Upload 18 files

Upload small random models for testing the UniDiffuserPipeline

image_encoder/config.json

@@ -0,0 +1,22 @@
+{
+  "architectures": [
+    "CLIPVisionModel"
+  ],
+  "attention_dropout": 0.1,
+  "dropout": 0.1,
+  "hidden_act": "quick_gelu",
+  "hidden_size": 32,
+  "image_size": 32,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 37,
+  "layer_norm_eps": 1e-05,
+  "model_type": "clip_vision_model",
+  "num_attention_heads": 4,
+  "num_channels": 3,
+  "num_hidden_layers": 5,
+  "patch_size": 2,
+  "projection_dim": 32,
+  "torch_dtype": "float32",
+  "transformers_version": "4.23.1"
+}

image_encoder/merges.txt

@@ -0,0 +1,647 @@
+#version: 0.2
+Ġ t
+Ġt h
+Ġ a
+Ġth e</w>
+i n
+Ġ o
+Ġ ,</w>
+Ġ s
+e d</w>
+Ġ w
+e r
+Ġ .</w>
+Ġ i
+r e
+Ġ c
+n d</w>
+Ġ f
+Ġ b
+a t
+Ġo f</w>
+e r</w>
+e n
+a r
+o r
+i t
+Ġ p
+Ġ h
+Ġa nd</w>
+o n
+in g</w>
+a n
+r o
+Ġ m
+Ġ d
+e s</w>
+Ġi n</w>
+o n</w>
+Ġt o</w>
+o u
+i s
+Ġ a</w>
+i c
+Ġ T
+a l
+Ġ l
+Ġ =</w>
+Ġ re
+Ġ "</w>
+e s
+Ġ S
+a s</w>
+a l</w>
+i l
+e l
+i on</w>
+Ġ A
+Ġ C
+Ġ 1
+Ġ Ċ</w>
+u r
+ĠT h
+Ġ n
+a s
+Ġ @
+e c
+o m
+a c
+Ġ e
+Ġw as</w>
+Ġ M
+o r</w>
+a n</w>
+a m
+e n</w>
+o l
+Ġ in
+Ġ g
+Ġ '</w>
+Ġ B
+l y</w>
+a t</w>
+i v
+t s</w>
+ĠTh e</w>
+u s
+- @</w>
+Ġ@ -@</w>
+i s</w>
+Ġ I
+Ġw h
+i g
+Ġ H
+Ġs t
+o s
+u n
+t h
+Ġ P
+Ġw it
+Ġth at</w>
+i r
+Ġa s</w>
+e m
+Ġo n</w>
+r a
+Ġf or</w>
+Ġ R
+e t
+o w
+Ġ 2
+i d
+Ġ D
+l e</w>
+Ġwit h</w>
+l a
+en t</w>
+i m
+Ġ F
+e a
+i on
+Ġb y</w>
+Ġ )</w>
+Ġ (</w>
+Ġa l
+Ġc on
+en t
+Ġ W
+Ġi s</w>
+er e</w>
+Ġ G
+Ġ N
+Ġ L
+Ġh a
+er s</w>
+r i
+t h</w>
+t ed</w>
+u c
+Ġ J
+Ġ1 9
+e v
+u l
+Ġ v
+c e</w>
+at ion</w>
+ro m</w>
+Ġb e
+Ġ E
+i n</w>
+Ġth e
+Ġf rom</w>
+Ġ O
+t er</w>
+Ġp ro
+Ġa r
+a d
+Ġc om
+i c</w>
+a g
+Ġh is</w>
+Ġs h
+Ġa t</w>
+o v
+i es</w>
+o o
+p p
+s t
+c h
+Ġ r
+Ġ2 0
+a y</w>
+i f
+Ġw ere</w>
+Ġc h
+u t</w>
+s t</w>
+u t
+d s</w>
+o p
+u m
+Ġi t</w>
+o c
+t er
+l e
+ig h
+u d
+Ġe x
+ion s</w>
+at e</w>
+it y</w>
+at ed</w>
+Ġ un
+e p
+q u
+Ġn o
+Ġ K
+iv e</w>
+is t
+Ġo n
+am e</w>
+ou n
+i r</w>
+a b
+Ġ â
+in g
+Ġh e</w>
+l d</w>
+u g
+ic h</w>
+Ġa n</w>
+e d
+Ġ k
+Ġâ Ģ
+Ġha d</w>
+v e</w>
+a in
+Ġs e
+t ion</w>
+or e</w>
+re s
+Ġwh ich</w>
+ĠI n</w>
+o d
+th er</w>
+a k
+Ġs p
+a r</w>
+Ġ y
+ĠC h
+on g</w>
+Ġa c
+es t</w>
+Ġ U
+a p
+f f
+al ly</w>
+r it
+ĠS t
+u b
+g e</w>
+b er</w>
+e t</w>
+Ġb e</w>
+e ar
+Ġre c
+er s
+Ġf ir
+o t
+Ġar e</w>
+Ġa n
+c h</w>
+o g
+i a</w>
+es t
+in e</w>
+il l
+an d
+e l</w>
+ar y</w>
+e w</w>
+i d</w>
+Ġf or
+Ġ ;</w>
+Ġcom p
+Ġ V
+Ġin c
+t r
+Ġ20 0
+Ġthe ir</w>
+u s</w>
+Ġb ut</w>
+r an
+ic al</w>
+Ġfir st</w>
+Ġd e
+Ġin t
+Ġ ro
+s o</w>
+ĠâĢ ĵ</w>
+Ġno t</w>
+d ing</w>
+f ter</w>
+ur e</w>
+Ġp ar
+Ġ :</w>
+i an</w>
+Ġt w
+ou ld</w>
+Ġal so</w>
+Ġi ts</w>
+Ġw or
+u m</w>
+Ġo r</w>
+os t</w>
+0 0</w>
+ou r
+ar d</w>
+Ġre s
+m p
+u e</w>
+Ġa b
+is h</w>
+Ġcon t
+Ġa d
+ow n</w>
+al l</w>
+ou g
+Ġh er</w>
+as t</w>
+Ġ en
+om e</w>
+al l
+d ed</w>
+o w</w>
+Ġha ve</w>
+Ġ us
+ea r</w>
+ac k</w>
+d uc
+i al</w>
+s s
+en ts</w>
+a in</w>
+t ing</w>
+Ġon e</w>
+es s
+Ġh as</w>
+igh t</w>
+a v
+Ġe v
+ou t</w>
+a y
+en ce</w>
+Ġbe en</w>
+e w
+Ġtw o</w>
+Ġc l
+d er</w>
+im e</w>
+k s</w>
+es s</w>
+is h
+. @</w>
+Ġ@ .@</w>
+Ġp la
+Ġp l
+Ġo r
+u p</w>
+m ent</w>
+ur ing</w>
+ol l
+ĠI n
+Ġth is</w>
+Ġb ec
+Ġcom m
+Ġd is
+at er</w>
+ag e</w>
+Ġa pp
+ou s</w>
+e y</w>
+i l</w>
+p er
+ĠA l
+ion al</w>
+l ud
+el y</w>
+t t
+il e</w>
+i z
+Ġ j
+Ġwh o</w>
+Ġa g
+i b
+Ġthe y</w>
+f or
+Ġo v
+at h
+e g
+Ġs c
+i p
+Ġ20 1
+Ġ 3
+Ġp er
+or y</w>
+Ġd es
+id e</w>
+Ġs er
+s e</w>
+ĠH e</w>
+la nd</w>
+at ions</w>
+r ic
+i t</w>
+re s</w>
+er ed</w>
+Ġp re
+ĠS h
+an ce</w>
+or t</w>
+an t</w>
+, @</w>
+Ġ@ ,@</w>
+el l</w>
+Ġ Y
+n ed</w>
+el l
+it e</w>
+Ġinc lud
+Ġre p
+Ġa fter</w>
+Ġs uc
+re e</w>
+an y</w>
+i m</w>
+or t
+Ġ1 8
+Ġs u
+ad e</w>
+ou r</w>
+ĠU n
+ĠI t</w>
+i k
+ĠM ar
+em ber</w>
+Ġ 1</w>
+e en</w>
+a nd</w>
+Ġs ec
+ic e</w>
+Ġt ime</w>
+ĠA n
+Ġint o</w>
+Ġf in
+Ġo ther</w>
+Ġa tt
+il l</w>
+re n
+ac h
+as s
+er al</w>
+es e</w>
+s h
+al s</w>
+it ion</w>
+oug h</w>
+l es</w>
+am p
+Ġw ould</w>
+Ġm ore</w>
+ro ug
+ri b
+er y</w>
+ac e</w>
+Ġ A</w>
+Ġpla y
+it ed</w>
+k ed</w>
+is t</w>
+i ed</w>
+Ġ 2</w>
+as ed</w>
+ing s</w>
+an g
+a m</w>
+i p</w>
+Ġb o
+ab le</w>
+t y</w>
+Ġch ar
+Ġc ent
+et w
+at es</w>
+ro p
+Ġ I</w>
+u nd</w>
+ĠA m
+c es</w>
+o in
+Ġin ter
+u p
+c t
+on e</w>
+Ġt ra
+an t
+ec t
+Ġal l</w>
+e f
+Ġcon s
+ub l
+n ing</w>
+an s</w>
+Ġf e
+us t</w>
+Ġ 0
+Ġre m
+as e</w>
+on g
+Ġwh en</w>
+e b
+ĠW h
+Ġe ar
+ev er</w>
+Ġov er</w>
+Ġk n
+a us
+Ġp os
+a d</w>
+er m
+Ġsh e</w>
+Ġ ra
+Ġd uring</w>
+as on</w>
+v i
+Ġex p
+Ġl ea
+Ġ el
+Ġ 4
+Ġon ly</w>
+o nd</w>
+Ġd ec
+Ġac c
+Ġo ff
+is s
+Ġf l
+ĠE n
+o t</w>
+en s
+os e</w>
+ak e</w>
+o m</w>
+Ġs ev
+ac h</w>
+etw een</w>
+er n
+Ġ 3</w>
+Ġp r
+Ġg ro
+r uc
+Ġd i
+Ġ19 9
+ĠA r
+Ġg ame</w>
+Ġh im</w>
+oo k</w>
+Ġ up</w>
+Ġab out</w>
+Ġre l
+for m
+Ġth ree</w>
+at t
+ĠC om
+Ġs a
+ear s</w>
+Ġ 5
+r y</w>
+Ġi mp
+Ġm ost</w>
+f er
+Ġp res
+Ġf il
+Ġb etween</w>
+Ġbe g
+p h
+or s</w>
+Ġth an</w>
+Ġrec or
+o b
+er ic
+at ing</w>
+Ġth roug
+k ing</w>
+Ġo ut</w>
+Ġn um
+oo d</w>
+oll ow
+ac t
+u il
+Ġc re
+ol og
+at ional</w>
+Ġpro duc
+Ġwh ile</w>
+Ġl ater</w>
+Ġw rit
+e x
+Ġst ar
+Ġsp ec
+e e
+ish ed</w>
+Ġre g
+is ion</w>
+ou th</w>
+Ġre le
+Ġa ss
+Ġse ason</w>
+Ġm ade</w>
+il y</w>
+r u
+o y
+t ur
+t e</w>
+Ġ qu
+Ġm ov
+ur y</w>
+ĠAm eric
+em ent</w>
+c c
+ou nd</w>
+Ġl ar
+Ġfor m
+ec t</w>
+Ġde f
+Ġm us
+ĠP ar
+Ġm e
+Ġs ub
+w ay</w>
+o p</w>
+o h
+el d</w>
+i e</w>
+em p
+am es</w>
+er n</w>
+Ġn or
+iv ed</w>
+ev el
+Ġsuc h</w>
+ar ds</w>
+Ġin d
+ik e</w>
+Ġg en
+er t
+Ġy ear</w>
+Ġus ed</w>
+Ġn ew</w>
+Ġ 5</w>
+Ġal b
+s p
+y p
+Ġwit h
+Ġwh ere</w>
+ic s</w>
+ĠTh is</w>
+Ġthe m</w>
+w n</w>

image_encoder/preprocessor_config.json

@@ -0,0 +1,21 @@
+{
+  "crop_size": 32,
+  "do_center_crop": true,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_resize": true,
+  "feature_extractor_type": "CLIPFeatureExtractor",
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "processor_class": "CLIPProcessor",
+  "resample": 3,
+  "size": 32
+}

image_encoder/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:217ebf83168efb58b9399b4dd0b0776af23c09ccd51bfd2d8d7720905addb7b9
+size 201209

image_encoder/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

image_encoder/tokenizer_config.json

@@ -0,0 +1,35 @@
+{
+  "add_prefix_space": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "do_lower_case": true,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "errors": "replace",
+  "model_max_length": 77,
+  "name_or_path": "hf-internal-testing/tiny-random-clip",
+  "pad_token": "<|endoftext|>",
+  "processor_class": "CLIPProcessor",
+  "special_tokens_map_file": "/home/lysandre/.cache/huggingface/transformers/18a566598f286c9139f88160c99f84eec492a26bd22738fa9cb44d5b7e0a5c76.cce1206abbad28826f000510f22f354e53e66a97f7c23745a7dfe27609cc07f5",
+  "tokenizer_class": "CLIPTokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

