End of training

README.md

@@ -14,7 +14,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model was trained from scratch on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.1577
+- Loss: 0.1153
 
 ## Model description
 
@@ -33,30 +33,27 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.0006047816549758072
-- train_batch_size: 8
-- eval_batch_size: 8
+- learning_rate: 1e-05
+- train_batch_size: 128
+- eval_batch_size: 128
 - seed: 42
-- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.964129172421366,0.8471340191802936) and epsilon=1.51279024695782e-08 and optimizer_args=No additional optimizer arguments
+- optimizer: Use OptimizerNames.ADAMW_APEX_FUSED with betas=(0.826646043090655,0.991636944120939) and epsilon=3.4341677539323e-07 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 2593
-- num_epochs: 500
+- lr_scheduler_warmup_steps: 5000
+- num_epochs: 200
 
 ### Training results
 
 | Training Loss | Epoch | Step  | Validation Loss |
 |:-------------:|:-----:|:-----:|:---------------:|
-| 0.2814        | 1.0   | 6351  | 0.1742          |
-| 0.2857        | 2.0   | 12702 | 0.1582          |
-| 0.2872        | 3.0   | 19053 | 0.1529          |
-| 0.2733        | 4.0   | 25404 | 0.1528          |
-| 0.3071        | 5.0   | 31755 | 0.1805          |
-| 0.297         | 6.0   | 38106 | 0.1577          |
+| 0.0237        | 1.0   | 18731 | 0.1124          |
+| 0.0216        | 2.0   | 37462 | 0.1128          |
+| 0.0201        | 3.0   | 56193 | 0.1153          |
 
 
 ### Framework versions
 
-- Transformers 4.49.0
-- Pytorch 2.6.0.dev20241217
-- Datasets 2.20.0
-- Tokenizers 0.21.0
+- Transformers 4.50.3
+- Pytorch 2.6.0+cu126
+- Datasets 3.3.0
+- Tokenizers 0.21.1

