Upload 14 files
Browse files- config.json +2 -3
- generation_config.json +1 -1
- onnx/decoder_model.onnx +3 -0
- onnx/decoder_model_merged.onnx +3 -0
- onnx/decoder_model_merged_quantized.onnx +3 -0
- onnx/decoder_model_quantized.onnx +3 -0
- onnx/decoder_with_past_model.onnx +3 -0
- onnx/decoder_with_past_model_quantized.onnx +3 -0
- onnx/encoder_model.onnx +3 -0
- onnx/encoder_model_quantized.onnx +3 -0
- quantize_config.json +139 -0
- special_tokens_map.json +21 -3
- tokenizer.json +202 -200
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "paust
|
| 3 |
"architectures": [
|
| 4 |
"T5ForConditionalGeneration"
|
| 5 |
],
|
|
@@ -25,8 +25,7 @@
|
|
| 25 |
"relative_attention_max_distance": 128,
|
| 26 |
"relative_attention_num_buckets": 32,
|
| 27 |
"tie_word_embeddings": false,
|
| 28 |
-
"
|
| 29 |
-
"transformers_version": "4.38.0.dev0",
|
| 30 |
"use_cache": true,
|
| 31 |
"vocab_size": 50358
|
| 32 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "kms7530/paust-t5-small-hatespeach",
|
| 3 |
"architectures": [
|
| 4 |
"T5ForConditionalGeneration"
|
| 5 |
],
|
|
|
|
| 25 |
"relative_attention_max_distance": 128,
|
| 26 |
"relative_attention_num_buckets": 32,
|
| 27 |
"tie_word_embeddings": false,
|
| 28 |
+
"transformers_version": "4.42.3",
|
|
|
|
| 29 |
"use_cache": true,
|
| 30 |
"vocab_size": 50358
|
| 31 |
}
|
generation_config.json
CHANGED
|
@@ -3,5 +3,5 @@
|
|
| 3 |
"decoder_start_token_id": 0,
|
| 4 |
"eos_token_id": 1,
|
| 5 |
"pad_token_id": 0,
|
| 6 |
-
"transformers_version": "4.
|
| 7 |
}
|
|
|
|
| 3 |
"decoder_start_token_id": 0,
|
| 4 |
"eos_token_id": 1,
|
| 5 |
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.42.3"
|
| 7 |
}
|
onnx/decoder_model.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:491e4a76a001bbfd617b73c927743c5c0de153e7e857b2412c948d45fd89ce9c
|
| 3 |
+
size 307223757
|
onnx/decoder_model_merged.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f182a42879355663b53d8549e464046dbbc26e00d12229bf33ba2fcbc73736c8
|
| 3 |
+
size 307454470
|
onnx/decoder_model_merged_quantized.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:419b9a3a827189d90f2baaa1ccb41bb2a1047579bf4eb3f9c2f63b51663bc460
|
| 3 |
+
size 78098252
|
onnx/decoder_model_quantized.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98cd799705e1aabe4da7f4a217b753cfb65768f51091ceea6961dc7665aa4623
|
| 3 |
+
size 77702074
|
onnx/decoder_with_past_model.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a56828dc5f6f611f4e84d97eb5c5e2cf92fc5cc0e50c59c679d5c0b463ad3914
|
| 3 |
+
size 294624064
|
onnx/decoder_with_past_model_quantized.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:52ccbe7c947a5f77aee456895874caad9ffbc828c6057b7e15cdbb1a49572841
|
| 3 |
+
size 74488068
|