image_encoder/vocab.json

@@ -0,0 +1,1002 @@
+{
+  "!": 2,
+  "!</w>": 345,
+  "\"": 3,
+  "\"</w>": 344,
+  "#": 4,
+  "#</w>": 325,
+  "$": 5,
+  "$</w>": 348,
+  "%": 6,
+  "%</w>": 351,
+  "&": 7,
+  "&</w>": 352,
+  "'": 8,
+  "'</w>": 296,
+  "(": 9,
+  "(</w>": 318,
+  ")": 10,
+  ")</w>": 330,
+  "*": 11,
+  "*</w>": 327,
+  "+": 12,
+  "+</w>": 341,
+  ",": 13,
+  ",</w>": 279,
+  ",@</w>": 754,
+  "-": 14,
+  "-</w>": 276,
+  "-@</w>": 439,
+  ".": 15,
+  ".</w>": 253,
+  ".@</w>": 695,
+  "/": 16,
+  "/</w>": 350,
+  "0": 17,
+  "00</w>": 647,
+  "0</w>": 216,
+  "1": 18,
+  "1</w>": 222,
+  "2": 19,
+  "2</w>": 231,
+  "3": 20,
+  "3</w>": 243,
+  "4": 21,
+  "4</w>": 233,
+  "5": 22,
+  "5</w>": 240,
+  "6": 23,
+  "6</w>": 226,
+  "7": 24,
+  "7</w>": 215,
+  "8": 25,
+  "8</w>": 236,
+  "9": 26,
+  "9</w>": 242,
+  ":": 27,
+  ":</w>": 353,
+  ";": 28,
+  ";</w>": 317,
+  "<": 29,
+  "<</w>": 340,
+  "<|endoftext|>": 1,
+  "<|startoftext|>": 0,
+  "=": 30,
+  "=</w>": 342,
+  ">": 31,
+  "></w>": 300,
+  "?": 32,
+  "?</w>": 346,
+  "@": 33,
+  "@</w>": 320,
+  "A": 34,
+  "A</w>": 227,
+  "B": 35,
+  "B</w>": 258,
+  "C": 36,
+  "C</w>": 239,
+  "D": 37,
+  "D</w>": 255,
+  "E": 38,
+  "E</w>": 246,
+  "F": 39,
+  "F</w>": 213,
+  "G": 40,
+  "G</w>": 283,
+  "H": 41,
+  "H</w>": 219,
+  "I": 42,
+  "I</w>": 237,
+  "J": 43,
+  "J</w>": 251,
+  "K": 44,
+  "K</w>": 254,
+  "L": 45,
+  "L</w>": 218,
+  "M": 46,
+  "M</w>": 234,
+  "N": 47,
+  "N</w>": 238,
+  "O": 48,
+  "O</w>": 265,
+  "P": 49,
+  "P</w>": 245,
+  "Q": 50,
+  "Q</w>": 309,
+  "R": 51,
+  "R</w>": 264,
+  "S": 52,
+  "S</w>": 230,
+  "T": 53,
+  "T</w>": 235,
+  "U": 54,
+  "U</w>": 268,
+  "V": 55,
+  "V</w>": 248,
+  "W": 56,
+  "W</w>": 274,
+  "X": 57,
+  "X</w>": 263,
+  "Y": 58,
+  "Y</w>": 310,
+  "Z": 59,
+  "Z</w>": 207,
+  "[": 60,
+  "[</w>": 270,
+  "\\": 61,
+  "\\</w>": 338,
+  "]": 62,
+  "]</w>": 289,
+  "^": 63,
+  "^</w>": 331,
+  "_": 64,
+  "_</w>": 334,
+  "`": 65,
+  "`</w>": 347,
+  "a": 66,
+  "a</w>": 197,
+  "ab": 555,
+  "able</w>": 820,
+  "ac": 420,
+  "ace</w>": 806,
+  "ach": 791,
+  "ach</w>": 885,
+  "ack</w>": 670,
+  "act": 929,
+  "ad": 508,
+  "ad</w>": 860,
+  "ade</w>": 771,
+  "ag": 511,
+  "age</w>": 710,
+  "ain": 568,
+  "ain</w>": 675,
+  "ak": 577,
+  "ake</w>": 882,
+  "al": 397,
+  "al</w>": 405,
+  "all": 664,
+  "all</w>": 658,
+  "ally</w>": 588,
+  "als</w>": 796,
+  "am": 426,
+  "am</w>": 817,
+  "ame</w>": 552,
+  "ames</w>": 976,
+  "amp": 800,
+  "an": 384,
+  "an</w>": 425,
+  "ance</w>": 751,
+  "and": 609,
+  "and</w>": 780,
+  "ang": 816,
+  "ans</w>": 844,
+  "ant": 837,
+  "ant</w>": 753,
+  "any</w>": 766,
+  "ap": 586,
+  "ar": 376,
+  "ar</w>": 579,
+  "ard</w>": 649,
+  "ards</w>": 982,
+  "ary</w>": 611,
+  "as": 416,
+  "as</w>": 404,
+  "ase</w>": 849,
+  "ased</w>": 814,
+  "ason</w>": 865,
+  "ass": 792,
+  "ast</w>": 661,
+  "at": 372,
+  "at</w>": 434,
+  "ate</w>": 541,
+  "ated</w>": 543,
+  "ater</w>": 709,
+  "ates</w>": 825,
+  "ath": 730,
+  "ating</w>": 922,
+  "ation</w>": 497,
+  "ational</w>": 933,
+  "ations</w>": 744,
+  "att": 903,
+  "aus": 858,
+  "av": 681,
+  "ay": 684,
+  "ay</w>": 523,
+  "b": 67,
+  "b</w>": 212,
+  "ber</w>": 593,
+  "c": 68,
+  "c</w>": 224,
+  "cc": 960,
+  "ce</w>": 496,
+  "ces</w>": 830,
+  "ch": 520,
+  "ch</w>": 603,
+  "ct": 834,
+  "d": 69,
+  "d</w>": 196,
+  "ded</w>": 665,
+  "der</w>": 690,
+  "ding</w>": 633,
+  "ds</w>": 530,
+  "duc": 671,
+  "e": 70,
+  "e</w>": 195,
+  "ea": 471,
+  "ear": 596,
+  "ear</w>": 669,
+  "ears</w>": 906,
+  "eb": 852,
+  "ec": 418,
+  "ect": 838,
+  "ect</w>": 964,
+  "ed": 563,
+  "ed</w>": 362,
+  "ee": 941,
+  "een</w>": 779,
+  "ef": 840,
+  "eg": 731,
+  "el": 407,
+  "el</w>": 610,
+  "eld</w>": 973,
+  "ell": 759,
+  "ell</w>": 756,
+  "ely</w>": 719,
+  "em": 455,
+  "ember</w>": 777,
+  "ement</w>": 959,
+  "emp": 975,
+  "en": 375,
+  "en</w>": 427,
+  "ence</w>": 685,
+  "ens": 880,
+  "ent": 478,
+  "ent</w>": 468,
+  "ents</w>": 674,
+  "ep": 545,
+  "er": 364,
+  "er</w>": 374,
+  "eral</w>": 793,
+  "ere</w>": 481,
+  "ered</w>": 748,
+  "eric": 921,
+  "erm": 861,
+  "ern": 887,
+  "ern</w>": 977,
+  "ers": 598,
+  "ers</w>": 486,
+  "ert": 986,
+  "ery</w>": 805,
+  "es": 402,
+  "es</w>": 388,
+  "ese</w>": 794,
+  "ess": 678,
+  "ess</w>": 693,
+  "est": 606,
+  "est</w>": 584,
+  "et": 460,
+  "et</w>": 594,
+  "etw": 824,
+  "etween</w>": 886,
+  "ev": 493,
+  "evel": 980,
+  "ever</w>": 855,
+  "ew": 687,
+  "ew</w>": 612,
+  "ex": 938,
+  "ey</w>": 713,
+  "f": 71,
+  "f</w>": 209,
+  "fer": 911,
+  "ff": 587,
+  "for": 728,
+  "form": 901,
+  "fter</w>": 634,
+  "g": 72,
+  "g</w>": 214,
+  "ge</w>": 592,
+  "h": 73,
+  "h</w>": 203,
+  "i": 74,
+  "i</w>": 205,
+  "ia</w>": 605,
+  "ial</w>": 672,
+  "ian</w>": 638,
+  "ib": 726,
+  "ic": 395,
+  "ic</w>": 510,
+  "ical</w>": 625,
+  "ice</w>": 782,
+  "ich</w>": 561,
+  "ics</w>": 996,
+  "id": 463,
+  "id</w>": 613,
+  "ide</w>": 739,
+  "ie</w>": 974,
+  "ied</w>": 812,
+  "ies</w>": 516,
+  "if": 524,
+  "ig": 444,
+  "igh": 537,
+  "ight</w>": 680,
+  "ik": 775,
+  "ike</w>": 984,
+  "il": 406,
+  "il</w>": 714,
+  "ile</w>": 721,
+  "ill": 608,
+  "ill</w>": 789,
+  "ily</w>": 950,
+  "im": 469,
+  "im</w>": 767,
+  "ime</w>": 691,
+  "in": 358,
+  "in</w>": 501,
+  "ine</w>": 607,
+  "ing": 557,
+  "ing</w>": 383,
+  "ings</w>": 815,
+  "ion": 472,
+  "ion</w>": 408,
+  "ional</w>": 717,
+  "ions</w>": 540,
+  "ip": 733,
+  "ip</w>": 818,
+  "ir": 453,
+  "ir</w>": 554,
+  "is": 393,
+  "is</w>": 441,
+  "ish": 694,
+  "ish</w>": 654,
+  "ished</w>": 942,
+  "ision</w>": 944,
+  "iss": 876,
+  "ist": 550,
+  "ist</w>": 811,
+  "it": 378,
+  "it</w>": 746,
+  "ite</w>": 760,
+  "ited</w>": 809,
+  "ition</w>": 797,
+  "ity</w>": 542,
+  "iv": 435,
+  "ive</w>": 549,
+  "ived</w>": 979,
+  "iz": 722,
+  "j": 75,
+  "j</w>": 288,
+  "k": 76,
+  "k</w>": 210,
+  "ked</w>": 810,
+  "king</w>": 924,
+  "ks</w>": 692,
+  "l": 77,
+  "l</w>": 201,
+  "la": 467,
+  "land</w>": 743,
+  "ld</w>": 559,
+  "le": 536,
+  "le</w>": 465,
+  "les</w>": 799,
+  "lud": 718,
+  "ly</w>": 433,
+  "m": 78,
+  "m</w>": 202,
+  "ment</w>": 701,
+  "mp": 651,
+  "n": 79,
+  "n</w>": 199,
+  "nd</w>": 369,
+  "ned</w>": 758,
+  "ning</w>": 843,
+  "o": 80,
+  "o</w>": 198,
+  "ob": 920,
+  "oc": 534,
+  "od": 575,
+  "og": 604,
+  "oh": 972,
+  "oin": 831,
+  "ol": 428,
+  "oll": 703,
+  "ollow": 928,
+  "olog": 932,
+  "om": 419,
+  "om</w>": 883,
+  "ome</w>": 663,
+  "on": 382,
+  "on</w>": 390,
+  "ond</w>": 872,
+  "one</w>": 835,
+  "ong": 850,
+  "ong</w>": 582,
+  "oo": 517,
+  "ood</w>": 927,
+  "ook</w>": 897,
+  "op": 531,
+  "op</w>": 971,
+  "or": 377,
+  "or</w>": 424,
+  "ore</w>": 571,
+  "ors</w>": 917,
+  "ort": 768,
+  "ort</w>": 752,
+  "ory</w>": 737,
+  "os": 447,
+  "ose</w>": 881,
+  "ost</w>": 646,
+  "ot": 600,
+  "ot</w>": 879,
+  "ou": 392,
+  "oug": 659,
+  "ough</w>": 798,
+  "ould</w>": 640,
+  "oun": 553,
+  "ound</w>": 961,
+  "our": 648,
+  "our</w>": 772,
+  "ous</w>": 712,
+  "out</w>": 683,
+  "outh</w>": 945,
+  "ov": 515,
+  "ow": 461,
+  "ow</w>": 666,
+  "own</w>": 657,
+  "oy": 952,
+  "p": 81,
+  "p</w>": 217,
+  "per": 715,
+  "ph": 916,
+  "pp": 518,
+  "q": 82,
+  "q</w>": 280,
+  "qu": 546,
+  "r": 83,
+  "r</w>": 204,
+  "ra": 457,
+  "ran": 624,
+  "re": 367,
+  "ree</w>": 765,
+  "ren": 790,
+  "res": 572,
+  "res</w>": 747,
+  "ri": 487,
+  "rib": 804,
+  "ric": 745,
+  "rit": 589,
+  "ro": 385,
+  "rom</w>": 498,
+  "rop": 826,
+  "roug": 803,
+  "ru": 951,
+  "ruc": 891,
+  "ry</w>": 908,
+  "s": 84,
+  "s</w>": 206,
+  "se</w>": 741,
+  "sh": 795,
+  "so</w>": 630,
+  "sp": 992,
+  "ss": 673,
+  "st": 519,
+  "st</w>": 528,
+  "t": 85,
+  "t</w>": 208,
+  "te</w>": 954,
+  "ted</w>": 489,
+  "ter": 535,
+  "ter</w>": 505,
+  "th": 449,
+  "th</w>": 488,
+  "ther</w>": 576,
+  "ting</w>": 676,
+  "tion</w>": 570,
+  "tr": 619,
+  "ts</w>": 436,
+  "tt": 720,
+  "tur": 953,