added_tokens.json

@@ -1,247 +1,415 @@
 {
-  "±": 32314,
-  "→": 32189,
-  "⌜": 32266,
-  "⌝": 32231,
-  "●": 32179,
-  "⟦": 32218,
-  "⟧": 32200,
-  "𐀀": 32299,
-  "𐀁": 32169,
-  "𐀂": 32110,
-  "𐀃": 32301,
-  "𐀄": 32261,
-  "𐀅": 32257,
-  "𐀆": 32102,
-  "𐀇": 32341,
-  "𐀈": 32136,
-  "𐀉": 32187,
-  "𐀊": 32286,
-  "𐀋": 32127,
-  "𐀍": 32108,
-  "𐀎": 32315,
-  "𐀏": 32162,
-  "𐀐": 32154,
-  "𐀑": 32306,
-  "𐀒": 32152,
-  "𐀓": 32156,
-  "𐀔": 32157,
-  "𐀕": 32239,
-  "𐀖": 32295,
-  "𐀗": 32210,
-  "𐀘": 32226,
-  "𐀙": 32340,
-  "𐀚": 32288,
-  "𐀛": 32304,
-  "𐀜": 32188,
-  "𐀝": 32272,
-  "𐀞": 32171,
-  "𐀟": 32268,
-  "𐀠": 32124,
-  "𐀡": 32175,
-  "𐀢": 32219,
-  "𐀣": 32293,
-  "𐀤": 32300,
-  "𐀥": 32287,
-  "𐀦": 32172,
-  "𐀨": 32106,
-  "𐀩": 32214,
-  "𐀪": 32115,
-  "𐀫": 32225,
-  "𐀬": 32309,
-  "𐀭": 32159,
-  "𐀮": 32290,
-  "𐀯": 32237,
-  "𐀰": 32313,
-  "𐀱": 32194,
-  "𐀲": 32281,
-  "𐀳": 32107,
-  "𐀴": 32265,
-  "𐀵": 32325,
-  "𐀶": 32126,
-  "𐀷": 32164,
-  "𐀸": 32130,
-  "𐀹": 32198,
-  "𐀺": 32318,
-  "𐀼": 32263,
-  "𐀽": 32321,
-  "𐀿": 32271,
-  "𐁀": 32216,
-  "𐁁": 32160,
-  "𐁂": 32144,
-  "𐁃": 32135,
-  "𐁄": 32331,
-  "𐁅": 32167,
-  "𐁆": 32208,
-  "𐁇": 32204,
-  "𐁈": 32282,
-  "𐁉": 32118,
-  "𐁊": 32213,
-  "𐁋": 32183,
-  "𐁌": 32311,
-  "𐁍": 32227,
-  "𐁐": 32298,
-  "𐁑": 32344,
-  "𐁒": 32134,
-  "𐁓": 32149,
-  "𐁔": 32228,
-  "𐁕": 32264,
-  "𐁖": 32180,
-  "𐁗": 32165,
-  "𐁘": 32101,
-  "𐁙": 32147,
-  "𐁚": 32173,
-  "𐁛": 32117,
-  "𐁜": 32100,
-  "𐁝": 32270,
-  "𐂀": 32267,
-  "𐂁": 32201,
-  "𐂂": 32291,
-  "𐂃": 32128,
-  "𐂄": 32285,
-  "𐂅": 32137,
-  "𐂆": 32235,
-  "𐂇": 32255,
-  "𐂈": 32199,
-  "𐂉": 32195,
-  "𐂊": 32191,
-  "𐂋": 32249,
-  "𐂌": 32177,
-  "𐂍": 32123,
-  "𐂎": 32302,
-  "𐂏": 32248,
-  "𐂐": 32192,
-  "𐂑": 32310,
-  "𐂒": 32109,
-  "𐂔": 32337,
-  "𐂕": 32276,
-  "𐂖": 32143,
-  "𐂗": 32166,
-  "𐂚": 32184,
-  "𐂛": 32338,
-  "𐂜": 32292,
-  "𐂝": 32215,
-  "𐂞": 32247,
-  "𐂟": 32240,
-  "𐂡": 32111,
-  "𐂢": 32168,
-  "𐂣": 32233,
-  "𐂥": 32125,
-  "𐂦": 32206,
-  "𐂧": 32141,
-  "𐂨": 32297,
-  "𐂩": 32241,
-  "𐂪": 32185,
-  "𐂫": 32343,
-  "𐂬": 32251,
-  "𐂭": 32284,
-  "𐂮": 32202,
-  "𐂯": 32140,
-  "𐂰": 32246,
-  "𐂱": 32336,
-  "𐂲": 32148,
-  "𐂳": 32256,
-  "𐂴": 32317,
-  "𐂵": 32104,
-  "𐂶": 32190,
-  "𐂷": 32329,
-  "𐂸": 32289,
-  "𐂹": 32323,
-  "𐂺": 32320,
-  "𐂻": 32280,
-  "𐂼": 32324,
-  "𐂽": 32150,
-  "𐂾": 32230,
-  "𐂿": 32296,
-  "𐃀": 32232,
-  "𐃁": 32129,
-  "𐃂": 32155,
-  "𐃄": 32163,
-  "𐃆": 32220,
-  "𐃇": 32133,
-  "𐃈": 32333,
-  "𐃉": 32203,