onnx/encoder_model.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb134ad8ab7d91cf5a153ad447cbd970c2e8dc2cd97be0d57ba33e81b277b8e1
|
| 3 |
+
size 178791398
|
onnx/encoder_model_quantized.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c6c46c3170de7a1990d83f8270efec59d856ac8afe200dcb87eb1a7826ebefa
|
| 3 |
+
size 45092928
|
quantize_config.json
ADDED
|
@@ -0,0 +1,139 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"per_channel": true,
|
| 3 |
+
"reduce_range": true,
|
| 4 |
+
"per_model_config": {
|
| 5 |
+
"decoder_with_past_model": {
|
| 6 |
+
"op_types": [
|
| 7 |
+
"Concat",
|
| 8 |
+
"Tanh",
|
| 9 |
+
"Add",
|
| 10 |
+
"Unsqueeze",
|
| 11 |
+
"Log",
|
| 12 |
+
"Less",
|
| 13 |
+
"Gather",
|
| 14 |
+
"Tile",
|
| 15 |
+
"Min",
|
| 16 |
+
"ReduceMean",
|
| 17 |
+
"Transpose",
|
| 18 |
+
"Slice",
|
| 19 |
+
"Div",
|
| 20 |
+
"Mul",
|
| 21 |
+
"Sqrt",
|
| 22 |
+
"LessOrEqual",
|
| 23 |
+
"Reshape",
|
| 24 |
+
"Shape",
|
| 25 |
+
"Where",
|
| 26 |
+
"Constant",
|
| 27 |
+
"MatMul",
|
| 28 |
+
"Pow",
|
| 29 |
+
"Cast",
|
| 30 |
+
"ConstantOfShape",
|
| 31 |
+
"Sub",
|
| 32 |
+
"Neg",
|
| 33 |
+
"Softmax",
|
| 34 |
+
"Expand",
|
| 35 |
+
"Range"
|
| 36 |
+
],
|
| 37 |
+
"weight_type": "QInt8"
|
| 38 |
+
},
|
| 39 |
+
"decoder_model": {
|
| 40 |
+
"op_types": [
|
| 41 |
+
"Concat",
|
| 42 |
+
"Tanh",
|
| 43 |
+
"Add",
|
| 44 |
+
"Unsqueeze",
|
| 45 |
+
"Log",
|
| 46 |
+
"Less",
|
| 47 |
+
"Gather",
|
| 48 |
+
"Tile",
|
| 49 |
+
"Min",
|
| 50 |
+
"ReduceMean",
|
| 51 |
+
"Transpose",
|
| 52 |
+
"Div",
|
| 53 |
+
"Mul",
|
| 54 |
+
"Sqrt",
|
| 55 |
+
"LessOrEqual",
|
| 56 |
+
"Reshape",
|
| 57 |
+
"Shape",
|
| 58 |
+
"Where",
|
| 59 |
+
"Constant",
|
| 60 |
+
"MatMul",
|
| 61 |
+
"Pow",
|
| 62 |
+
"Cast",
|
| 63 |
+
"ConstantOfShape",
|
| 64 |
+
"Sub",
|
| 65 |
+
"Neg",
|
| 66 |
+
"Softmax",
|
| 67 |
+
"Expand",
|
| 68 |
+
"Range"
|
| 69 |
+
],
|
| 70 |
+
"weight_type": "QInt8"
|
| 71 |
+
},
|
| 72 |
+
"encoder_model": {
|
| 73 |
+
"op_types": [
|
| 74 |
+
"Concat",
|
| 75 |
+
"Abs",
|
| 76 |
+
"Tanh",
|
| 77 |
+
"Add",
|
| 78 |
+
"Unsqueeze",
|
| 79 |
+
"Log",
|
| 80 |
+
"Greater",
|
| 81 |
+
"Less",
|
| 82 |
+
"Gather",
|
| 83 |
+
"Min",
|
| 84 |
+
"ReduceMean",
|
| 85 |
+
"Transpose",
|
| 86 |
+
"Div",
|
| 87 |
+
"Mul",
|
| 88 |
+
"Sqrt",
|
| 89 |
+
"Reshape",
|
| 90 |
+
"Shape",
|
| 91 |
+
"Where",
|
| 92 |
+
"Constant",
|
| 93 |
+
"MatMul",
|
| 94 |
+
"Pow",
|
| 95 |
+
"Cast",
|
| 96 |
+
"ConstantOfShape",
|
| 97 |
+
"Sub",
|
| 98 |
+
"Softmax",
|
| 99 |
+
"Range"
|
| 100 |
+
],
|
| 101 |
+
"weight_type": "QInt8"
|
| 102 |
+
},
|
| 103 |
+
"decoder_model_merged": {
|
| 104 |
+
"op_types": [
|
| 105 |
+
"Concat",
|
| 106 |
+
"Tanh",
|
| 107 |
+
"Add",
|
| 108 |
+
"Unsqueeze",
|
| 109 |
+
"If",
|
| 110 |
+
"Log",
|
| 111 |
+
"Less",
|
| 112 |
+
"Gather",
|
| 113 |
+
"Tile",
|
| 114 |