+  "ty</w>": 821,
+  "u": 86,
+  "u</w>": 229,
+  "ub": 591,
+  "ubl": 842,
+  "uc": 490,
+  "ud": 538,
+  "ue</w>": 652,
+  "ug": 560,
+  "uil": 930,
+  "ul": 494,
+  "um": 532,
+  "um</w>": 644,
+  "un": 448,
+  "und</w>": 828,
+  "up": 833,
+  "up</w>": 700,
+  "ur": 413,
+  "ure</w>": 635,
+  "uring</w>": 702,
+  "ury</w>": 957,
+  "us": 438,
+  "us</w>": 622,
+  "ust</w>": 846,
+  "ut": 529,
+  "ut</w>": 527,
+  "v": 87,
+  "v</w>": 232,
+  "ve</w>": 567,
+  "vi": 866,
+  "w": 88,
+  "w</w>": 250,
+  "way</w>": 970,
+  "wn</w>": 999,
+  "x": 89,
+  "x</w>": 269,
+  "y": 90,
+  "y</w>": 211,
+  "yp": 993,
+  "z": 91,
+  "z</w>": 228,
+  "|": 92,
+  "|</w>": 304,
+  "}": 93,
+  "}</w>": 336,
+  "~": 94,
+  "~</w>": 343,
+  "¡": 95,
+  "¡</w>": 220,
+  "¢": 96,
+  "¢</w>": 306,
+  "£": 97,
+  "£</w>": 323,
+  "¤": 98,
+  "¤</w>": 292,
+  "¥": 99,
+  "¥</w>": 339,
+  "¦": 100,
+  "¦</w>": 303,
+  "§": 101,
+  "§</w>": 275,
+  "¨": 102,
+  "¨</w>": 282,
+  "©": 103,
+  "©</w>": 259,
+  "ª": 104,
+  "ª</w>": 286,
+  "«": 105,
+  "«</w>": 266,
+  "¬": 106,
+  "¬</w>": 319,
+  "®": 107,
+  "®</w>": 329,
+  "¯": 108,
+  "¯</w>": 287,
+  "°": 109,
+  "°</w>": 298,
+  "±": 110,
+  "±</w>": 200,
+  "²": 111,
+  "²</w>": 284,
+  "³": 112,
+  "³</w>": 272,
+  "´": 113,
+  "´</w>": 307,
+  "µ": 114,
+  "µ</w>": 261,
+  "¶": 115,
+  "¶</w>": 301,
+  "·": 116,
+  "·</w>": 326,
+  "¸": 117,
+  "¸</w>": 257,
+  "¹": 118,
+  "¹</w>": 241,
+  "º": 119,
+  "º</w>": 260,
+  "»": 120,
+  "»</w>": 247,
+  "¼": 121,
+  "¼</w>": 305,
+  "½": 122,
+  "½</w>": 294,
+  "¾": 123,
+  "¾</w>": 316,
+  "¿": 124,
+  "¿</w>": 271,
+  "Â": 125,
+  "Ã": 126,
+  "Ä": 127,
+  "Å": 128,
+  "Æ": 129,
+  "Ç": 130,
+  "È": 131,
+  "É": 132,
+  "Ê": 133,
+  "Ë": 134,
+  "Ì": 135,
+  "Í": 136,
+  "Î": 137,
+  "Ï": 138,
+  "Ð": 139,
+  "Ñ": 140,
+  "Ö": 141,
+  "×": 142,
+  "Ø": 143,
+  "Ù": 144,
+  "Ü": 145,
+  "à": 146,
+  "á": 147,
+  "â": 148,
+  "ã": 149,
+  "ä": 150,
+  "å": 151,
+  "æ": 152,
+  "ç": 153,
+  "è": 154,
+  "é": 155,
+  "ë": 156,
+  "ì": 157,
+  "ï": 158,
+  "Ċ": 159,
+  "Ċ</w>": 349,
+  "Ġ": 160,
+  "Ġ\"</w>": 401,
+  "Ġ'</w>": 431,
+  "Ġ(</w>": 475,
+  "Ġ)</w>": 474,
+  "Ġ,</w>": 360,
+  "Ġ.</w>": 365,
+  "Ġ0": 847,
+  "Ġ1": 411,
+  "Ġ18": 769,
+  "Ġ19": 492,
+  "Ġ199": 893,
+  "Ġ1</w>": 778,
+  "Ġ2": 462,
+  "Ġ20": 522,
+  "Ġ200": 620,
+  "Ġ201": 734,
+  "Ġ2</w>": 813,
+  "Ġ3": 735,
+  "Ġ3</w>": 888,
+  "Ġ4": 870,
+  "Ġ5": 907,
+  "Ġ5</w>": 990,
+  "Ġ:</w>": 637,
+  "Ġ;</w>": 615,
+  "Ġ</w>": 333,
+  "Ġ=</w>": 399,
+  "Ġ@": 417,
+  "Ġ@,@</w>": 755,
+  "Ġ@-@</w>": 440,
+  "Ġ@.@</w>": 696,
+  "ĠA": 409,
+  "ĠA</w>": 807,
+  "ĠAl": 716,
+  "ĠAm": 829,
+  "ĠAmeric": 958,
+  "ĠAn": 784,
+  "ĠAr": 894,
+  "ĠB": 432,
+  "ĠC": 410,
+  "ĠCh": 581,
+  "ĠCom": 904,
+  "ĠD": 464,
+  "ĠE": 500,
+  "ĠEn": 878,
+  "ĠF": 470,
+  "ĠG": 482,
+  "ĠH": 445,
+  "ĠHe</w>": 742,
+  "ĠI": 442,
+  "ĠI</w>": 827,
+  "ĠIn": 704,
+  "ĠIn</w>": 574,
+  "ĠIt</w>": 774,
+  "ĠJ": 491,
+  "ĠK": 548,
+  "ĠL": 484,
+  "ĠM": 423,
+  "ĠMar": 776,
+  "ĠN": 483,
+  "ĠO": 504,
+  "ĠP": 450,
+  "ĠPar": 967,
+  "ĠR": 459,
+  "ĠS": 403,
+  "ĠSh": 750,
+  "ĠSt": 590,
+  "ĠT": 396,
+  "ĠTh": 414,
+  "ĠThe</w>": 437,
+  "ĠThis</w>": 997,
+  "ĠU": 585,
+  "ĠUn": 773,
+  "ĠV": 617,
+  "ĠW": 479,
+  "ĠWh": 853,
+  "ĠY": 757,
+  "Ġa": 356,
+  "Ġa</w>": 394,
+  "Ġab": 653,
+  "Ġabout</w>": 899,
+  "Ġac": 583,
+  "Ġacc": 874,
+  "Ġad": 656,
+  "Ġafter</w>": 763,
+  "Ġag": 725,
+  "Ġal": 476,
+  "Ġalb": 991,
+  "Ġall</w>": 839,
+  "Ġalso</w>": 641,
+  "Ġan": 602,
+  "Ġan</w>": 562,
+  "Ġand</w>": 381,
+  "Ġapp": 711,
+  "Ġar": 507,
+  "Ġare</w>": 601,
+  "Ġas</w>": 454,
+  "Ġass": 947,
+  "Ġat</w>": 514,
+  "Ġatt": 788,
+  "Ġb": 371,
+  "Ġbe": 499,
+  "Ġbe</w>": 595,
+  "Ġbec": 706,
+  "Ġbeen</w>": 686,
+  "Ġbeg": 915,
+  "Ġbetween</w>": 914,
+  "Ġbo": 819,
+  "Ġbut</w>": 623,
+  "Ġby</w>": 473,
+  "Ġc": 368,
+  "Ġcent": 823,
+  "Ġch": 526,
+  "Ġchar": 822,
+  "Ġcl": 689,
+  "Ġcom": 509,
+  "Ġcomm": 707,
+  "Ġcomp": 616,
+  "Ġcon": 477,
+  "Ġcons": 841,
+  "Ġcont": 655,
+  "Ġcre": 931,
+  "Ġd": 387,
+  "Ġde": 627,
+  "Ġdec": 873,
+  "Ġdef": 965,
+  "Ġdes": 738,
+  "Ġdi": 892,
+  "Ġdis": 708,
+  "Ġduring</w>": 864,
+  "Ġe": 421,
+  "Ġear": 854,
+  "Ġel": 869,
+  "Ġen": 662,
+  "Ġev": 682,
+  "Ġex": 539,
+  "Ġexp": 867,
+  "Ġf": 370,
+  "Ġfe": 845,
+  "Ġfil": 913,
+  "Ġfin": 786,
+  "Ġfir": 599,
+  "Ġfirst</w>": 626,
+  "Ġfl": 877,
+  "Ġfor": 614,
+  "Ġfor</w>": 458,
+  "Ġform": 963,
+  "Ġfrom</w>": 503,
+  "Ġg": 430,
+  "Ġgame</w>": 895,
+  "Ġgen": 985,
+  "Ġgro": 890,
+  "Ġh": 380,
+  "Ġha": 485,
+  "Ġhad</w>": 566,
+  "Ġhas</w>": 679,
+  "Ġhave</w>": 667,
+  "Ġhe</w>": 558,
+  "Ġher</w>": 660,
+  "Ġhim</w>": 896,
+  "Ġhis</w>": 512,
+  "Ġi": 366,
+  "Ġimp": 909,
+  "Ġin": 429,
+  "Ġin</w>": 389,
+  "Ġinc": 618,
+  "Ġinclud": 761,
+  "Ġind": 983,
+  "Ġint": 628,
+  "Ġinter": 832,
+  "Ġinto</w>": 785,
+  "Ġis</w>": 480,
+  "Ġit</w>": 533,
+  "Ġits</w>": 642,
+  "Ġj": 723,
+  "Ġk": 564,
+  "Ġkn": 857,
+  "Ġl": 398,
+  "Ġlar": 962,
+  "Ġlater</w>": 936,
+  "Ġlea": 868,
+  "Ġm": 386,
+  "Ġmade</w>": 949,
+  "Ġme": 968,
+  "Ġmore</w>": 802,
+  "Ġmost</w>": 910,
+  "Ġmov": 956,
+  "Ġmus": 966,
+  "Ġn": 415,
+  "Ġnew</w>": 989,
+  "Ġno": 547,
+  "Ġnor": 978,
+  "Ġnot</w>": 632,
+  "Ġnum": 926,
+  "Ġo": 359,
+  "Ġof</w>": 373,
+  "Ġoff": 875,
+  "Ġon": 551,
+  "Ġon</w>": 456,
+  "Ġone</w>": 677,
+  "Ġonly</w>": 871,
+  "Ġor": 699,
+  "Ġor</w>": 645,
+  "Ġother</w>": 787,
+  "Ġout</w>": 925,
+  "Ġov": 729,
+  "Ġover</w>": 856,
+  "Ġp": 379,
+  "Ġpar": 636,
+  "Ġper": 736,
+  "Ġpl": 698,
+  "Ġpla": 697,
+  "Ġplay": 808,
+  "Ġpos": 859,
+  "Ġpr": 889,
+  "Ġpre": 749,
+  "Ġpres": 912,
+  "Ġpro": 506,
+  "Ġproduc": 934,
+  "Ġqu": 955,
+  "Ġr": 521,
+  "Ġra": 863,
+  "Ġre": 400,
+  "Ġrec": 597,
+  "Ġrecor": 919,
+  "Ġreg": 943,
+  "Ġrel": 900,
+  "Ġrele": 946,
+  "Ġrem": 848,
+  "Ġrep": 762,
+  "Ġres": 650,
+  "Ġro": 629,
+  "Ġs": 361,
+  "Ġsa": 905,
+  "Ġsc": 732,
+  "Ġse": 569,
+  "Ġseason</w>": 948,
+  "Ġsec": 781,
+  "Ġser": 740,
+  "Ġsev": 884,
+  "Ġsh": 513,
+  "Ġshe</w>": 862,
+  "Ġsp": 578,
+  "Ġspec": 940,
+  "Ġst": 446,
+  "Ġstar": 939,
+  "Ġsu": 770,
+  "Ġsub": 969,
+  "Ġsuc": 764,
+  "Ġsuch</w>": 981,
+  "Ġt": 354,
+  "Ġth": 355,
+  "Ġthan</w>": 918,
+  "Ġthat</w>": 452,
+  "Ġthe": 502,
+  "Ġthe</w>": 357,
+  "Ġtheir</w>": 621,
+  "Ġthem</w>": 998,
+  "Ġthey</w>": 727,
+  "Ġthis</w>": 705,
+  "Ġthree</w>": 902,
+  "Ġthroug": 923,
+  "Ġtime</w>": 783,
+  "Ġto</w>": 391,
+  "Ġtra": 836,
+  "Ġtw": 639,
+  "Ġtwo</w>": 688,
+  "Ġun": 544,
+  "Ġup</w>": 898,
+  "Ġus": 668,
+  "Ġused</w>": 988,
+  "Ġv": 495,
+  "Ġw": 363,
+  "Ġwas</w>": 422,
+  "Ġwere</w>": 525,
+  "Ġwh": 443,
+  "Ġwhen</w>": 851,
+  "Ġwhere</w>": 995,
+  "Ġwhich</w>": 573,
+  "Ġwhile</w>": 935,
+  "Ġwho</w>": 724,
+  "Ġwit": 451,
+  "Ġwith": 994,
+  "Ġwith</w>": 466,
+  "Ġwor": 643,
+  "Ġwould</w>": 801,
+  "Ġwrit": 937,
+  "Ġy": 580,
+  "Ġyear</w>": 987,
+  "Ġâ": 556,
+  "ĠâĢ": 565,
+  "ĠâĢĵ</w>": 631,
+  "ĠĊ</w>": 412,
+  "Ģ": 161,
+  "Ģ</w>": 223,
+  "ģ": 162,
+  "ģ</w>": 273,
+  "Ĥ": 163,
+  "Ĥ</w>": 262,
+  "ĥ": 164,
+  "ĥ</w>": 337,
+  "Ħ": 165,
+  "Ħ</w>": 278,
+  "ħ": 166,
+  "ħ</w>": 281,
+  "Ĩ": 167,
+  "Ĩ</w>": 308,
+  "ĩ": 168,
+  "ĩ</w>": 225,
+  "Ī": 169,
+  "Ī</w>": 221,
+  "ī": 170,
+  "ī</w>": 244,
+  "Ĭ": 171,
+  "Ĭ</w>": 315,
+  "ĭ": 172,
+  "ĭ</w>": 321,
+  "Į": 173,
+  "Į</w>": 324,
+  "į": 174,
+  "į</w>": 302,
+  "İ": 175,
+  "İ</w>": 249,
+  "ı": 176,
+  "ı</w>": 332,
+  "IJ": 177,
+  "IJ</w>": 295,
+  "ij": 178,
+  "ij</w>": 313,
+  "Ĵ": 179,
+  "Ĵ</w>": 328,
+  "ĵ": 180,
+  "ĵ</w>": 312,
+  "Ķ": 181,
+  "Ķ</w>": 256,
+  "ķ": 182,
+  "ķ</w>": 314,
+  "ĸ": 183,
+  "ĸ</w>": 277,
+  "Ĺ": 184,
+  "Ĺ</w>": 322,
+  "ĺ": 185,
+  "ĺ</w>": 285,
+  "Ļ": 186,
+  "Ļ</w>": 267,
+  "ļ": 187,
+  "ļ</w>": 290,
+  "Ľ": 188,
+  "Ľ</w>": 311,
+  "ľ": 189,
+  "ľ</w>": 299,
+  "Ŀ": 190,
+  "Ŀ</w>": 291,
+  "ŀ": 191,
+  "ŀ</w>": 293,
+  "Ł": 192,
+  "Ł</w>": 335,
+  "ł": 193,
+  "ł</w>": 252,
+  "Ń": 194,
+  "Ń</w>": 297
+}