-  "𐃊": 32121,
-  "𐃌": 32120,
-  "𐃍": 32131,
-  "𐃎": 32196,
-  "𐃏": 32158,
-  "𐃐": 32113,
-  "𐃑": 32316,
-  "𐃓": 32258,
-  "𐃔": 32274,
-  "𐃕": 32217,
-  "𐃗": 32322,
-  "𐃙": 32223,
-  "𐃚": 32260,
-  "𐃜": 32342,
-  "𐃝": 32283,
-  "𐃞": 32305,
-  "𐃟": 32119,
-  "𐃠": 32193,
-  "𐃡": 32103,
-  "𐃢": 32253,
-  "𐃣": 32234,
-  "𐃤": 32312,
-  "𐃥": 32153,
-  "𐃦": 32146,
-  "𐃧": 32114,
-  "𐃨": 32209,
-  "𐃩": 32139,
-  "𐃪": 32335,
-  "𐃫": 32151,
-  "𐃬": 32332,
-  "𐃭": 32244,
-  "𐃮": 32279,
-  "𐃯": 32116,
-  "𐃰": 32197,
-  "𐃱": 32205,
-  "𐃲": 32207,
-  "𐃳": 32330,
-  "𐃴": 32181,
-  "𐃵": 32211,
-  "𐃶": 32236,
-  "𐃷": 32254,
-  "𐃸": 32308,
-  "𐃹": 32339,
-  "𐄈": 32334,
-  "𐄉": 32328,
-  "𐄊": 32262,
-  "𐄋": 32138,
-  "𐄌": 32178,
-  "𐄍": 32319,
-  "𐄎": 32269,
-  "𐄏": 32122,
-  "𐄐": 32170,
-  "𐄑": 32221,
-  "𐄒": 32275,
-  "𐄓": 32245,
-  "𐄔": 32174,
-  "𐄕": 32224,
-  "𐄖": 32182,
-  "𐄗": 32142,
-  "𐄘": 32250,
-  "𐄙": 32212,
-  "𐄚": 32294,
-  "𐄛": 32277,
-  "𐄜": 32242,
-  "𐄝": 32186,
-  "𐄞": 32132,
-  "𐄟": 32105,
-  "𐄡": 32112,
-  "𐄢": 32243,
-  "𐄣": 32273,
-  "𐄤": 32238,
-  "𐄥": 32145,
-  "𐄧": 32259,
-  "𐄪": 32327,
-  "𐄫": 32222,
-  "𐄷": 32229,
-  "𐄸": 32326,
-  "𐄹": 32176,
-  "𐄺": 32303,
-  "𐄼": 32307,
-  "𐄽": 32161,
-  "𐄾": 32252,
-  "𐄿": 32278
+  "×": 32424,
+  "ŋ": 32303,
+  "ʾ": 32444,
+  "⌈": 32402,
+  "𒀀": 32121,
+  "𒀁": 32445,
+  "𒀉": 32105,
+  "𒀊": 32131,
+  "𒀏": 32147,
+  "𒀕": 32415,
+  "𒀖": 32219,
+  "𒀘": 32479,
+  "𒀚": 32156,
+  "𒀜": 32183,
+  "𒀝": 32244,
+  "𒀞": 32176,
+  "𒀠": 32369,
+  "𒀩": 32323,
+  "𒀪": 32364,
+  "𒀫": 32228,
+  "𒀬": 32207,
+  "𒀭": 32405,
+  "𒀮": 32242,
+  "𒀯": 32123,
+  "𒀲": 32391,
+  "𒀳": 32221,
+  "𒀴": 32142,
+  "𒀸": 32220,
+  "𒀹": 32448,
+  "𒀾": 32417,
+  "𒀿": 32465,
+  "𒁀": 32175,
+  "𒁁": 32299,
+  "𒁃": 32494,
+  "𒁄": 32265,
+  "𒁆": 32483,
+  "𒁇": 32421,
+  "𒁈": 32288,
+  "𒁉": 32174,
+  "𒁍": 32387,
+  "𒁑": 32294,
+  "𒁓": 32326,
+  "𒁔": 32342,
+  "𒁕": 32172,
+  "𒁖": 32161,
+  "𒁜": 32447,
+  "𒁦": 32497,
+  "𒁮": 32141,
+  "𒁯": 32206,
+  "𒁰": 32453,
+  "𒁲": 32130,
+  "𒁳": 32280,
+  "𒁴": 32311,
+  "𒁵": 32367,
+  "𒁶": 32251,
+  "𒁷": 32232,
+  "𒁹": 32135,
+  "𒁺": 32297,
+  "𒁻": 32502,
+  "𒁼": 32396,
+  "𒁽": 32499,
+  "𒁾": 32340,
+  "𒂀": 32488,
+  "𒂁": 32198,
+  "𒂂": 32143,
+  "𒂃": 32506,
+  "𒂄": 32270,
+  "𒂅": 32108,
+  "𒂆": 32487,
+  "𒂇": 32249,
+  "𒂈": 32234,
+  "𒂉": 32259,
+  "𒂊": 32363,
+  "𒂍": 32245,
+  "𒂔": 32171,
+  "𒂕": 32380,
+  "𒂖": 32209,
+  "𒂗": 32160,
+  "𒂙": 32144,
+  "𒂞": 32456,
+  "𒂟": 32413,
+  "𒂠": 32378,
+  "𒂡": 32401,
