xC0>": 195,
350
+ "<0xC1>": 196,
351
+ "<0xC2>": 197,
352
+ "<0xC3>": 198,
353
+ "<0xC4>": 199,
354
+ "<0xC5>": 200,
355
+ "<0xC6>": 201,
356
+ "<0xC7>": 202,
357
+ "<0xC8>": 203,
358
+ "<0xC9>": 204,
359
+ "<0xCA>": 205,
360
+ "<0xCB>": 206,
361
+ "<0xCC>": 207,
362
+ "<0xCD>": 208,
363
+ "<0xCE>": 209,
364
+ "<0xCF>": 210,
365
+ "<0xD0>": 211,
366
+ "<0xD1>": 212,
367
+ "<0xD2>": 213,
368
+ "<0xD3>": 214,
369
+ "<0xD4>": 215,
370
+ "<0xD5>": 216,
371
+ "<0xD6>": 217,
372
+ "<0xD7>": 218,
373
+ "<0xD8>": 219,
374
+ "<0xD9>": 220,
375
+ "<0xDA>": 221,
376
+ "<0xDB>": 222,
377
+ "<0xDC>": 223,
378
+ "<0xDD>": 224,
379
+ "<0xDE>": 225,
380
+ "<0xDF>": 226,
381
+ "<0xE0>": 227,
382
+ "<0xE1>": 228,
383
+ "<0xE2>": 229,
384
+ "<0xE3>": 230,
385
+ "<0xE4>": 231,
386
+ "<0xE5>": 232,
387
+ "<0xE6>": 233,
388
+ "<0xE7>": 234,
389
+ "<0xE8>": 235,
390
+ "<0xE9>": 236,
391
+ "<0xEA>": 237,
392
+ "<0xEB>": 238,
393
+ "<0xEC>": 239,
394
+ "<0xED>": 240,
395
+ "<0xEE>": 241,
396
+ "<0xEF>": 242,
397
+ "<0xF0>": 243,
398
+ "<0xF1>": 244,
399
+ "<0xF2>": 245,
400
+ "<0xF3>": 246,
401
+ "<0xF4>": 247,
402
+ "<0xF5>": 248,
403
+ "<0xF6>": 249,
404
+ "<0xF7>": 250,
405
+ "<0xF8>": 251,
406
+ "<0xF9>": 252,
407
+ "<0xFA>": 253,
408
+ "<0xFB>": 254,
409
+ "<0xFC>": 255,
410
+ "<0xFD>": 256,
411
+ "<0xFE>": 257,
412
+ "<0xFF>": 258,
413
+ "▁▁": 259,
414
+ "▁t": 260,
415
+ "er": 261,
416
+ "in": 262,
417
+ "▁a": 263,
418
+ "en": 264,
419
+ "on": 265,
420
+ "▁th": 266,
421
+ "es": 267,
422
+ "▁▁▁▁": 268,
423
+ "▁s": 269,
424
+ "▁d": 270,
425
+ "at": 271,
426
+ "or": 272,
427
+ "an": 273,
428
+ "▁c": 274,
429
+ "is": 275,
430
+ "re": 276,
431
+ "it": 277,
432
+ "▁the": 278,
433
+ "ar": 279,
434
+ "le": 280,
435
+ "▁w": 281,
436
+ "▁p": 282,
437
+ "ou": 283,
438
+ "al": 284,
439
+ "▁f": 285,
440
+ "▁m": 286,
441
+ "ed": 287,
442
+ "▁o": 288,
443
+ "▁b": 289,
444
+ "om": 290,
445
+ "ion": 291,
446
+ "ing": 292,
447
+ "ic": 293,
448
+ "as": 294,
449
+ "el": 295,
450
+ "ent": 296,
451
+ "▁in": 297,
452
+ "▁h": 298,
453
+ "nd": 299,
454
+ "et": 300,
455
+ "▁l": 301,
456
+ "▁n": 302,
457
+ "st": 303,
458
+ "▁to": 304,
459
+ "ch": 305,
460
+ "▁I": 306,
461
+ "ro": 307,
462
+ "▁▁▁▁▁▁▁▁": 308,
463
+ "il": 309,
464
+ "▁of": 310,
465
+ "de": 311,
466
+ "ct": 312,
467
+ "▁(": 313,
468
+ "am": 314,
469
+ "▁C": 315,
470
+ "▁de": 316,
471
+ "▁S": 317,
472
+ "▁u": 318,
473
+ "▁A": 319,
474
+ "▁\\": 320,
475
+ "▁e": 321,
476
+ "▁and": 322,
477
+ "▁T": 323,
478
+ "ol": 324,
479
+ "▁v": 325,
480
+ "im": 326,
481
+ "ot": 327,
482
+ "ad": 328,
483
+ "ut": 329,
484
+ "▁g": 330,
485
+ "em": 331,
486
+ "ur": 332,
487
+ "id": 333,
488
+ "▁*": 334,
489
+ "ig": 335,
490
+ "ra": 336,
491
+ "▁re": 337,
492
+ "▁is": 338,
493
+ "qu": 339,
494
+ "ow": 340,