text_encoder/config.json

@@ -0,0 +1,23 @@
+{
+  "architectures": [
+    "CLIPTextModel"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "dropout": 0.0,
+  "eos_token_id": 2,
+  "hidden_act": "quick_gelu",
+  "hidden_size": 32,
+  "initializer_factor": 1.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 37,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 77,
+  "model_type": "clip_text_model",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 5,
+  "pad_token_id": 1,
+  "torch_dtype": "float32",
+  "transformers_version": "4.23.1",
+  "vocab_size": 1000
+}

text_encoder/merges.txt

@@ -0,0 +1,647 @@
+#version: 0.2
+Ġ t
+Ġt h
+Ġ a
+Ġth e</w>
+i n
+Ġ o
+Ġ ,</w>
+Ġ s
+e d</w>
+Ġ w
+e r
+Ġ .</w>
+Ġ i
+r e
+Ġ c
+n d</w>
+Ġ f
+Ġ b
+a t
+Ġo f</w>
+e r</w>
+e n
+a r
+o r
+i t
+Ġ p
+Ġ h
+Ġa nd</w>
+o n
+in g</w>
+a n
+r o
+Ġ m
+Ġ d
+e s</w>
+Ġi n</w>
+o n</w>
+Ġt o</w>
+o u
+i s
+Ġ a</w>
+i c
+Ġ T
+a l
+Ġ l
+Ġ =</w>
+Ġ re
+Ġ "</w>
+e s
+Ġ S
+a s</w>
+a l</w>
+i l
+e l
+i on</w>
+Ġ A
+Ġ C
+Ġ 1
+Ġ Ċ</w>
+u r
+ĠT h
+Ġ n
+a s
+Ġ @
+e c
+o m
+a c
+Ġ e
+Ġw as</w>
+Ġ M
+o r</w>
+a n</w>
+a m
+e n</w>
+o l
+Ġ in
+Ġ g
+Ġ '</w>
+Ġ B
+l y</w>
+a t</w>
+i v
+t s</w>
+ĠTh e</w>
+u s
+- @</w>
+Ġ@ -@</w>
+i s</w>
+Ġ I
+Ġw h
+i g
+Ġ H
+Ġs t
+o s
+u n
+t h
+Ġ P
+Ġw it
+Ġth at</w>
+i r
+Ġa s</w>
+e m
+Ġo n</w>
+r a
+Ġf or</w>
+Ġ R
+e t
+o w
+Ġ 2
+i d
+Ġ D
+l e</w>
+Ġwit h</w>
+l a
+en t</w>
+i m
+Ġ F
+e a
+i on
+Ġb y</w>
+Ġ )</w>
+Ġ (</w>
+Ġa l
+Ġc on
+en t
+Ġ W
+Ġi s</w>
+er e</w>
+Ġ G
+Ġ N
+Ġ L
+Ġh a
+er s</w>
+r i
+t h</w>
+t ed</w>
+u c
+Ġ J
+Ġ1 9
+e v
+u l
+Ġ v
+c e</w>
+at ion</w>
+ro m</w>
+Ġb e
+Ġ E
+i n</w>
+Ġth e
+Ġf rom</w>
+Ġ O
+t er</w>
+Ġp ro
+Ġa r
+a d
+Ġc om
+i c</w>
+a g
+Ġh is</w>
+Ġs h
+Ġa t</w>
+o v
+i es</w>
+o o
+p p
+s t
+c h
+Ġ r
+Ġ2 0
+a y</w>
+i f
+Ġw ere</w>
+Ġc h
+u t</w>
+s t</w>
+u t
+d s</w>
+o p
+u m
+Ġi t</w>
+o c
+t er
+l e
+ig h
+u d
+Ġe x
+ion s</w>
+at e</w>
+it y</w>
+at ed</w>
+Ġ un
+e p
+q u
+Ġn o
+Ġ K
+iv e</w>
+is t
+Ġo n
+am e</w>
+ou n
+i r</w>
+a b
+Ġ â
+in g
+Ġh e</w>
+l d</w>
+u g
+ic h</w>
+Ġa n</w>
+e d
+Ġ k
+Ġâ Ģ
+Ġha d</w>
+v e</w>
+a in
+Ġs e
+t ion</w>
+or e</w>
+re s
+Ġwh ich</w>
+ĠI n</w>
+o d
+th er</w>
+a k
+Ġs p
+a r</w>
+Ġ y
+ĠC h
+on g</w>
+Ġa c
+es t</w>
+Ġ U
+a p
+f f
+al ly</w>
+r it
+ĠS t
+u b
+g e</w>
+b er</w>
+e t</w>
+Ġb e</w>
+e ar
+Ġre c
+er s
+Ġf ir
+o t
+Ġar e</w>
+Ġa n
+c h</w>
+o g
+i a</w>
+es t
+in e</w>
+il l
+an d
+e l</w>
+ar y</w>
+e w</w>
+i d</w>
+Ġf or
+Ġ ;</w>
+Ġcom p
+Ġ V
+Ġin c
+t r
+Ġ20 0
+Ġthe ir</w>
+u s</w>
+Ġb ut</w>
+r an
+ic al</w>
+Ġfir st</w>
+Ġd e
+Ġin t
+Ġ ro
+s o</w>
+ĠâĢ ĵ</w>
+Ġno t</w>
+d ing</w>
+f ter</w>
+ur e</w>
+Ġp ar
+Ġ :</w>
+i an</w>
+Ġt w
+ou ld</w>
+Ġal so</w>
+Ġi ts</w>
+Ġw or
+u m</w>
+Ġo r</w>
+os t</w>
+0 0</w>
+ou r
+ar d</w>
+Ġre s
+m p
+u e</w>
+Ġa b
+is h</w>
+Ġcon t
+Ġa d
+ow n</w>
+al l</w>
+ou g
+Ġh er</w>
+as t</w>
+Ġ en
+om e</w>
+al l
+d ed</w>
+o w</w>
+Ġha ve</w>
+Ġ us
+ea r</w>
+ac k</w>
+d uc
+i al</w>
+s s
+en ts</w>
+a in</w>
+t ing</w>
+Ġon e</w>
+es s
+Ġh as</w>
+igh t</w>
+a v
+Ġe v
+ou t</w>
+a y
+en ce</w>
+Ġbe en</w>
+e w
+Ġtw o</w>
+Ġc l
+d er</w>
+im e</w>
+k s</w>
+es s</w>
+is h
+. @</w>
+Ġ@ .@</w>
+Ġp la
+Ġp l
+Ġo r
+u p</w>
+m ent</w>
+ur ing</w>
+ol l
+ĠI n
+Ġth is</w>
+Ġb ec
+Ġcom m
+Ġd is
+at er</w>
+ag e</w>
+Ġa pp
+ou s</w>
+e y</w>
+i l</w>
+p er
+ĠA l
+ion al</w>
+l ud
+el y</w>
+t t
+il e</w>
+i z
+Ġ j
+Ġwh o</w>
+Ġa g
+i b
+Ġthe y</w>
+f or
+Ġo v
+at h
+e g
+Ġs c
+i p
+Ġ20 1
+Ġ 3
+Ġp er
+or y</w>
+Ġd es
+id e</w>
+Ġs er
+s e</w>
+ĠH e</w>
+la nd</w>
+at ions</w>
+r ic
+i t</w>
+re s</w>
+er ed</w>
+Ġp re
+ĠS h
+an ce</w>
+or t</w>
+an t</w>
+, @</w>
+Ġ@ ,@</w>
+el l</w>
+Ġ Y
+n ed</w>
+el l
+it e</w>
+Ġinc lud
+Ġre p
+Ġa fter</w>
+Ġs uc
+re e</w>
+an y</w>
+i m</w>
+or t
+Ġ1 8
+Ġs u
+ad e</w>
+ou r</w>
+ĠU n
+ĠI t</w>
+i k
+ĠM ar
+em ber</w>
+Ġ 1</w>
+e en</w>
+a nd</w>
+Ġs ec
+ic e</w>
+Ġt ime</w>
+ĠA n
+Ġint o</w>
+Ġf in
+Ġo ther</w>
+Ġa tt
+il l</w>
+re n
+ac h
+as s
+er al</w>
+es e</w>
+s h
+al s</w>
+it ion</w>
+oug h</w>
+l es</w>
+am p
+Ġw ould</w>
+Ġm ore</w>
+ro ug
+ri b
+er y</w>
+ac e</w>
+Ġ A</w>
+Ġpla y
+it ed</w>
+k ed</w>
+is t</w>
+i ed</w>
+Ġ 2</w>
+as ed</w>
+ing s</w>
+an g
+a m</w>
+i p</w>
+Ġb o
+ab le</w>
+t y</w>
+Ġch ar
+Ġc ent
+et w
+at es</w>
+ro p
+Ġ I</w>
+u nd</w>
+ĠA m
+c es</w>
+o in
+Ġin ter
+u p
+c t
+on e</w>
+Ġt ra
+an t
+ec t
+Ġal l</w>
+e f
+Ġcon s
+ub l
+n ing</w>
+an s</w>
+Ġf e
+us t</w>
+Ġ 0
+Ġre m
+as e</w>
+on g
+Ġwh en</w>
+e b
+ĠW h
+Ġe ar
+ev er</w>
+Ġov er</w>
+Ġk n
+a us
+Ġp os
+a d</w>
+er m
+Ġsh e</w>
+Ġ ra
+Ġd uring</w>
+as on</w>
+v i
+Ġex p
+Ġl ea
+Ġ el
+Ġ 4
+Ġon ly</w>
+o nd</w>
+Ġd ec
+Ġac c
+Ġo ff
+is s
+Ġf l
+ĠE n
+o t</w>
+en s
+os e</w>
+ak e</w>
+o m</w>
+Ġs ev
+ac h</w>
+etw een</w>
+er n
+Ġ 3</w>
+Ġp r
+Ġg ro
+r uc
+Ġd i
+Ġ19 9
+ĠA r
+Ġg ame</w>
+Ġh im</w>
+oo k</w>
+Ġ up</w>
+Ġab out</w>
+Ġre l
+for m
+Ġth ree</w>
+at t
+ĠC om
+Ġs a
+ear s</w>
+Ġ 5
+r y</w>
+Ġi mp
+Ġm ost</w>
+f er
+Ġp res
+Ġf il
+Ġb etween</w>
+Ġbe g
+p h
+or s</w>
+Ġth an</w>
+Ġrec or
+o b
+er ic
+at ing</w>
+Ġth roug
+k ing</w>
+Ġo ut</w>
+Ġn um
+oo d</w>
+oll ow
+ac t
+u il
+Ġc re
+ol og
+at ional</w>
+Ġpro duc
+Ġwh ile</w>
+Ġl ater</w>
+Ġw rit
+e x
+Ġst ar
+Ġsp ec
+e e
+ish ed</w>
+Ġre g
+is ion</w>
+ou th</w>
+Ġre le
+Ġa ss
+Ġse ason</w>
+Ġm ade</w>
+il y</w>
+r u
+o y
+t ur
+t e</w>
+Ġ qu
+Ġm ov
+ur y</w>
+ĠAm eric
+em ent</w>
+c c
+ou nd</w>
+Ġl ar
+Ġfor m
+ec t</w>
+Ġde f
+Ġm us
+ĠP ar
+Ġm e
+Ġs ub
+w ay</w>
+o p</w>
+o h
+el d</w>
+i e</w>
+em p
+am es</w>
+er n</w>
+Ġn or
+iv ed</w>
+ev el
+Ġsuc h</w>
+ar ds</w>
+Ġin d
+ik e</w>
+Ġg en
+er t
+Ġy ear</w>
+Ġus ed</w>
+Ġn ew</w>
+Ġ 5</w>
+Ġal b
+s p
+y p
+Ġwit h
+Ġwh ere</w>
+ic s</w>
+ĠTh is</w>
+Ġthe m</w>
+w n</w>

text_encoder/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8db031dc7a69f514ab2b725e1653abb62f13146b92fd9a1c0a6258b63a4d71eb
+size 301680

text_encoder/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

text_encoder/tokenizer_config.json

@@ -0,0 +1,34 @@
+{
+  "add_prefix_space": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "do_lower_case": true,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "errors": "replace",
+  "model_max_length": 77,
+  "name_or_path": "hf-internal-testing/tiny-random-clip",
+  "pad_token": "<|endoftext|>",
+  "special_tokens_map_file": "/home/lysandre/.cache/huggingface/transformers/18a566598f286c9139f88160c99f84eec492a26bd22738fa9cb44d5b7e0a5c76.cce1206abbad28826f000510f22f354e53e66a97f7c23745a7dfe27609cc07f5",
+  "tokenizer_class": "CLIPTokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