+  "𒂤": 32468,
+  "𒂦": 32301,
+  "𒂬": 32508,
+  "𒂮": 32443,
+  "𒂵": 32252,
+  "𒂷": 32269,
+  "𒂼": 32185,
+  "𒃌": 32336,
+  "𒃞": 32429,
+  "𒃡": 32510,
+  "𒃢": 32461,
+  "𒃣": 32275,
+  "𒃮": 32129,
+  "𒃰": 32110,
+  "𒃲": 32103,
+  "𒃴": 32477,
+  "𒃵": 32267,
+  "𒃶": 32390,
+  "𒃷": 32100,
+  "𒃸": 32455,
+  "𒃻": 32273,
+  "𒃼": 32375,
+  "𒃽": 32307,
+  "𒃾": 32261,
+  "𒄀": 32149,
+  "𒄃": 32305,
+  "𒄄": 32481,
+  "𒄆": 32482,
+  "𒄇": 32309,
+  "𒄈": 32404,
+  "𒄉": 32476,
+  "𒄊": 32127,
+  "𒄋": 32423,
+  "𒄑": 32139,
+  "𒄒": 32411,
+  "𒄖": 32324,
+  "𒄗": 32117,
+  "𒄘": 32268,
+  "𒄙": 32434,
+  "𒄛": 32485,
+  "𒄝": 32509,
+  "𒄞": 32210,
+  "𒄠": 32355,
+  "𒄢": 32193,
+  "𒄣": 32203,
+  "𒄤": 32406,
+  "𒄥": 32113,
+  "𒄦": 32432,
+  "𒄧": 32189,
+  "𒄨": 32165,
+  "𒄩": 32202,
+  "𒄫": 32358,
+  "𒄬": 32140,
+  "𒄭": 32238,
+  "𒄯": 32361,
+  "𒄰": 32241,
+  "𒄲": 32503,
+  "𒄴": 32354,
+  "𒄵": 32322,
+  "𒄷": 32166,
+  "𒄸": 32505,
+  "𒄽": 32383,
+  "𒄾": 32151,
+  "𒄿": 32295,
+  "𒅀": 32395,
+  "𒅁": 32418,
+  "𒅂": 32351,
+  "𒅅": 32180,
+  "𒅆": 32337,
+  "𒅇": 32376,
+  "𒅈": 32359,
+  "𒅊": 32460,
+  "𒅋": 32145,
+  "𒅍": 32212,
+  "𒅎": 32385,
+  "𒅓": 32449,
+  "𒅔": 32237,
+  "𒅕": 32195,
+  "𒅖": 32122,
+  "𒅗": 32155,
+  "𒅘": 32360,
+  "𒅜": 32240,
+  "𒅢": 32439,
+  "𒅤": 32412,
+  "𒅥": 32200,
+  "𒅮": 32484,
+  "𒅴": 32253,
+  "𒅸": 32222,
+  "𒅻": 32116,
+  "𒅾": 32462,
+  "𒆃": 32315,
+  "𒆍": 32325,
+  "𒆏": 32178,
+  "𒆐": 32469,
+  "𒆑": 32126,
+  "𒆒": 32134,
+  "𒆓": 32491,
+  "𒆕": 32137,
+  "𒆗": 32343,
+  "𒆘": 32492,
+  "𒆚": 32422,
+  "𒆛": 32472,
+  "𒆜": 32302,
+  "𒆟": 32328,
+  "𒆠": 32398,
+  "𒆢": 32128,
+  "𒆤": 32196,
+  "𒆥": 32262,
+  "𒆦": 32266,
+  "𒆧": 32235,
+  "𒆪": 32146,
+  "𒆬": 32416,
+  "𒆭": 32349,
+  "𒆯": 32489,
+  "𒆰": 32285,
+  "𒆲": 32283,
+  "𒆳": 32257,
+  "𒆵": 32426,
+  "𒆷": 32194,
+  "𒆸": 32353,
+  "𒆹": 32304,
+  "𒇀": 32495,
+  "𒇅": 32112,
+  "𒇆": 32478,
+  "𒇇": 32260,
+  "𒇉": 32496,
+  "𒇒": 32441,
+  "𒇡": 32236,
+  "𒇥": 32230,
+  "𒇧": 32392,
+  "𒇬": 32500,
+  "𒇭": 32450,
+  "𒇯": 32227,
+  "𒇲": 32403,
+  "𒇳": 32427,
+  "𒇴": 32470,
+  "𒇵": 32473,
+  "𒇷": 32345,
+  "𒇸": 32119,
+  "𒇹": 32410,
+  "𒇺": 32339,
+  "𒇻": 32314,
+  "𒇼": 32504,
+  "𒇽": 32306,
+  "𒇿": 32329,
+  "𒈕": 32184,
+  "𒈖": 32438,
+  "𒈗": 32258,
+  "𒈛": 32286,
+  "𒈜": 32247,
+  "𒈝": 32352,
+  "𒈠": 32321,
+  "𒈢": 32454,
+  "𒈣": 32163,
+  "𒈤": 32168,
+  "𒈥": 32313,
+  "𒈦": 32188,
+  "𒈧": 32138,
+  "𒈨": 32366,
+  "𒈩": 32409,
+  "𒈪": 32223,
+  "𒈫": 32271,
+  "𒈬": 32208,
+  "𒈭": 32279,
+  "𒈮": 32397,