495
+ "▁M": 341,
496
+ "est": 342,
497
+ "▁y": 343,
498
+ "se": 344,
499
+ "ve": 345,
500
+ "ce": 346,
501
+ "ie": 347,
502
+ "un": 348,
503
+ "▁P": 349,
504
+ "▁B": 350,
505
+ "ag": 351,
506
+ "ul": 352,
507
+ "▁=": 353,
508
+ "he": 354,
509
+ "end": 355,
510
+ "ode": 356,
511
+ "ter": 357,
512
+ "ment": 358,
513
+ "os": 359,
514
+ "▁D": 360,
515
+ "if": 361,
516
+ "ation": 362,
517
+ "▁for": 363,
518
+ "▁r": 364,
519
+ "▁L": 365,
520
+ "▁you": 366,
521
+ "▁be": 367,
522
+ "ly": 368,
523
+ "ver": 369,
524
+ "ab": 370,
525
+ "te": 371,
526
+ "▁it": 372,
527
+ "▁on": 373,
528
+ "ri": 374,
529
+ "us": 375,
530
+ "▁\"": 376,
531
+ "▁wh": 377,
532
+ "▁con": 378,
533
+ "▁H": 379,
534
+ "▁st": 380,
535
+ "ir": 381,
536
+ "▁E": 382,
537
+ "▁F": 383,
538
+ "ck": 384,
539
+ "▁an": 385,
540
+ "th": 386,
541
+ "eg": 387,
542
+ "ay": 388,
543
+ "ith": 389,
544
+ "▁R": 390,
545
+ "ist": 391,
546
+ "and": 392,
547
+ "▁that": 393,
548
+ "▁al": 394,
549
+ "▁$": 395,
550
+ "▁#": 396,
551
+ "od": 397,
552
+ "um": 398,
553
+ "▁W": 399,
554
+ "ht": 400,
555
+ "code": 401,
556
+ "▁G": 402,
557
+ "ate": 403,
558
+ "ess": 404,
559
+ "▁N": 405,
560
+ "ere": 406,
561
+ "pp": 407,
562
+ "▁as": 408,
563
+ "▁se": 409,
564
+ "▁pro": 410,
565
+ "▁with": 411,
566
+ "pe": 412,
567
+ "▁k": 413,
568
+ "ers": 414,
569
+ "pt": 415,
570
+ ");": 416,
571
+ "lo": 417,
572
+ "▁▁▁▁▁": 418,
573
+ "▁com": 419,
574
+ "ame": 420,
575
+ "▁`": 421,
576
+ "▁Com": 422,
577
+ "ia": 423,
578
+ "ant": 424,
579
+ "▁la": 425,
580
+ "▁{": 426,
581
+ "▁en": 427,
582
+ "ction": 428,
583
+ "▁ex": 429,
584
+ "ld": 430,
585
+ "ub": 431,
586
+ "▁j": 432,
587
+ "la": 433,
588
+ "ue": 434,
589
+ "▁J": 435,
590
+ "ich": 436,
591
+ "▁do": 437,
592
+ "▁O": 438,
593
+ "▁qu": 439,
594
+ "iv": 440,
595
+ "ort": 441,
596
+ "art": 442,
597
+ "▁un": 443,
598
+ "▁##": 444,
599
+ "▁this": 445,
600
+ "ke": 446,
601
+ "▁ha": 447,
602
+ "▁-": 448,
603
+ "out": 449,
604
+ "▁The": 450,
605
+ "▁not": 451,
606
+ "▁ne": 452,
607
+ "ill": 453,
608
+ "▁le": 454,
609
+ "ci": 455,
610
+ "rom": 456,
611
+ "ine": 457,
612
+ "//": 458,
613
+ "op": 459,
614
+ "egin": 460,
615
+ "▁Comment": 461,
616
+ "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁": 462,
617
+ "begin": 463,
618
+ "ст": 464,
619
+ "ass": 465,
620
+ "iz": 466,
621
+ ").": 467,
622
+ "og": 468,
623
+ "▁п": 469,
624
+ "▁or": 470,
625
+ "▁was": 471,
626
+ "▁at": 472,
627
+ "our": 473,
628
+ "▁i": 474,
629
+ "ain": 475,
630
+ "▁K": 476,
631
+ "на": 477,
632
+ "▁V": 478,
633
+ "ge": 479,
634
+ "▁su": 480,
635
+ "ap": 481,
636
+ "age": 482,
637
+ "ould": 483,
638
+ "ne": 484,
639
+ "av": 485,
640
+ "xt": 486,
641
+ "ore": 487,
642
+ "ile": 488,
643
+ "--": 489,
644
+ "▁в": 490,
645
+ "▁by": 491,
646
+ "li": 492,
647
+ "ath": 493,