text_encoder/vocab.json

@@ -0,0 +1,1002 @@
+{
+  "!": 2,
+  "!</w>": 345,
+  "\"": 3,
+  "\"</w>": 344,
+  "#": 4,
+  "#</w>": 325,
+  "$": 5,
+  "$</w>": 348,
+  "%": 6,
+  "%</w>": 351,
+  "&": 7,
+  "&</w>": 352,
+  "'": 8,
+  "'</w>": 296,
+  "(": 9,
+  "(</w>": 318,
+  ")": 10,
+  ")</w>": 330,
+  "*": 11,
+  "*</w>": 327,
+  "+": 12,
+  "+</w>": 341,
+  ",": 13,
+  ",</w>": 279,
+  ",@</w>": 754,
+  "-": 14,
+  "-</w>": 276,
+  "-@</w>": 439,
+  ".": 15,
+  ".</w>": 253,
+  ".@</w>": 695,
+  "/": 16,
+  "/</w>": 350,
+  "0": 17,
+  "00</w>": 647,
+  "0</w>": 216,
+  "1": 18,
+  "1</w>": 222,
+  "2": 19,
+  "2</w>": 231,
+  "3": 20,
+  "3</w>": 243,
+  "4": 21,
+  "4</w>": 233,
+  "5": 22,
+  "5</w>": 240,
+  "6": 23,
+  "6</w>": 226,
+  "7": 24,
+  "7</w>": 215,
+  "8": 25,
+  "8</w>": 236,
+  "9": 26,
+  "9</w>": 242,
+  ":": 27,
+  ":</w>": 353,
+  ";": 28,
+  ";</w>": 317,
+  "<": 29,
+  "<</w>": 340,
+  "<|endoftext|>": 1,
+  "<|startoftext|>": 0,
+  "=": 30,
+  "=</w>": 342,
+  ">": 31,
+  "></w>": 300,
+  "?": 32,
+  "?</w>": 346,
+  "@": 33,
+  "@</w>": 320,
+  "A": 34,
+  "A</w>": 227,
+  "B": 35,
+  "B</w>": 258,
+  "C": 36,
+  "C</w>": 239,
+  "D": 37,
+  "D</w>": 255,
+  "E": 38,
+  "E</w>": 246,
+  "F": 39,
+  "F</w>": 213,
+  "G": 40,
+  "G</w>": 283,
+  "H": 41,
+  "H</w>": 219,
+  "I": 42,
+  "I</w>": 237,
+  "J": 43,
+  "J</w>": 251,
+  "K": 44,
+  "K</w>": 254,
+  "L": 45,
+  "L</w>": 218,
+  "M": 46,
+  "M</w>": 234,
+  "N": 47,
+  "N</w>": 238,
+  "O": 48,
+  "O</w>": 265,
+  "P": 49,
+  "P</w>": 245,
+  "Q": 50,
+  "Q</w>": 309,
+  "R": 51,
+  "R</w>": 264,
+  "S": 52,
+  "S</w>": 230,
+  "T": 53,
+  "T</w>": 235,
+  "U": 54,
+  "U</w>": 268,
+  "V": 55,
+  "V</w>": 248,
+  "W": 56,
+  "W</w>": 274,
+  "X": 57,
+  "X</w>": 263,
+  "Y": 58,
+  "Y</w>": 310,
+  "Z": 59,
+  "Z</w>": 207,
+  "[": 60,
+  "[</w>": 270,
+  "\\": 61,
+  "\\</w>": 338,
+  "]": 62,
+  "]</w>": 289,
+  "^": 63,
+  "^</w>": 331,
+  "_": 64,
+  "_</w>": 334,
+  "`": 65,
+  "`</w>": 347,
+  "a": 66,
+  "a</w>": 197,
+  "ab": 555,
+  "able</w>": 820,
+  "ac": 420,
+  "ace</w>": 806,
+  "ach": 791,
+  "ach</w>": 885,
+  "ack</w>": 670,
+  "act": 929,
+  "ad": 508,
+  "ad</w>": 860,
+  "ade</w>": 771,
+  "ag": 511,
+  "age</w>": 710,
+  "ain": 568,
+  "ain</w>": 675,
+  "ak": 577,
+  "ake</w>": 882,
+  "al": 397,
+  "al</w>": 405,
+  "all": 664,
+  "all</w>": 658,
+  "ally</w>": 588,
+  "als</w>": 796,
+  "am": 426,
+  "am</w>": 817,
+  "ame</w>": 552,
+  "ames</w>": 976,
+  "amp": 800,
+  "an": 384,
+  "an</w>": 425,
+  "ance</w>": 751,
+  "and": 609,
+  "and</w>": 780,
+  "ang": 816,
+  "ans</w>": 844,
+  "ant": 837,
+  "ant</w>": 753,
+  "any</w>": 766,
+  "ap": 586,
+  "ar": 376,
+  "ar</w>": 579,
+  "ard</w>": 649,
+  "ards</w>": 982,
+  "ary</w>": 611,
+  "as": 416,
+  "as</w>": 404,
+  "ase</w>": 849,
+  "ased</w>": 814,
+  "ason</w>": 865,
+  "ass": 792,
+  "ast</w>": 661,
+  "at": 372,
+  "at</w>": 434,
+  "ate</w>": 541,
+  "ated</w>": 543,
+  "ater</w>": 709,
+  "ates</w>": 825,
+  "ath": 730,
+  "ating</w>": 922,
+  "ation</w>": 497,
+  "ational</w>": 933,
+  "ations</w>": 744,
+  "att": 903,
+  "aus": 858,
+  "av": 681,
+  "ay": 684,
+  "ay</w>": 523,
+  "b": 67,
+  "b</w>": 212,
+  "ber</w>": 593,
+  "c": 68,
+  "c</w>": 224,
+  "cc": 960,
+  "ce</w>": 496,
+  "ces</w>": 830,
+  "ch": 520,
+  "ch</w>": 603,
+  "ct": 834,
+  "d": 69,
+  "d</w>": 196,
+  "ded</w>": 665,
+  "der</w>": 690,
+  "ding</w>": 633,
+  "ds</w>": 530,
+  "duc": 671,
+  "e": 70,
+  "e</w>": 195,
+  "ea": 471,
+  "ear": 596,
+  "ear</w>": 669,
+  "ears</w>": 906,
+  "eb": 852,
+  "ec": 418,
+  "ect": 838,
+  "ect</w>": 964,
+  "ed": 563,
+  "ed</w>": 362,
+  "ee": 941,
+  "een</w>": 779,
+  "ef": 840,
+  "eg": 731,
+  "el": 407,
+  "el</w>": 610,
+  "eld</w>": 973,
+  "ell": 759,
+  "ell</w>": 756,
+  "ely</w>": 719,
+  "em": 455,
+  "ember</w>": 777,
+  "ement</w>": 959,
+  "emp": 975,
+  "en": 375,
+  "en</w>": 427,
+  "ence</w>": 685,
+  "ens": 880,
+  "ent": 478,
+  "ent</w>": 468,
+  "ents</w>": 674,
+  "ep": 545,
+  "er": 364,
+  "er</w>": 374,
+  "eral</w>": 793,
+  "ere</w>": 481,
+  "ered</w>": 748,
+  "eric": 921,
+  "erm": 861,
+  "ern": 887,
+  "ern</w>": 977,
+  "ers": 598,
+  "ers</w>": 486,
+  "ert": 986,
+  "ery</w>": 805,
+  "es": 402,
+  "es</w>": 388,
+  "ese</w>": 794,
+  "ess": 678,
+  "ess</w>": 693,
+  "est": 606,
+  "est</w>": 584,
+  "et": 460,
+  "et</w>": 594,
+  "etw": 824,
+  "etween</w>": 886,
+  "ev": 493,
+  "evel": 980,
+  "ever</w>": 855,
+  "ew": 687,
+  "ew</w>": 612,
+  "ex": 938,
+  "ey</w>": 713,
+  "f": 71,
+  "f</w>": 209,
+  "fer": 911,
+  "ff": 587,
+  "for": 728,
+  "form": 901,
+  "fter</w>": 634,
+  "g": 72,
+  "g</w>": 214,
+  "ge</w>": 592,
+  "h": 73,
+  "h</w>": 203,
+  "i": 74,
+  "i</w>": 205,
+  "ia</w>": 605,
+  "ial</w>": 672,
+  "ian</w>": 638,
+  "ib": 726,
+  "ic": 395,
+  "ic</w>": 510,
+  "ical</w>": 625,
+  "ice</w>": 782,
+  "ich</w>": 561,
+  "ics</w>": 996,
+  "id": 463,
+  "id</w>": 613,
+  "ide</w>": 739,
+  "ie</w>": 974,
+  "ied</w>": 812,
+  "ies</w>": 516,
+  "if": 524,
+  "ig": 444,
+  "igh": 537,
+  "ight</w>": 680,
+  "ik": 775,
+  "ike</w>": 984,
+  "il": 406,
+  "il</w>": 714,
+  "ile</w>": 721,
+  "ill": 608,
+  "ill</w>": 789,
+  "ily</w>": 950,
+  "im": 469,
+  "im</w>": 767,
+  "ime</w>": 691,
+  "in": 358,
+  "in</w>": 501,
+  "ine</w>": 607,
+  "ing": 557,
+  "ing</w>": 383,
+  "ings</w>": 815,
+  "ion": 472,
+  "ion</w>": 408,
+  "ional</w>": 717,
+  "ions</w>": 540,
+  "ip": 733,
+  "ip</w>": 818,
+  "ir": 453,
+  "ir</w>": 554,
+  "is": 393,
+  "is</w>": 441,
+  "ish": 694,
+  "ish</w>": 654,
+  "ished</w>": 942,
+  "ision</w>": 944,
+  "iss": 876,
+  "ist": 550,
+  "ist</w>": 811,
+  "it": 378,
+  "it</w>": 746,
+  "ite</w>": 760,
+  "ited</w>": 809,
+  "ition</w>": 797,
+  "ity</w>": 542,
+  "iv": 435,
+  "ive</w>": 549,
+  "ived</w>": 979,
+  "iz": 722,
+  "j": 75,
+  "j</w>": 288,
+  "k": 76,
+  "k</w>": 210,
+  "ked</w>": 810,
+  "king</w>": 924,
+  "ks</w>": 692,
+  "l": 77,
+  "l</w>": 201,
+  "la": 467,
+  "land</w>": 743,
+  "ld</w>": 559,
+  "le": 536,
+  "le</w>": 465,
+  "les</w>": 799,
+  "lud": 718,
+  "ly</w>": 433,
+  "m": 78,
+  "m</w>": 202,
+  "ment</w>": 701,
+  "mp": 651,
+  "n": 79,
+  "n</w>": 199,
+  "nd</w>": 369,
+  "ned</w>": 758,
+  "ning</w>": 843,
+  "o": 80,
+  "o</w>": 198,
+  "ob": 920,
+  "oc": 534,
+  "od": 575,
+  "og": 604,
+  "oh": 972,
+  "oin": 831,
+  "ol": 428,
+  "oll": 703,
+  "ollow": 928,
+  "olog": 932,
+  "om": 419,
+  "om</w>": 883,
+  "ome</w>": 663,
+  "on": 382,
+  "on</w>": 390,
+  "ond</w>": 872,
+  "one</w>": 835,
+  "ong": 850,
+  "ong</w>": 582,
+  "oo": 517,
+  "ood</w>": 927,
+  "ook</w>": 897,
+  "op": 531,
+  "op</w>": 971,
+  "or": 377,
+  "or</w>": 424,
+  "ore</w>": 571,
+  "ors</w>": 917,
+  "ort": 768,
+  "ort</w>": 752,
+  "ory</w>": 737,
+  "os": 447,
+  "ose</w>": 881,
+  "ost</w>": 646,
+  "ot": 600,
+  "ot</w>": 879,
+  "ou": 392,
+  "oug": 659,
+  "ough</w>": 798,
+  "ould</w>": 640,
+  "oun": 553,
+  "ound</w>": 961,
+  "our": 648,
+  "our</w>": 772,
+  "ous</w>": 712,
+  "out</w>": 683,
+  "outh</w>": 945,
+  "ov": 515,
+  "ow": 461,
+  "ow</w>": 666,
+  "own</w>": 657,
+  "oy": 952,
+  "p": 81,
+  "p</w>": 217,
+  "per": 715,
+  "ph": 916,
+  "pp": 518,
+  "q": 82,
+  "q</w>": 280,
+  "qu": 546,
+  "r": 83,
+  "r</w>": 204,
+  "ra": 457,
+  "ran": 624,
+  "re": 367,
+  "ree</w>": 765,
+  "ren": 790,
+  "res": 572,
+  "res</w>": 747,
+  "ri": 487,
+  "rib": 804,
+  "ric": 745,
+  "rit": 589,
+  "ro": 385,
+  "rom</w>": 498,
+  "rop": 826,
+  "roug": 803,
+  "ru": 951,
+  "ruc": 891,
+  "ry</w>": 908,
+  "s": 84,
+  "s</w>": 206,
+  "se</w>": 741,
+  "sh": 795,
+  "so</w>": 630,
+  "sp": 992,
+  "ss": 673,
+  "st": 519,
+  "st</w>": 528,
+  "t": 85,
+  "t</w>": 208,
+  "te</w>": 954,
+  "ted</w>": 489,
+  "ter": 535,
+  "ter</w>": 505,
+  "th": 449,
+  "th</w>": 488,
+  "ther</w>": 576,
+  "ting</w>": 676,
+  "tion</w>": 570,
+  "tr": 619,
+  "ts</w>": 436,
+  "tt": 720,
+  "tur": 953,