+  "𒈯": 32152,
+  "𒈲": 32381,
+  "𒈹": 32437,
+  "𒈻": 32486,
+  "𒈽": 32312,
+  "𒈾": 32310,
+  "𒈿": 32463,
+  "𒉀": 32498,
+  "𒉄": 32319,
+  "𒉅": 32278,
+  "𒉆": 32224,
+  "𒉇": 32132,
+  "𒉈": 32111,
+  "𒉋": 32106,
+  "𒉌": 32169,
+  "𒉎": 32250,
+  "𒉏": 32162,
+  "𒉐": 32436,
+  "𒉑": 32428,
+  "𒉒": 32474,
+  "𒉓": 32451,
+  "𒉘": 32107,
+  "𒉚": 32333,
+  "𒉠": 32320,
+  "𒉡": 32290,
+  "𒉢": 32493,
+  "𒉣": 32199,
+  "𒉪": 32159,
+  "𒉭": 32243,
+  "𒉺": 32308,
+  "𒉻": 32211,
+  "𒉼": 32512,
+  "𒉽": 32102,
+  "𒉾": 32226,
+  "𒉿": 32186,
+  "𒊊": 32104,
+  "𒊌": 32407,
+  "𒊍": 32300,
+  "𒊏": 32281,
+  "𒊐": 32382,
+  "𒊑": 32292,
+  "𒊒": 32341,
+  "𒊓": 32158,
+  "𒊕": 32217,
+  "𒊚": 32459,
+  "𒊨": 32471,
+  "𒊩": 32214,
+  "𒊬": 32389,
+  "𒊭": 32182,
+  "𒊮": 32239,
+  "𒊯": 32467,
+  "𒊷": 32457,
+  "𒊹": 32204,
+  "𒊺": 32191,
+  "𒊻": 32347,
+  "𒊿": 32350,
+  "𒋀": 32231,
+  "𒋁": 32187,
+  "𒋃": 32330,
+  "𒋆": 32296,
+  "𒋇": 32480,
+  "𒋋": 32501,
+  "𒋓": 32490,
+  "𒋗": 32408,
+  "𒋙": 32216,
+  "𒋚": 32357,
+  "𒋛": 32372,
+  "𒋜": 32373,
+  "𒋝": 32346,
+  "𒋞": 32327,
+  "𒋠": 32274,
+  "𒋡": 32173,
+  "𒋢": 32136,
+  "𒋤": 32414,
+  "𒋥": 32289,
+  "𒋦": 32379,
+  "𒋧": 32170,
+  "𒋩": 32374,
+  "𒋫": 32368,
+  "𒋬": 32133,
+  "𒋭": 32293,
+  "𒋰": 32388,
+  "𒋳": 32338,
+  "𒋺": 32466,
+  "𒋻": 32356,
+  "𒋼": 32335,
+  "𒋽": 32440,
+  "𒋾": 32154,
+  "𒌀": 32344,
+  "𒌁": 32287,
+  "𒌅": 32316,
+  "𒌆": 32365,
+  "𒌇": 32101,
+  "𒌈": 32318,
+  "𒌉": 32190,
+  "𒌋": 32177,
+  "𒌌": 32157,
+  "𒌍": 32150,
+  "𒌑": 32120,
+  "𒌒": 32377,
+  "𒌓": 32400,
+  "𒌔": 32148,
+  "𒌗": 32317,
+  "𒌜": 32218,
+  "𒌝": 32277,
+  "𒌢": 32256,
+  "𒌣": 32371,
+  "𒌤": 32464,
+  "𒌦": 32197,
+  "𒌨": 32386,
+  "𒌫": 32233,
+  "𒌴": 32446,
+  "𒌵": 32124,
+  "𒌶": 32442,
+  "𒌷": 32164,
+  "𒌺": 32167,
+  "𒍀": 32425,
+  "𒍂": 32435,
+  "𒍇": 32284,
+  "𒍍": 32264,
+  "𒍎": 32431,
+  "𒍏": 32109,
+  "𒍑": 32334,
+  "𒍒": 32118,
+  "𒍗": 32125,
+  "𒍚": 32430,
+  "𒍜": 32254,
+  "𒍝": 32229,
+  "𒍞": 32507,
+  "𒍠": 32205,
+  "𒍢": 32179,
+  "𒍣": 32394,
+  "𒍤": 32458,
+  "𒍥": 32255,
+  "𒍦": 32114,
+  "𒍨": 32213,
+  "𒍩": 32475,
+  "𒍪": 32201,
+  "𒍬": 32511,
+  "𒍮": 32291,
+  "𒐈": 32298,
+  "𒐉": 32192,
+  "𒐊": 32419,
+  "𒐋": 32181,
+  "𒐌": 32115,
+  "𒐍": 32282,
+  "𒐏": 32393,
+  "𒐐": 32246,
+  "𒐕": 32348,
+  "𒐖": 32248,
+  "𒐗": 32215,
+  "𒐚": 32433,
+  "𒐞": 32384,
+  "𒐻": 32331,
+  "𒑆": 32332,
+  "𒑏": 32420,
+  "𒑐": 32276,
+  "𒑑": 32370,
+  "𒑒": 32263,
+  "𒑔": 32153,
+  "𒑖": 32225,
+  "𒑚": 32362,
+  "𒑛": 32272,
+  "𒑱": 32452,
+  "𒑳": 32399
 }