648
+ "ра": 494,
649
+ "ber": 495,
650
+ "ach": 496,
651
+ "all": 497,
652
+ "▁Th": 498,
653
+ "ult": 499
654
+ },
655
+ "merges": [
656
+ "▁t he",
657
+ "▁a nd",
658
+ "at ion",
659
+ "▁f or",
660
+ "▁y ou",
661
+ "▁c on",
662
+ "▁th at",
663
+ "▁p ro",
664
+ "▁w ith",
665
+ "▁c om",
666
+ "▁C om",
667
+ "ct ion",
668
+ "▁th is",
669
+ "▁T he",
670
+ "▁n ot",
671
+ "eg in",
672
+ "▁Com ment",
673
+ "▁w as",
674
+ "ou ld",
675
+ "▁▁ ▁▁",
676
+ "▁▁▁▁ ▁▁▁▁",
677
+ "▁▁▁▁▁▁▁▁ ▁▁▁▁▁▁▁▁"
678
+ ]
679
+ }
680
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": true,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ },
30
+ "32000": {
31
+ "content": "<video>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ },
38
+ "32001": {
39
+ "content": "<image>",
40
+ "lstrip": false,
41
+ "normalized": false,
42
+ "rstrip": false,
43
+ "single_word": false,
44
+ "special": true
45
+ }
46
+ },
47
+ "bos_token": "<s>",
48
+ "clean_up_tokenization_spaces": false,
49
+ "eos_token": "</s>",
50
+ "extra_special_tokens": {
51
+ "image_token": "<image>",
52
+ "video_token": "<video>"
53
+ },
54
+ "image_token": "<image>",
55
+ "legacy": false,
56
+ "model_max_length": 10250,
57
+ "pad_token": "<unk>",
58
+ "padding_side": "left",
59
+ "processor_class": "LlavaNextVideoProcessor",
60
+ "sp_model_kwargs": {},
61
+ "spaces_between_special_tokens": false,
62
+ "tokenizer_class": "LlamaTokenizer",
63
+ "unk_token": "<unk>",
64
+ "use_default_system_prompt": false,
65
+ "video_token": "<video>"
66
+ }
video_preprocessor_config.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": {
3
+ "height": 336,
4
+ "width": 336
5
+ },
6
+ "data_format": "channels_first",
7
+ "default_to_square": false,
8
+ "device": null,
9
+ "do_center_crop": true,
10
+ "do_convert_rgb": true,
11
+ "do_normalize": true,
12
+ "do_pad": true,
13
+ "do_rescale": true,
14
+ "do_resize": true,
15
+ "do_sample_frames": false,
16
+ "fps": null,
17
+ "image_grid_pinpoints": [
18
+ [
19
+ 336,
20
+ 672
21
+ ],
22
+ [
23
+ 672,
24
+ 336
25
+ ],
26
+ [
27
+ 672,
28
+ 672
29
+ ],
30
+ [
31
+ 1008,
32
+ 336
33
+ ],
34
+ [
35
+ 336,
36
+ 1008
37
+ ]
38
+ ],
39
+ "image_mean": [
40
+ 0.48145466,
41
+ 0.4578275,
42
+ 0.40821073
43
+ ],
44
+ "image_processor_type": "LlavaNextImageProcessor",
45
+ "image_std": [
46
+ 0.26862954,
47
+ 0.26130258,
48
+ 0.27577711
49
+ ],
50
+ "input_data_format": null,
51
+ "num_frames": null,
52
+ "pad_size": null,
53
+ "processor_class": "LlavaNextVideoProcessor",
54
+ "resample": 3,
55
+ "rescale_factor": 0.00392156862745098,
56
+ "return_metadata": false,
57
+ "return_tensors": null,
58
+ "size": {
59
+ "shortest_edge": 336
60
+ },
61
+ "video_metadata": null,
62
+ "video_processor_type": "LlavaNextVideoVideoProcessor"
63
+ }