+  "ty</w>": 821,
+  "u": 86,
+  "u</w>": 229,
+  "ub": 591,
+  "ubl": 842,
+  "uc": 490,
+  "ud": 538,
+  "ue</w>": 652,
+  "ug": 560,
+  "uil": 930,
+  "ul": 494,
+  "um": 532,
+  "um</w>": 644,
+  "un": 448,
+  "und</w>": 828,
+  "up": 833,
+  "up</w>": 700,
+  "ur": 413,
+  "ure</w>": 635,
+  "uring</w>": 702,
+  "ury</w>": 957,
+  "us": 438,
+  "us</w>": 622,
+  "ust</w>": 846,
+  "ut": 529,
+  "ut</w>": 527,
+  "v": 87,
+  "v</w>": 232,
+  "ve</w>": 567,
+  "vi": 866,
+  "w": 88,
+  "w</w>": 250,
+  "way</w>": 970,
+  "wn</w>": 999,
+  "x": 89,
+  "x</w>": 269,
+  "y": 90,
+  "y</w>": 211,
+  "yp": 993,
+  "z": 91,
+  "z</w>": 228,
+  "|": 92,
+  "|</w>": 304,
+  "}": 93,
+  "}</w>": 336,
+  "~": 94,
+  "~</w>": 343,
+  "¡": 95,
+  "¡</w>": 220,
+  "¢": 96,
+  "¢</w>": 306,
+  "£": 97,
+  "£</w>": 323,
+  "¤": 98,
+  "¤</w>": 292,
+  "¥": 99,
+  "¥</w>": 339,
+  "¦": 100,
+  "¦</w>": 303,
+  "§": 101,
+  "§</w>": 275,
+  "¨": 102,
+  "¨</w>": 282,
+  "©": 103,
+  "©</w>": 259,
+  "ª": 104,
+  "ª</w>": 286,
+  "«": 105,
+  "«</w>": 266,
+  "¬": 106,
+  "¬</w>": 319,
+  "®": 107,
+  "®</w>": 329,
+  "¯": 108,
+  "¯</w>": 287,
+  "°": 109,
+  "°</w>": 298,
+  "±": 110,
+  "±</w>": 200,
+  "²": 111,
+  "²</w>": 284,
+  "³": 112,
+  "³</w>": 272,
+  "´": 113,
+  "´</w>": 307,
+  "µ": 114,
+  "µ</w>": 261,
+  "¶": 115,
+  "¶</w>": 301,
+  "·": 116,
+  "·</w>": 326,
+  "¸": 117,
+  "¸</w>": 257,
+  "¹": 118,
+  "¹</w>": 241,
+  "º": 119,
+  "º</w>": 260,
+  "»": 120,
+  "»</w>": 247,
+  "¼": 121,
+  "¼</w>": 305,
+  "½": 122,
+  "½</w>": 294,
+  "¾": 123,
+  "¾</w>": 316,
+  "¿": 124,
+  "¿</w>": 271,
+  "Â": 125,
+  "Ã": 126,
+  "Ä": 127,
+  "Å": 128,
+  "Æ": 129,
+  "Ç": 130,
+  "È": 131,
+  "É": 132,
+  "Ê": 133,
+  "Ë": 134,
+  "Ì": 135,
+  "Í": 136,
+  "Î": 137,
+  "Ï": 138,
+  "Ð": 139,
+  "Ñ": 140,
+  "Ö": 141,
+  "×": 142,
+  "Ø": 143,
+  "Ù": 144,
+  "Ü": 145,
+  "à": 146,
+  "á": 147,
+  "â": 148,
+  "ã": 149,
+  "ä": 150,
+  "å": 151,
+  "æ": 152,
+  "ç": 153,
+  "è": 154,
+  "é": 155,
+  "ë": 156,
+  "ì": 157,
+  "ï": 158,
+  "Ċ": 159,
+  "Ċ</w>": 349,
+  "Ġ": 160,
+  "Ġ\"</w>": 401,
+  "Ġ'</w>": 431,
+  "Ġ(</w>": 475,
+  "Ġ)</w>": 474,
+  "Ġ,</w>": 360,
+  "Ġ.</w>": 365,
+  "Ġ0": 847,
+  "Ġ1": 411,
+  "Ġ18": 769,
+  "Ġ19": 492,
+  "Ġ199": 893,
+  "Ġ1</w>": 778,
+  "Ġ2": 462,
+  "Ġ20": 522,
+  "Ġ200": 620,
+  "Ġ201": 734,
+  "Ġ2</w>": 813,
+  "Ġ3": 735,
+  "Ġ3</w>": 888,
+  "Ġ4": 870,
+  "Ġ5": 907,
+  "Ġ5</w>": 990,
+  "Ġ:</w>": 637,
+  "Ġ;</w>": 615,
+  "Ġ</w>": 333,
+  "Ġ=</w>": 399,
+  "Ġ@": 417,
+  "Ġ@,@</w>": 755,
+  "Ġ@-@</w>": 440,
+  "Ġ@.@</w>": 696,
+  "ĠA": 409,
+  "ĠA</w>": 807,
+  "ĠAl": 716,
+  "ĠAm": 829,
+  "ĠAmeric": 958,
+  "ĠAn": 784,
+  "ĠAr": 894,
+  "ĠB": 432,
+  "ĠC": 410,
+  "ĠCh": 581,
+  "ĠCom": 904,
+  "ĠD": 464,
+  "ĠE": 500,
+  "ĠEn": 878,
+  "ĠF": 470,
+  "ĠG": 482,
+  "ĠH": 445,
+  "ĠHe</w>": 742,
+  "ĠI": 442,
+  "ĠI</w>": 827,
+  "ĠIn": 704,
+  "ĠIn</w>": 574,
+  "ĠIt</w>": 774,
+  "ĠJ": 491,
+  "ĠK": 548,
+  "ĠL": 484,
+  "ĠM": 423,
+  "ĠMar": 776,
+  "ĠN": 483,
+  "ĠO": 504,
+  "ĠP": 450,
+  "ĠPar": 967,
+  "ĠR": 459,
+  "ĠS": 403,
+  "ĠSh": 750,
+  "ĠSt": 590,
+  "ĠT": 396,
+  "ĠTh": 414,
+  "ĠThe</w>": 437,
+  "ĠThis</w>": 997,
+  "ĠU": 585,
+  "ĠUn": 773,
+  "ĠV": 617,
+  "ĠW": 479,
+  "ĠWh": 853,
+  "ĠY": 757,
+  "Ġa": 356,
+  "Ġa</w>": 394,
+  "Ġab": 653,
+  "Ġabout</w>": 899,
+  "Ġac": 583,
+  "Ġacc": 874,
+  "Ġad": 656,
+  "Ġafter</w>": 763,
+  "Ġag": 725,
+  "Ġal": 476,
+  "Ġalb": 991,
+  "Ġall</w>": 839,
+  "Ġalso</w>": 641,
+  "Ġan": 602,
+  "Ġan</w>": 562,
+  "Ġand</w>": 381,
+  "Ġapp": 711,
+  "Ġar": 507,
+  "Ġare</w>": 601,
+  "Ġas</w>": 454,
+  "Ġass": 947,
+  "Ġat</w>": 514,
+  "Ġatt": 788,
+  "Ġb": 371,
+  "Ġbe": 499,
+  "Ġbe</w>": 595,
+  "Ġbec": 706,
+  "Ġbeen</w>": 686,
+  "Ġbeg": 915,
+  "Ġbetween</w>": 914,
+  "Ġbo": 819,
+  "Ġbut</w>": 623,
+  "Ġby</w>": 473,
+  "Ġc": 368,
+  "Ġcent": 823,
+  "Ġch": 526,
+  "Ġchar": 822,
+  "Ġcl": 689,
+  "Ġcom": 509,
+  "Ġcomm": 707,
+  "Ġcomp": 616,
+  "Ġcon": 477,
+  "Ġcons": 841,
+  "Ġcont": 655,
+  "Ġcre": 931,
+  "Ġd": 387,
+  "Ġde": 627,
+  "Ġdec": 873,
+  "Ġdef": 965,
+  "Ġdes": 738,
+  "Ġdi": 892,
+  "Ġdis": 708,
+  "Ġduring</w>": 864,
+  "Ġe": 421,
+  "Ġear": 854,
+  "Ġel": 869,
+  "Ġen": 662,
+  "Ġev": 682,
+  "Ġex": 539,
+  "Ġexp": 867,
+  "Ġf": 370,
+  "Ġfe": 845,
+  "Ġfil": 913,
+  "Ġfin": 786,
+  "Ġfir": 599,
+  "Ġfirst</w>": 626,
+  "Ġfl": 877,
+  "Ġfor": 614,
+  "Ġfor</w>": 458,
+  "Ġform": 963,
+  "Ġfrom</w>": 503,
+  "Ġg": 430,
+  "Ġgame</w>": 895,
+  "Ġgen": 985,
+  "Ġgro": 890,
+  "Ġh": 380,
+  "Ġha": 485,
+  "Ġhad</w>": 566,
+  "Ġhas</w>": 679,
+  "Ġhave</w>": 667,
+  "Ġhe</w>": 558,
+  "Ġher</w>": 660,
+  "Ġhim</w>": 896,
+  "Ġhis</w>": 512,
+  "Ġi": 366,
+  "Ġimp": 909,
+  "Ġin": 429,
+  "Ġin</w>": 389,
+  "Ġinc": 618,
+  "Ġinclud": 761,
+  "Ġind": 983,
+  "Ġint": 628,
+  "Ġinter": 832,
+  "Ġinto</w>": 785,
+  "Ġis</w>": 480,
+  "Ġit</w>": 533,
+  "Ġits</w>": 642,
+  "Ġj": 723,
+  "Ġk": 564,
+  "Ġkn": 857,
+  "Ġl": 398,
+  "Ġlar": 962,
+  "Ġlater</w>": 936,
+  "Ġlea": 868,
+  "Ġm": 386,
+  "Ġmade</w>": 949,
+  "Ġme": 968,
+  "Ġmore</w>": 802,
+  "Ġmost</w>": 910,
+  "Ġmov": 956,
+  "Ġmus": 966,
+  "Ġn": 415,
+  "Ġnew</w>": 989,
+  "Ġno": 547,
+  "Ġnor": 978,
+  "Ġnot</w>": 632,
+  "Ġnum": 926,
+  "Ġo": 359,
+  "Ġof</w>": 373,
+  "Ġoff": 875,
+  "Ġon": 551,
+  "Ġon</w>": 456,
+  "Ġone</w>": 677,
+  "Ġonly</w>": 871,
+  "Ġor": 699,
+  "Ġor</w>": 645,
+  "Ġother</w>": 787,
+  "Ġout</w>": 925,
+  "Ġov": 729,
+  "Ġover</w>": 856,
+  "Ġp": 379,
+  "Ġpar": 636,
+  "Ġper": 736,
+  "Ġpl": 698,
+  "Ġpla": 697,
+  "Ġplay": 808,
+  "Ġpos": 859,
+  "Ġpr": 889,
+  "Ġpre": 749,
+  "Ġpres": 912,
+  "Ġpro": 506,
+  "Ġproduc": 934,
+  "Ġqu": 955,
+  "Ġr": 521,
+  "Ġra": 863,
+  "Ġre": 400,
+  "Ġrec": 597,
+  "Ġrecor": 919,
+  "Ġreg": 943,
+  "Ġrel": 900,
+  "Ġrele": 946,
+  "Ġrem": 848,
+  "Ġrep": 762,
+  "Ġres": 650,
+  "Ġro": 629,
+  "Ġs": 361,
+  "Ġsa": 905,
+  "Ġsc": 732,
+  "Ġse": 569,
+  "Ġseason</w>": 948,
+  "Ġsec": 781,
+  "Ġser": 740,
+  "Ġsev": 884,
+  "Ġsh": 513,
+  "Ġshe</w>": 862,
+  "Ġsp": 578,
+  "Ġspec": 940,
+  "Ġst": 446,
+  "Ġstar": 939,
+  "Ġsu": 770,
+  "Ġsub": 969,
+  "Ġsuc": 764,
+  "Ġsuch</w>": 981,
+  "Ġt": 354,
+  "Ġth": 355,
+  "Ġthan</w>": 918,
+  "Ġthat</w>": 452,
+  "Ġthe": 502,
+  "Ġthe</w>": 357,
+  "Ġtheir</w>": 621,
+  "Ġthem</w>": 998,
+  "Ġthey</w>": 727,
+  "Ġthis</w>": 705,
+  "Ġthree</w>": 902,
+  "Ġthroug": 923,
+  "Ġtime</w>": 783,
+  "Ġto</w>": 391,
+  "Ġtra": 836,
+  "Ġtw": 639,
+  "Ġtwo</w>": 688,
+  "Ġun": 544,
+  "Ġup</w>": 898,
+  "Ġus": 668,
+  "Ġused</w>": 988,
+  "Ġv": 495,
+  "Ġw": 363,
+  "Ġwas</w>": 422,
+  "Ġwere</w>": 525,
+  "Ġwh": 443,
+  "Ġwhen</w>": 851,
+  "Ġwhere</w>": 995,
+  "Ġwhich</w>": 573,
+  "Ġwhile</w>": 935,
+  "Ġwho</w>": 724,
+  "Ġwit": 451,
+  "Ġwith": 994,
+  "Ġwith</w>": 466,
+  "Ġwor": 643,
+  "Ġwould</w>": 801,
+  "Ġwrit": 937,
+  "Ġy": 580,
+  "Ġyear</w>": 987,
+  "Ġâ": 556,
+  "ĠâĢ": 565,
+  "ĠâĢĵ</w>": 631,
+  "ĠĊ</w>": 412,
+  "Ģ": 161,
+  "Ģ</w>": 223,
+  "ģ": 162,
+  "ģ</w>": 273,
+  "Ĥ": 163,
+  "Ĥ</w>": 262,
+  "ĥ": 164,
+  "ĥ</w>": 337,
+  "Ħ": 165,
+  "Ħ</w>": 278,
+  "ħ": 166,
+  "ħ</w>": 281,
+  "Ĩ": 167,
+  "Ĩ</w>": 308,
+  "ĩ": 168,
+  "ĩ</w>": 225,
+  "Ī": 169,
+  "Ī</w>": 221,
+  "ī": 170,
+  "ī</w>": 244,
+  "Ĭ": 171,
+  "Ĭ</w>": 315,
+  "ĭ": 172,
+  "ĭ</w>": 321,
+  "Į": 173,
+  "Į</w>": 324,
+  "į": 174,
+  "į</w>": 302,
+  "İ": 175,
+  "İ</w>": 249,
+  "ı": 176,
+  "ı</w>": 332,
+  "IJ": 177,
+  "IJ</w>": 295,
+  "ij": 178,
+  "ij</w>": 313,
+  "Ĵ": 179,
+  "Ĵ</w>": 328,
+  "ĵ": 180,
+  "ĵ</w>": 312,
+  "Ķ": 181,
+  "Ķ</w>": 256,
+  "ķ": 182,
+  "ķ</w>": 314,
+  "ĸ": 183,
+  "ĸ</w>": 277,
+  "Ĺ": 184,
+  "Ĺ</w>": 322,
+  "ĺ": 185,
+  "ĺ</w>": 285,
+  "Ļ": 186,
+  "Ļ</w>": 267,
+  "ļ": 187,
+  "ļ</w>": 290,
+  "Ľ": 188,
+  "Ľ</w>": 311,
+  "ľ": 189,
+  "ľ</w>": 299,
+  "Ŀ": 190,
+  "Ŀ</w>": 291,
+  "ŀ": 191,
+  "ŀ</w>": 293,
+  "Ł": 192,
+  "Ł</w>": 335,
+  "ł": 193,
+  "ł</w>": 252,
+  "Ń": 194,
+  "Ń</w>": 297
+}