+
"Min",
|
| 115 |
+
"ReduceMean",
|
| 116 |
+
"Transpose",
|
| 117 |
+
"Slice",
|
| 118 |
+
"Div",
|
| 119 |
+
"Mul",
|
| 120 |
+
"Sqrt",
|
| 121 |
+
"LessOrEqual",
|
| 122 |
+
"Reshape",
|
| 123 |
+
"Shape",
|
| 124 |
+
"Where",
|
| 125 |
+
"Constant",
|
| 126 |
+
"MatMul",
|
| 127 |
+
"Pow",
|
| 128 |
+
"Cast",
|
| 129 |
+
"ConstantOfShape",
|
| 130 |
+
"Sub",
|
| 131 |
+
"Neg",
|
| 132 |
+
"Softmax",
|
| 133 |
+
"Expand",
|
| 134 |
+
"Range"
|
| 135 |
+
],
|
| 136 |
+
"weight_type": "QInt8"
|
| 137 |
+
}
|
| 138 |
+
}
|
| 139 |
+
}
|
special_tokens_map.json
CHANGED
|
@@ -101,7 +101,25 @@
|
|
| 101 |
"<extra_id_98>",
|
| 102 |
"<extra_id_99>"
|
| 103 |
],
|
| 104 |
-
"eos_token":
|
| 105 |
-
|
| 106 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 107 |
}
|
|
|
|
| 101 |
"<extra_id_98>",
|
| 102 |
"<extra_id_99>"
|
| 103 |
],
|
| 104 |
+
"eos_token": {
|
| 105 |
+
"content": "</s>",
|
| 106 |
+
"lstrip": false,
|
| 107 |
+
"normalized": false,
|
| 108 |
+
"rstrip": false,
|
| 109 |
+
"single_word": false
|
| 110 |
+
},
|
| 111 |
+
"pad_token": {
|
| 112 |
+
"content": "<pad>",
|
| 113 |
+
"lstrip": false,
|
| 114 |
+
"normalized": false,
|
| 115 |
+
"rstrip": false,
|
| 116 |
+
"single_word": false
|
| 117 |
+
},
|
| 118 |
+
"unk_token": {
|
| 119 |
+
"content": "<pad>",
|
| 120 |
+
"lstrip": false,
|
| 121 |
+
"normalized": false,
|
| 122 |
+
"rstrip": false,
|
| 123 |
+
"single_word": false
|
| 124 |
+
}
|
| 125 |
}
|
tokenizer.json
CHANGED
|
@@ -32,8 +32,8 @@
|
|
| 32 |
"id": 50258,
|
| 33 |
"content": "<extra_id_0>",
|
| 34 |
"single_word": false,
|
| 35 |
-
"lstrip":
|
| 36 |
-
"rstrip":
|
| 37 |
"normalized": false,
|
| 38 |
"special": true
|
| 39 |
},
|
|
@@ -41,8 +41,8 @@
|
|
| 41 |
"id": 50259,
|
| 42 |
"content": "<extra_id_1>",
|
| 43 |
"single_word": false,
|
| 44 |
-
"lstrip":
|
| 45 |
-
"rstrip":
|
| 46 |
"normalized": false,
|
| 47 |
"special": true
|
| 48 |
},
|
|
@@ -50,8 +50,8 @@
|
|
| 50 |
"id": 50260,
|
| 51 |
"content": "<extra_id_2>",
|
| 52 |
"single_word": false,
|
| 53 |
-
"lstrip":
|
| 54 |
-
"rstrip":
|
| 55 |
"normalized": false,
|
| 56 |
"special": true
|
| 57 |
},
|
|
@@ -59,8 +59,8 @@
|
|
| 59 |
"id": 50261,
|
| 60 |
"content": "<extra_id_3>",
|
| 61 |
"single_word": false,
|
| 62 |
-
"lstrip":
|
| 63 |
-
"rstrip":
|
| 64 |
"normalized": false,
|
| 65 |
"special": true
|
| 66 |
},
|
|
@@ -68,8 +68,8 @@
|
|
| 68 |
"id": 50262,
|
| 69 |
"content": "<extra_id_4>",
|
| 70 |
"single_word": false,
|
| 71 |
-
"lstrip":
|
| 72 |
-
"rstrip":
|
| 73 |
"normalized": false,
|
| 74 |
"special": true
|
| 75 |
},
|
|
@@ -77,8 +77,8 @@
|
|
| 77 |
"id": 50263,
|
| 78 |
"content": "<extra_id_5>",
|
| 79 |
"single_word": false,
|
| 80 |
-
"lstrip":
|
| 81 |
-
"rstrip":
|
| 82 |
"normalized": false,
|
| 83 |
"special": true