config.json

@@ -1,12 +1,11 @@
 {
-  "_name_or_path": "/Users/lee/GitHub/results/GMY-T5Small/train_1/checkpoint-65840",
   "architectures": [
     "T5ForConditionalGeneration"
   ],
   "classifier_dropout": 0.0,
-  "d_ff": 2048,
+  "d_ff": 3072,
   "d_kv": 64,
-  "d_model": 512,
+  "d_model": 768,
   "decoder_start_token_id": 0,
   "dense_act_fn": "relu",
   "dropout_rate": 0.1,
@@ -18,9 +17,9 @@
   "layer_norm_epsilon": 1e-06,
   "model_type": "t5",
   "n_positions": 512,
-  "num_decoder_layers": 6,
-  "num_heads": 8,
-  "num_layers": 6,
+  "num_decoder_layers": 12,
+  "num_heads": 12,
+  "num_layers": 12,
   "output_past": true,
   "pad_token_id": 0,
   "relative_attention_max_distance": 128,
@@ -55,7 +54,7 @@
     }
   },
   "torch_dtype": "float32",
-  "transformers_version": "4.49.0",
+  "transformers_version": "4.50.3",
   "use_cache": true,
-  "vocab_size": 32345
+  "vocab_size": 32513
 }

generation_config.json

@@ -2,5 +2,5 @@
   "decoder_start_token_id": 0,
   "eos_token_id": 1,
   "pad_token_id": 0,
-  "transformers_version": "4.49.0"
+  "transformers_version": "4.50.3"
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fb7f5acc0e0a8661f5cfc17f326e922b7a39a32e92f43973c0b13a9bb42bed48
-size 242486312
+oid sha256:1b355c8952aa6b646e2ceaedda7950541d09bfbea54f24486af95e5935373c5a
+size 892827432

tokenizer.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4645dfb7555ec9f79235ff2195ef1775751eab22cd17671731d376a2edca89ab
-size 2466261
+oid sha256:8f1a24a97d5195455dd1fbf567d54e18c575a09f3b208ca957c2d319909b36be
+size 2496502

tokenizer_config.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:226471339474e6ac717acdd053bf20542f82bc2c6871f4cdc128bed5acd7516c
-size 62465
+oid sha256:3757494a511ac5c6706bf61d0b94c86e25b0c88c477932122634597858deb533
+size 91055

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:332ee1e3d81fcfdb71d17c2ea4ccecf168f3c2a602aad1e2850c9c02f1268160
-size 5560
+oid sha256:9643441c6cbb1024a83de6334eb0c291be2906f1780a037691fe0da36f334e10
+size 5496