text_tokenizer/added_tokens.json

@@ -0,0 +1,3 @@
+{
+  "<|EOS|>": 1024
+}

text_tokenizer/merges.txt

@@ -0,0 +1,768 @@
+#version: 0.2
+Ġ t
+h e
+Ġ a
+i n
+Ġt he
+e r
+o n
+Ġ ,
+r e
+Ġ s
+e d
+Ġ o
+Ġ w
+n d
+a t
+Ġ .
+o r
+i t
+Ġ c
+e n
+Ġ f
+i s
+e s
+a r
+Ġo f
+Ġ b
+a n
+Ġ in
+a l
+in g
+Ġ p
+Ġa nd
+a s
+Ġt o
+r o
+i c
+Ġ m
+Ġ d
+Ġ h
+i on
+l e
+o u
+Ġ T
+Ġ re
+Ġ =
+Ġ "
+Ġ A
+Ġ S
+en t
+i l
+Ġt h
+Ġ 1
+s t
+Ġ C
+e l
+o m
+Ġ l
+a m
+Ġ Ċ
+Ġ e
+Ġ n
+Ġ @
+a d
+a c
+Ġw as
+Ġ M
+u r
+ĠT he
+e c
+Ġ on
+l y
+Ġ B
+Ġ I
+Ġ g
+Ġ '
+e t
+o l
+i d
+i v
+i m
+Ġf or
+i r
+- @
+Ġ@ -@
+i g
+o t
+t er
+Ġa s
+Ġ H
+u s
+o w
+Ġs t
+u t
+it h
+a y
+Ġ 2
+Ġ P
+at ion
+v er
+Ġb e
+he r
+Ġth at
+Ġw ith
+Ġ R
+c e
+t h
+Ġ D
+Ġ is
+u n
+e m
+Ġ F
+Ġw h
+u l
+Ġb y
+Ġa l
+c h
+Ġ )
+Ġ (
+Ġ W
+Ġc on
+r a
+Ġ G
+o s
+Ġ L
+Ġ N
+Ġa t
+er s
+c t
+Ġ it
+Ġ1 9
+ro m
+a nd
+Ġa n
+u m
+es t
+Ġ J
+a g
+Ġ he
+0 0
+is t
+a in
+o d
+a v
+r i
+Ġ E
+Ġ O
+Ġf rom
+Ġc om
+Ġh is
+o p
+Ġp ro
+re s
+i es
+i f
+Ġ v
+or t
+er e
+il l
+l d
+Ġd e
+p p
+Ġs u
+o re
+ĠI n
+Ġ r
+Ġs e
+Ġw ere
+e w
+on g
+ig h
+ar d
+at e
+al l
+ar t
+a k
+ic h
+Ġc h
+Ġo r
+a b
+an t
+u d
+o c
+b er
+Ġe x
+g h
+it y
+at ed
+p t
+es s
+e ar
+Ġ K
+Ġp l
+am e
+q u
+iv e
+ro u
+Ġa re
+Ġ â
+Ġs h
+Ġ k
+ac k
+ec t
+Ġâ Ģ
+Ġ U
+Ġh ad
+s e
+Ġwh ich
+re d
+o v
+ĠS t
+as t
+Ġs p
+i an
+Ġ y
+m ent
+Ġ le
+Ġn ot
+g e
+or d
+r it
+i p
+in e
+el l
+al ly
+ou r
+o st
+igh t
+t her
+a p
+Ġ u
+is h
+ĠC h
+ou n
+i a
+Ġ 3
+av e
+ar y
+u st
+o g
+Ġ2 00
+Ġ un
+ou s
+ir st
+Ġ V
+c c
+Ġin c
+Ġ ;
+Ġcom p
+r u
+ion s
+Ġthe ir
+Ġb ut
+id e
+u re
+s o
+Ġcon t
+Ġin t
+f ter
+ic al
+i al
+Ġa r
+Ġf irst
+ou ld
+Ġit s
+he d
+ĠâĢ ĵ
+Ġw he
+w o
+ou t
+u b
+Ġ2 0
+f f
+Ġ :
+u e
+Ġ her
+ow n
+o k
+Ġal so
+Ġc l
+p er
+ig n
+at er
+r an
+or m
+i e
+om e
+or k
+as s
+i re
+e nd
+Ġre s
+Ġa b
+Ġa d
+Ġ us
+r y
+Ġre c
+Ġh ave
+ag e
+ĠH e
+Ġ 4
+Ġ ro
+m er
+Ġon e
+on d
+l ow
+Ġh as
+ĠT h
+d u
+Ġ 5
+Ġp er
+Ġbe en
+im e
+Ġt wo
+en ce
+l and
+Ġ1 8
+. @
+Ġ@ .@
+ul t
+re e
+ou gh
+i le
+Ġwh o
+ĠA l
+Ġs c
+ur ing
+p l
+or y
+it ion
+r ic
+ation s
+Ġd is
+Ġth is
+Ġb ec
+Ġa pp
+i z
+ĠI t
+a re
+ac h
+l ud
+ad e
+Ġpl ay
+Ġ j
+Ġm an
+ac t
+el y
+Ġp art
+Ġd es
+Ġa g
+Ġthe y
+Ġy ear
+oun t
+Ġ20 1
+Ġo ver
+Ġo ther
+ou nd
+Ġa fter
+i b
+o ver
+Ġs er
+Ġ en
+Ġof f
+Ġ im
+ct ion
+Ġ Y
+k e
+it e
+, @
+Ġ@ ,@
+t e
+ur n
+Ġinc lud
+res s
+an ce
+an g
+Ġat t
+ic e
+ac e
+ar k
+Ġo ut
+w n
+p h
+em ber
+Ġp re
+Ġu p
+en s
+m an
+Ġe v
+Ġt ime
+nd er
+rou gh
+c ed
+Ġf in
+Ġint o
+on e
+p ort
+rou nd
+w e
+re n
+l es
+in t
+ĠO n
+v el
+Ġcom m
+Ġs he
+as on
+am p
+Ġt e
+Ġw ould
+w ard
+Ġm ore
+Ġ 6
+i ed
+os e
+ri b
+ĠU n
+Ġal l
+ing s
+ter n
+c es
+ab le
+Ġw e
+it ed
+e ver
+ent s
+Ġh im
+as ed
+or s
+o y
+o od
+Ġc ent
+i x
+as e
+il d
+ĠA n
+Ġ 7
+Ġw ork
+at es
+i ous
+at h
+Ġp o
+ro p
+ol d
+al s
+is s
+e y
+ic t
+Ġf e
+Ġthe m
+g an
+Ġs ec
+Ġb et
+Ġwhe n
+Ġs ong
+Ġre m
+e p
+f orm
+a il
+f er
+Ġe ar
+ub l
+a w
+Ġk n
+ak e
+a us
+Ġm ost
+Ġcon s
+Ġd uring
+ĠA s
+or th
+Ġn ew
+er ed
+il m
+v ed
+at t
+Ġon ly
+Ġ 9
+Ġd ec
+Ġ 8
+ic k
+Ġg ame
+on s
+u g
+Ġt r
+f t
+ot h
+o ok
+ĠM ar
+re at
+w ay
+Ġc an
+ol low
+ou th
+we en
+ĠE n
+Ġ19 9
+ter s
+Ġre l
+in d
+Ġab out
+Ġse ason
+Ġag ain
+r al
+Ġth ree
+ation al
+Ġu nder
+ul ar
+Ġm e
+Ġth an
+ĠC om
+ĠA r
+h ip
+o b
+Ġn e
+Ġbet ween
+Ġf l
+h n
+v e
+Ġch ar
+Ġc ol
+Ġrec ord
+i ew
+r on
+f ore
+Ġth rough
+is ion
+or n
+Ġ 00
+oc k
+Ġ ver
+Ġl ater
+Ġn um
+Ġe nd
+ol og
+am es
+Ġp os
+Ġw rit
+Ġpro du
+Ġwh ile
+Ġa ct
+Ġre le
+Ġf ilm
+is hed
+Ġp r
+an s
+Ġre g
+Ġfor m
+Ġas s
+ĠS e
+ur y
+t ed
+t s
+Ġm ade
+Ġsu b
+Ġp e
+Ġs o
+or ld
+Ġre t
+ĠN ew
+Ġsp ec
+Ġa cc
+Ġ qu
+Ġwhe re
+en er
+Ġm ov
+he s
+mer ic
+at ing
+Ġin ter
+ĠL e
+ĠA meric
+Ġ ra
+Ġs ome
+Ġc o
+Ġl ar
+Ġb u
+Ġde f
+b um
+Ġa c
+Ġm us
+Ġf ollow
+ĠA t
+in s
+iv ed
+if ic
+u al
+Ġa m
+Ġsu ch
+Ġsec ond
+i ke
+Ġf our
+Ġin d
+an n
+he n
+Ġus ed
+ĠR e
+ic s
+le ct
+Ġd ay
+i el
+il y
+ĠTh is
+Ġ 0
+Ġp ubl
+Ġc all
+ĠJ o
+l l
+Ġal bum
+Ġ00 0
+ran s
+Ġd o
+an y
+Ġbe fore
+ro s
+ĠS h
+Ġs y
+a id
+ĠEn g
+Ġbe ing
+Ġ1 0
+u c
+Ġe p
+Ġsu pp
+Ġthe re
+Ġyear s
+ar s
+ow ever
+Ġ ent
+if e
+Ġh igh
+Ġf ound
+ir d
+Ġn o
+Ġs et
+in es
+iv er
+i o
+ot her
+j ect
+Ġs ur
+a j
+t en
+Ġt ra
+Ġ1 2
+is ed
+it ies
+vel op
+Ġb l
+al e
+Ġser ies
+Ġl oc
+Ġnum ber
+Ġp res
+an e
+aus e
+od e
+e k
+t on
+ĠS c
+i er
+is e
+Ġse ver
+in ce
+Ġb oth
+an k
+ro w
+ire ct
+s on
+Ġthe n
+ĠB rit
+i et
+Ġ1 6
+Ġep is
+Ġinclud ing
+it s
+ig in
+p r
+Ġ /
+Ġagain st
+Ġw ell
+Ġbec ame
+Ġex p
+Ġkn own
+Ġt rans
+Ġchar ac
+ĠâĢ Ķ
+r am
+Ġb ack
+Ġad d
+Ġp op
+Ġg o
+ur ch
+Ġdes c
+Ġs ing
+iel d
+Ġper form
+ain ed
+Ġre ce
+id ent
+Ġe m
+er t
+u res
+Ġin v
+Ġde p
+Ġ19 8
+a ir
+er n
+at her
+f ul
+Ġ Z
+Ġm on
+Ġman y
+Ġm ain
+Ġst ud
+Ġl ong
+in n
+th ough
+u p
+o ol
+ĠUn ited
+l ed
+em ent
+Ġ1 5
+ow er
+ĠJo hn
+Ġo p
+Ġ1 1
+in ed
+Ġm et
+o ber
+le y
+Ġ1 7
+Ġcent ury
+Ġte am
+Ġ est
+ĠA fter
+y l
+Ġm in
+u ch
+ut e
+Ġde velop
+ĠS he
+i am
+Ġsh ow
+el f
+Ġre p
+Ġcon c
+at ive
+Ġc re
+over n
+a red
+Ġ19 4
+Ġor igin
+Ġs m
+iv ers
+a z
+Ġle ad
+Ġsever al
+a h
+Ġo b
+Ġre v
+Ġm ill
+er m
+u ally
+o ot
+Ġbe gan
+Ġ19 6

text_tokenizer/special_tokens_map.json

@@ -0,0 +1,5 @@
+{
+  "bos_token": "<|endoftext|>",
+  "eos_token": "<|EOS|>",
+  "unk_token": "<|endoftext|>"
+}

text_tokenizer/tokenizer_config.json

@@ -0,0 +1,34 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "errors": "replace",
+  "model_max_length": 512,
+  "name_or_path": "models/clip_caption_model",
+  "pad_token": null,
+  "special_tokens_map_file": null,
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

text_tokenizer/vocab.json

@@ -0,0 +1,1026 @@
+{
+  "!": 1,
+  "\"": 2,
+  "#": 3,
+  "$": 4,
+  "%": 5,
+  "&": 6,
+  "'": 7,
+  "(": 8,
+  ")": 9,
+  "*": 10,
+  "+": 11,
+  ",": 12,
+  ",@": 624,
+  "-": 13,
+  "-@": 339,
+  ".": 14,
+  ".@": 573,
+  "/": 15,
+  "0": 16,
+  "00": 395,
+  "1": 17,
+  "2": 18,
+  "3": 19,
+  "4": 20,
+  "5": 21,
+  "6": 22,
+  "7": 23,
+  "8": 24,
+  "9": 25,
+  ":": 26,
+  ";": 27,
+  "<": 28,
+  "<|endoftext|>": 0,
+  "=": 29,
+  ">": 30,
+  "?": 31,
+  "@": 32,
+  "A": 33,
+  "B": 34,
+  "C": 35,
+  "D": 36,
+  "E": 37,
+  "F": 38,
+  "G": 39,
+  "H": 40,
+  "I": 41,
+  "J": 42,
+  "K": 43,
+  "L": 44,
+  "M": 45,
+  "N": 46,
+  "O": 47,
+  "P": 48,
+  "Q": 49,
+  "R": 50,
+  "S": 51,
+  "T": 52,
+  "U": 53,
+  "V": 54,
+  "W": 55,
+  "X": 56,
+  "Y": 57,
+  "Z": 58,
+  "[": 59,
+  "\\": 60,
+  "]": 61,
+  "^": 62,
+  "_": 63,
+  "`": 64,
+  "a": 65,
+  "ab": 435,
+  "able": 677,
+  "ac": 320,
+  "ace": 634,
+  "ach": 595,
+  "ack": 457,
+  "act": 601,
+  "ad": 319,
+  "ade": 597,
+  "ag": 393,
+  "age": 554,
+  "ah": 1015,
+  "aid": 873,
+  "ail": 714,
+  "ain": 397,
+  "ained": 953,
+  "air": 962,
+  "aj": 897,
+  "ak": 431,
+  "ake": 720,
+  "al": 285,
+  "ale": 905,
+  "all": 429,
+  "ally": 480,
+  "als": 700,
+  "am": 314,
+  "ame": 449,
+  "ames": 791,
+  "amp": 663,
+  "an": 283,
+  "ance": 630,
+  "and": 388,
+  "ane": 910,
+  "ang": 631,
+  "ank": 921,
+  "ann": 849,
+  "ans": 801,
+  "ant": 436,
+  "any": 868,
+  "ap": 485,
+  "ar": 280,
+  "ard": 427,
+  "are": 594,
+  "ared": 1007,
+  "ark": 635,
+  "ars": 882,
+  "art": 430,
+  "ary": 493,
+  "as": 289,
+  "ase": 689,
+  "ased": 683,
+  "ason": 662,
+  "ass": 544,
+  "ast": 467,
+  "at": 271,
+  "ate": 428,
+  "ated": 443,
+  "ater": 538,
+  "ates": 694,
+  "ath": 696,
+  "ather": 964,
+  "ating": 824,
+  "ation": 354,
+  "ational": 761,
+  "ations": 587,
+  "ative": 1004,
+  "att": 731,
+  "aus": 721,
+  "ause": 911,
+  "av": 399,
+  "ave": 492,
+  "aw": 718,
+  "ay": 351,
+  "az": 1012,
+  "b": 66,
+  "ber": 439,
+  "bum": 834,
+  "c": 67,
+  "cc": 501,
+  "ce": 361,
+  "ced": 648,
+  "ces": 676,
+  "ch": 372,
+  "ct": 384,
+  "ction": 620,
+  "d": 68,
+  "du": 564,
+  "e": 69,
+  "ear": 446,
+  "ec": 325,
+  "ect": 458,
+  "ed": 267,
+  "ek": 913,
+  "el": 311,
+  "elf": 1001,
+  "ell": 479,
+  "ely": 602,
+  "em": 366,
+  "ember": 639,
+  "ement": 978,
+  "en": 276,
+  "ence": 570,
+  "end": 546,
+  "ener": 820,
+  "ens": 642,
+  "ent": 305,
+  "ents": 681,
+  "ep": 712,
+  "er": 262,
+  "ere": 413,
+  "ered": 728,
+  "erm": 1019,
+  "ern": 963,
+  "ers": 383,
+  "ert": 957,
+  "es": 279,
+  "ess": 445,
+  "est": 391,
+  "et": 332,
+  "ever": 680,
+  "ew": 424,
+  "ey": 702,
+  "f": 70,
+  "fer": 715,
+  "ff": 528,
+  "fore": 780,
+  "form": 713,
+  "ft": 741,
+  "fter": 514,
+  "ful": 965,
+  "g": 71,
+  "gan": 706,
+  "ge": 474,
+  "gh": 441,
+  "h": 72,
+  "he": 258,
+  "hed": 521,
+  "hen": 850,
+  "her": 357,
+  "hes": 822,
+  "hip": 768,
+  "hn": 773,
+  "i": 73,
+  "ia": 490,
+  "ial": 516,
+  "iam": 999,
+  "ian": 469,
+  "ib": 614,
+  "ic": 292,
+  "ical": 515,
+  "ice": 633,
+  "ich": 432,
+  "ick": 736,
+  "ics": 853,
+  "ict": 703,
+  "id": 334,
+  "ide": 509,
+  "ident": 955,
+  "ie": 541,
+  "ied": 669,
+  "iel": 856,
+  "ield": 951,
+  "ier": 916,
+  "ies": 409,
+  "iet": 927,
+  "iew": 778,
+  "if": 410,
+  "ife": 885,
+  "ific": 841,
+  "ig": 341,
+  "igh": 426,
+  "ight": 483,
+  "igin": 932,
+  "ign": 537,
+  "ike": 846,
+  "il": 306,
+  "ild": 690,
+  "ile": 578,
+  "ill": 414,
+  "ilm": 729,
+  "ily": 857,
+  "im": 336,
+  "ime": 568,
+  "in": 260,
+  "ince": 919,
+  "ind": 755,
+  "ine": 478,
+  "ined": 984,
+  "ines": 891,
+  "ing": 286,
+  "ings": 674,
+  "inn": 972,
+  "ins": 839,
+  "int": 657,
+  "io": 893,
+  "ion": 296,
+  "ions": 506,
+  "ious": 695,
+  "ip": 477,
+  "ir": 338,
+  "ird": 888,
+  "ire": 545,
+  "irect": 923,
+  "irst": 499,
+  "is": 278,
+  "ise": 917,
+  "ised": 901,
+  "ish": 487,
+  "ished": 799,
+  "ision": 782,
+  "iss": 701,
+  "ist": 396,
+  "it": 274,
+  "ite": 623,
+  "ited": 679,
+  "ith": 350,
+  "ities": 902,
+  "ition": 585,
+  "its": 931,
+  "ity": 442,
+  "iv": 335,
+  "ive": 451,
+  "ived": 840,
+  "iver": 892,
+  "ivers": 1011,
+  "ix": 688,
+  "iz": 592,
+  "j": 74,
+  "ject": 895,
+  "k": 75,
+  "ke": 622,
+  "l": 76,
+  "land": 571,
+  "ld": 415,
+  "le": 297,
+  "lect": 854,
+  "led": 977,
+  "les": 656,
+  "ley": 987,
+  "ll": 863,
+  "low": 561,
+  "lud": 596,
+  "ly": 327,
+  "m": 77,
+  "man": 643,
+  "ment": 471,
+  "mer": 558,
+  "meric": 823,
+  "n": 78,
+  "nd": 270,
+  "nder": 646,
+  "o": 79,
+  "ob": 769,
+  "ober": 986,
+  "oc": 438,
+  "ock": 785,
+  "od": 398,
+  "ode": 912,
+  "og": 495,
+  "ok": 533,
+  "ol": 333,
+  "old": 699,
+  "ollow": 748,
+  "olog": 790,
+  "om": 312,
+  "ome": 542,
+  "on": 263,
+  "ond": 560,
+  "one": 651,
+  "ong": 425,
+  "ons": 738,
+  "ood": 686,
+  "ook": 743,
+  "ool": 975,
+  "oot": 1021,
+  "op": 406,
+  "or": 273,
+  "ord": 475,
+  "ore": 419,
+  "ork": 543,
+  "orld": 813,
+  "orm": 540,
+  "orn": 783,
+  "ors": 684,
+  "ort": 412,
+  "orth": 726,
+  "ory": 584,
+  "os": 379,
+  "ose": 670,
+  "ost": 482,
+  "ot": 342,
+  "oth": 742,
+  "other": 894,
+  "ou": 298,
+  "ough": 577,
+  "ould": 519,
+  "oun": 489,
+  "ound": 612,
+  "ount": 608,
+  "our": 481,
+  "ous": 498,
+  "out": 525,
+  "outh": 749,
+  "ov": 465,
+  "over": 615,
+  "overn": 1006,
+  "ow": 347,
+  "ower": 980,
+  "owever": 883,
+  "own": 532,
+  "oy": 685,
+  "p": 80,
+  "per": 536,
+  "ph": 638,
+  "pl": 583,
+  "port": 652,
+  "pp": 417,
+  "pr": 933,
+  "pt": 444,
+  "q": 81,
+  "qu": 450,
+  "r": 82,
+  "ra": 377,
+  "ral": 759,
+  "ram": 943,
+  "ran": 539,
+  "rans": 866,
+  "re": 265,
+  "reat": 745,
+  "red": 464,
+  "ree": 576,
+  "ren": 655,
+  "res": 408,
+  "ress": 629,
+  "ri": 400,
+  "rib": 671,
+  "ric": 586,
+  "rit": 476,
+  "ro": 291,
+  "rom": 387,
+  "ron": 779,
+  "rop": 698,
+  "ros": 870,
+  "rou": 452,
+  "rough": 647,
+  "round": 653,
+  "row": 922,
+  "ru": 505,
+  "ry": 551,
+  "s": 83,
+  "se": 462,
+  "so": 511,
+  "son": 924,
+  "st": 309,
+  "t": 84,
+  "te": 626,
+  "ted": 807,
+  "ten": 898,
+  "ter": 343,
+  "tern": 675,
+  "ters": 753,
+  "th": 362,
+  "ther": 484,
+  "though": 973,
+  "ton": 914,
+  "ts": 808,
+  "u": 85,
+  "ual": 842,
+  "ually": 1020,
+  "ub": 526,
+  "ubl": 717,
+  "uc": 877,
+  "uch": 995,
+  "ud": 437,
+  "ue": 530,
+  "ug": 739,
+  "ul": 369,
+  "ular": 763,
+  "ult": 575,
+  "um": 390,
+  "un": 365,
+  "up": 974,
+  "ur": 323,
+  "urch": 948,
+  "ure": 510,
+  "ures": 958,
+  "uring": 582,
+  "urn": 627,
+  "ury": 806,
+  "us": 346,
+  "ust": 494,
+  "ut": 349,
+  "ute": 996,
+  "v": 86,
+  "ve": 774,
+  "ved": 730,
+  "vel": 659,
+  "velop": 903,
+  "ver": 355,
+  "w": 87,
+  "ward": 666,
+  "way": 746,
+  "we": 654,
+  "ween": 750,
+  "wn": 637,
+  "wo": 524,
+  "x": 88,
+  "y": 89,
+  "yl": 993,
+  "z": 90,
+  "{": 91,
+  "|": 92,
+  "}": 93,
+  "~": 94,
+  "¡": 95,
+  "¢": 96,
+  "£": 97,
+  "¤": 98,
+  "¥": 99,
+  "¦": 100,
+  "§": 101,
+  "¨": 102,
+  "©": 103,
+  "ª": 104,
+  "«": 105,
+  "¬": 106,
+  "®": 107,
+  "¯": 108,
+  "°": 109,
+  "±": 110,
+  "²": 111,
+  "³": 112,
+  "´": 113,
+  "µ": 114,
+  "¶": 115,
+  "·": 116,
+  "¸": 117,
+  "¹": 118,
+  "º": 119,
+  "»": 120,
+  "¼": 121,
+  "½": 122,
+  "¾": 123,
+  "¿": 124,
+  "À": 125,
+  "Á": 126,
+  "Â": 127,
+  "Ã": 128,
+  "Ä": 129,
+  "Å": 130,
+  "Æ": 131,
+  "Ç": 132,
+  "È": 133,
+  "É": 134,
+  "Ê": 135,
+  "Ë": 136,
+  "Ì": 137,
+  "Í": 138,
+  "Î": 139,
+  "Ï": 140,
+  "Ð": 141,
+  "Ñ": 142,
+  "Ò": 143,
+  "Ó": 144,
+  "Ô": 145,
+  "Õ": 146,
+  "Ö": 147,
+  "×": 148,
+  "Ø": 149,
+  "Ù": 150,
+  "Ú": 151,
+  "Û": 152,
+  "Ü": 153,
+  "Ý": 154,
+  "Þ": 155,
+  "ß": 156,
+  "à": 157,
+  "á": 158,
+  "â": 159,
+  "ã": 160,
+  "ä": 161,
+  "å": 162,
+  "æ": 163,
+  "ç": 164,
+  "è": 165,
+  "é": 166,
+  "ê": 167,
+  "ë": 168,
+  "ì": 169,
+  "í": 170,
+  "î": 171,
+  "ï": 172,
+  "ð": 173,
+  "ñ": 174,
+  "ò": 175,
+  "ó": 176,
+  "ô": 177,
+  "õ": 178,
+  "ö": 179,
+  "÷": 180,
+  "ø": 181,
+  "ù": 182,
+  "ú": 183,
+  "û": 184,
+  "ü": 185,
+  "ý": 186,
+  "þ": 187,
+  "ÿ": 188,
+  "Ā": 189,
+  "ā": 190,
+  "Ă": 191,
+  "ă": 192,
+  "Ą": 193,
+  "ą": 194,
+  "Ć": 195,
+  "ć": 196,
+  "Ĉ": 197,
+  "ĉ": 198,
+  "Ċ": 199,
+  "ċ": 200,
+  "Č": 201,
+  "č": 202,
+  "Ď": 203,
+  "ď": 204,
+  "Đ": 205,
+  "đ": 206,
+  "Ē": 207,
+  "ē": 208,
+  "Ĕ": 209,
+  "ĕ": 210,
+  "Ė": 211,
+  "ė": 212,
+  "Ę": 213,
+  "ę": 214,
+  "Ě": 215,
+  "ě": 216,
+  "Ĝ": 217,
+  "ĝ": 218,
+  "Ğ": 219,
+  "ğ": 220,
+  "Ġ": 221,
+  "Ġ\"": 302,
+  "Ġ'": 331,
+  "Ġ(": 374,
+  "Ġ)": 373,
+  "Ġ,": 264,
+  "Ġ.": 272,
+  "Ġ/": 934,
+  "Ġ0": 859,
+  "Ġ00": 784,
+  "Ġ000": 865,
+  "Ġ1": 308,
+  "Ġ10": 876,
+  "Ġ11": 983,
+  "Ġ12": 900,
+  "Ġ15": 979,
+  "Ġ16": 928,
+  "Ġ17": 988,
+  "Ġ18": 572,
+  "Ġ19": 386,
+  "Ġ194": 1008,
+  "Ġ196": 1023,
+  "Ġ198": 961,
+  "Ġ199": 752,
+  "Ġ2": 352,
+  "Ġ20": 527,
+  "Ġ200": 496,
+  "Ġ201": 609,
+  "Ġ3": 491,
+  "Ġ4": 556,
+  "Ġ5": 565,
+  "Ġ6": 668,
+  "Ġ7": 692,
+  "Ġ8": 735,
+  "Ġ9": 733,
+  "Ġ:": 529,
+  "Ġ;": 503,
+  "Ġ=": 301,
+  "Ġ@": 318,
+  "Ġ@,@": 625,
+  "Ġ@-@": 340,
+  "Ġ@.@": 574,
+  "ĠA": 303,
+  "ĠAfter": 992,
+  "ĠAl": 580,
+  "ĠAmeric": 827,
+  "ĠAn": 691,
+  "ĠAr": 767,
+  "ĠAs": 725,
+  "ĠAt": 838,
+  "ĠB": 328,
+  "ĠBrit": 926,
+  "ĠC": 310,
+  "ĠCh": 488,
+  "ĠCom": 766,
+  "ĠD": 363,
+  "ĠE": 401,
+  "ĠEn": 751,
+  "ĠEng": 874,
+  "ĠF": 367,
+  "ĠG": 378,
+  "ĠH": 345,
+  "ĠHe": 555,
+  "ĠI": 329,
+  "ĠIn": 420,
+  "ĠIt": 593,
+  "ĠJ": 392,
+  "ĠJo": 862,
+  "ĠJohn": 981,
+  "ĠK": 447,
+  "ĠL": 380,
+  "ĠLe": 826,
+  "ĠM": 322,
+  "ĠMar": 744,
+  "ĠN": 381,
+  "ĠNew": 815,
+  "ĠO": 402,
+  "ĠOn": 658,
+  "ĠP": 353,
+  "ĠR": 360,
+  "ĠRe": 852,
+  "ĠS": 304,
+  "ĠSc": 915,
+  "ĠSe": 805,
+  "ĠSh": 871,
+  "ĠShe": 998,
+  "ĠSt": 466,
+  "ĠT": 299,
+  "ĠTh": 563,
+  "ĠThe": 324,
+  "ĠThis": 858,
+  "ĠU": 460,
+  "��Un": 672,
+  "ĠUnited": 976,
+  "ĠV": 500,
+  "ĠW": 375,
+  "ĠY": 621,
+  "ĠZ": 966,
+  "Ġa": 259,
+  "Ġab": 548,
+  "Ġabout": 756,
+  "Ġac": 835,
+  "Ġacc": 817,
+  "Ġact": 796,
+  "Ġad": 549,
+  "Ġadd": 945,
+  "Ġafter": 613,
+  "Ġag": 605,
+  "Ġagain": 758,
+  "Ġagainst": 935,
+  "Ġal": 371,
+  "Ġalbum": 864,
+  "Ġall": 673,
+  "Ġalso": 534,
+  "Ġam": 843,
+  "Ġan": 389,
+  "Ġand": 288,
+  "Ġapp": 591,
+  "Ġar": 517,
+  "Ġare": 453,
+  "Ġas": 344,
+  "Ġass": 804,
+  "Ġat": 382,
+  "Ġatt": 632,
+  "Ġb": 282,
+  "Ġback": 944,
+  "Ġbe": 356,
+  "Ġbec": 590,
+  "Ġbecame": 937,
+  "Ġbeen": 567,
+  "Ġbefore": 869,
+  "Ġbegan": 1022,
+  "Ġbeing": 875,
+  "Ġbet": 708,
+  "Ġbetween": 771,
+  "Ġbl": 904,
+  "Ġboth": 920,
+  "Ġbu": 832,
+  "Ġbut": 508,
+  "Ġby": 370,
+  "Ġc": 275,
+  "Ġcall": 861,
+  "Ġcan": 747,
+  "Ġcent": 687,
+  "Ġcentury": 989,
+  "Ġch": 433,
+  "Ġchar": 775,
+  "Ġcharac": 941,
+  "Ġcl": 535,
+  "Ġco": 830,
+  "Ġcol": 776,
+  "Ġcom": 404,
+  "Ġcomm": 660,
+  "Ġcomp": 504,
+  "Ġcon": 376,
+  "Ġconc": 1003,
+  "Ġcons": 723,
+  "Ġcont": 512,
+  "Ġcre": 1005,
+  "Ġd": 294,
+  "Ġday": 855,
+  "Ġde": 416,
+  "Ġdec": 734,
+  "Ġdef": 833,
+  "Ġdep": 960,
+  "Ġdes": 604,
+  "Ġdesc": 949,
+  "Ġdevelop": 997,
+  "Ġdis": 588,
+  "Ġdo": 867,
+  "Ġduring": 724,
+  "Ġe": 316,
+  "Ġear": 716,
+  "Ġem": 956,
+  "Ġen": 617,
+  "Ġend": 789,
+  "Ġent": 884,
+  "Ġep": 878,
+  "Ġepis": 929,
+  "Ġest": 991,
+  "Ġev": 644,
+  "Ġex": 440,
+  "Ġexp": 938,
+  "Ġf": 277,
+  "Ġfe": 704,
+  "Ġfilm": 798,
+  "Ġfin": 649,
+  "Ġfirst": 518,
+  "Ġfl": 772,
+  "Ġfollow": 837,
+  "Ġfor": 337,
+  "Ġform": 803,
+  "Ġfound": 887,
+  "Ġfour": 847,
+  "Ġfrom": 403,
+  "Ġg": 330,
+  "Ġgame": 737,
+  "Ġgo": 947,
+  "Ġh": 295,
+  "Ġhad": 461,
+  "Ġhas": 562,
+  "Ġhave": 553,
+  "Ġhe": 394,
+  "Ġher": 531,
+  "Ġhigh": 886,
+  "Ġhim": 682,
+  "Ġhis": 405,
+  "Ġim": 619,
+  "Ġin": 284,
+  "Ġinc": 502,
+  "Ġinclud": 628,
+  "Ġincluding": 930,
+  "Ġind": 848,
+  "Ġint": 513,
+  "Ġinter": 825,
+  "Ġinto": 650,
+  "Ġinv": 959,
+  "Ġis": 364,
+  "Ġit": 385,
+  "Ġits": 520,
+  "Ġj": 599,
+  "Ġk": 456,
+  "Ġkn": 719,
+  "Ġknown": 939,
+  "Ġl": 313,
+  "Ġlar": 831,
+  "Ġlater": 787,
+  "Ġle": 472,
+  "Ġlead": 1013,
+  "Ġloc": 907,
+  "Ġlong": 971,
+  "Ġm": 293,
+  "Ġmade": 809,
+  "Ġmain": 969,
+  "Ġman": 600,
+  "Ġmany": 968,
+  "Ġme": 764,
+  "Ġmet": 985,
+  "Ġmill": 1018,
+  "Ġmin": 994,
+  "Ġmon": 967,
+  "Ġmore": 667,
+  "Ġmost": 722,
+  "Ġmov": 821,
+  "Ġmus": 836,
+  "Ġn": 317,
+  "Ġne": 770,
+  "Ġnew": 727,
+  "Ġno": 889,
+  "Ġnot": 473,
+  "Ġnum": 788,
+  "Ġnumber": 908,
+  "Ġo": 268,
+  "Ġob": 1016,
+  "Ġof": 281,
+  "Ġoff": 618,
+  "Ġon": 326,
+  "Ġone": 559,
+  "Ġonly": 732,
+  "Ġop": 982,
+  "Ġor": 434,
+  "Ġorigin": 1009,
+  "Ġother": 611,
+  "Ġout": 636,
+  "Ġover": 610,
+  "Ġp": 287,
+  "Ġpart": 603,
+  "Ġpe": 811,
+  "Ġper": 566,
+  "Ġperform": 952,
+  "Ġpl": 448,
+  "Ġplay": 598,
+  "Ġpo": 697,
+  "Ġpop": 946,
+  "Ġpos": 792,
+  "Ġpr": 800,
+  "Ġpre": 640,
+  "Ġpres": 909,
+  "Ġpro": 407,
+  "Ġprodu": 794,
+  "Ġpubl": 860,
+  "Ġqu": 818,
+  "Ġr": 421,
+  "Ġra": 828,
+  "Ġre": 300,
+  "Ġrec": 552,
+  "Ġrece": 954,
+  "Ġrecord": 777,
+  "Ġreg": 802,
+  "Ġrel": 754,
+  "Ġrele": 797,
+  "Ġrem": 711,
+  "Ġrep": 1002,
+  "Ġres": 547,
+  "Ġret": 814,
+  "Ġrev": 1017,
+  "Ġro": 557,
+  "Ġs": 266,
+  "Ġsc": 581,
+  "Ġse": 422,
+  "Ġseason": 757,
+  "Ġsec": 707,
+  "Ġsecond": 845,
+  "Ġser": 616,
+  "Ġseries": 906,
+  "Ġset": 890,
+  "Ġsever": 918,
+  "Ġseveral": 1014,
+  "Ġsh": 455,
+  "Ġshe": 661,
+  "Ġshow": 1000,
+  "Ġsing": 950,
+  "Ġsm": 1010,
+  "Ġso": 812,
+  "Ġsome": 829,
+  "Ġsong": 710,
+  "Ġsp": 468,
+  "Ġspec": 816,
+  "Ġst": 348,
+  "Ġstud": 970,
+  "Ġsu": 418,
+  "Ġsub": 810,
+  "Ġsuch": 844,
+  "Ġsupp": 879,
+  "Ġsur": 896,
+  "Ġsy": 872,
+  "Ġt": 257,
+  "Ġte": 664,
+  "Ġteam": 990,
+  "Ġth": 307,
+  "Ġthan": 765,
+  "Ġthat": 358,
+  "Ġthe": 261,
+  "Ġtheir": 507,
+  "Ġthem": 705,
+  "Ġthen": 925,
+  "Ġthere": 880,
+  "Ġthey": 606,
+  "Ġthis": 589,
+  "Ġthree": 760,
+  "Ġthrough": 781,
+  "Ġtime": 645,
+  "Ġto": 290,
+  "Ġtr": 740,
+  "Ġtra": 899,
+  "Ġtrans": 940,
+  "Ġtwo": 569,
+  "Ġu": 486,
+  "Ġun": 497,
+  "Ġunder": 762,
+  "Ġup": 641,
+  "Ġus": 550,
+  "Ġused": 851,
+  "Ġv": 411,
+  "Ġver": 786,
+  "Ġw": 269,
+  "Ġwas": 321,
+  "Ġwe": 678,
+  "Ġwell": 936,
+  "Ġwere": 423,
+  "Ġwh": 368,
+  "Ġwhe": 523,
+  "Ġwhen": 709,
+  "Ġwhere": 819,
+  "Ġwhich": 463,
+  "Ġwhile": 795,
+  "Ġwho": 579,
+  "Ġwith": 359,
+  "Ġwork": 693,
+  "Ġwould": 665,
+  "Ġwrit": 793,
+  "Ġy": 470,
+  "Ġyear": 607,
+  "Ġyears": 881,
+  "Ġâ": 454,
+  "ĠâĢ": 459,
+  "ĠâĢĵ": 522,
+  "ĠâĢĶ": 942,
+  "ĠĊ": 315,
+  "ġ": 222,
+  "Ģ": 223,
+  "ģ": 224,
+  "Ĥ": 225,
+  "ĥ": 226,
+  "Ħ": 227,
+  "ħ": 228,
+  "Ĩ": 229,
+  "ĩ": 230,
+  "Ī": 231,
+  "ī": 232,
+  "Ĭ": 233,
+  "ĭ": 234,
+  "Į": 235,
+  "į": 236,
+  "İ": 237,
+  "ı": 238,
+  "IJ": 239,
+  "ij": 240,
+  "Ĵ": 241,
+  "ĵ": 242,
+  "Ķ": 243,
+  "ķ": 244,
+  "ĸ": 245,
+  "Ĺ": 246,
+  "ĺ": 247,
+  "Ļ": 248,
+  "ļ": 249,
+  "Ľ": 250,
+  "ľ": 251,
+  "Ŀ": 252,
+  "ŀ": 253,
+  "Ł": 254,
+  "ł": 255,
+  "Ń": 256
+}