|
| 84 |
},
|
|
@@ -86,8 +86,8 @@
|
|
| 86 |
"id": 50264,
|
| 87 |
"content": "<extra_id_6>",
|
| 88 |
"single_word": false,
|
| 89 |
-
"lstrip":
|
| 90 |
-
"rstrip":
|
| 91 |
"normalized": false,
|
| 92 |
"special": true
|
| 93 |
},
|
|
@@ -95,8 +95,8 @@
|
|
| 95 |
"id": 50265,
|
| 96 |
"content": "<extra_id_7>",
|
| 97 |
"single_word": false,
|
| 98 |
-
"lstrip":
|
| 99 |
-
"rstrip":
|
| 100 |
"normalized": false,
|
| 101 |
"special": true
|
| 102 |
},
|
|
@@ -104,8 +104,8 @@
|
|
| 104 |
"id": 50266,
|
| 105 |
"content": "<extra_id_8>",
|
| 106 |
"single_word": false,
|
| 107 |
-
"lstrip":
|
| 108 |
-
"rstrip":
|
| 109 |
"normalized": false,
|
| 110 |
"special": true
|
| 111 |
},
|
|
@@ -113,8 +113,8 @@
|
|
| 113 |
"id": 50267,
|
| 114 |
"content": "<extra_id_9>",
|
| 115 |
"single_word": false,
|
| 116 |
-
"lstrip":
|
| 117 |
-
"rstrip":
|
| 118 |
"normalized": false,
|
| 119 |
"special": true
|
| 120 |
},
|
|
@@ -122,8 +122,8 @@
|
|
| 122 |
"id": 50268,
|
| 123 |
"content": "<extra_id_10>",
|
| 124 |
"single_word": false,
|
| 125 |
-
"lstrip":
|
| 126 |
-
"rstrip":
|
| 127 |
"normalized": false,
|
| 128 |
"special": true
|
| 129 |
},
|
|
@@ -131,8 +131,8 @@
|
|
| 131 |
"id": 50269,
|
| 132 |
"content": "<extra_id_11>",
|
| 133 |
"single_word": false,
|
| 134 |
-
"lstrip":
|
| 135 |
-
"rstrip":
|
| 136 |
"normalized": false,
|
| 137 |
"special": true
|
| 138 |
},
|
|
@@ -140,8 +140,8 @@
|
|
| 140 |
"id": 50270,
|
| 141 |
"content": "<extra_id_12>",
|
| 142 |
"single_word": false,
|
| 143 |
-
"lstrip":
|
| 144 |
-
"rstrip":
|
| 145 |
"normalized": false,
|
| 146 |
"special": true
|
| 147 |
},
|
|
@@ -149,8 +149,8 @@
|
|
| 149 |
"id": 50271,
|
| 150 |
"content": "<extra_id_13>",
|
| 151 |
"single_word": false,
|
| 152 |
-
"lstrip":
|
| 153 |
-
"rstrip":
|
| 154 |
"normalized": false,
|
| 155 |
"special": true
|
| 156 |
},
|
|
@@ -158,8 +158,8 @@
|
|
| 158 |
"id": 50272,
|
| 159 |
"content": "<extra_id_14>",
|
| 160 |
"single_word": false,
|
| 161 |
-
"lstrip":
|
| 162 |
-
"rstrip":
|
| 163 |
"normalized": false,
|
| 164 |
"special": true
|
| 165 |
},
|
|
@@ -167,8 +167,8 @@
|
|
| 167 |
"id": 50273,
|
| 168 |
"content": "<extra_id_15>",
|
| 169 |
"single_word": false,
|
| 170 |
-
"lstrip":
|
| 171 |
-
"rstrip":
|
| 172 |
"normalized": false,
|
| 173 |
"special": true
|
| 174 |
},
|
|
@@ -176,8 +176,8 @@
|
|
| 176 |
"id": 50274,
|
| 177 |
"content": "<extra_id_16>",
|
| 178 |
"single_word": false,
|
| 179 |
-
"lstrip":
|
| 180 |
-
"rstrip":
|
| 181 |
"normalized": false,
|
| 182 |
"special": true
|
| 183 |
},
|
|
@@ -185,8 +185,8 @@
|
|
| 185 |
"id": 50275,
|
| 186 |
"content": "<extra_id_17>",
|
| 187 |
"single_word": false,
|
| 188 |
-
"lstrip":
|
| 189 |
-
"rstrip":
|
| 190 |
"normalized": false,
|
| 191 |
"special": true
|
| 192 |
},
|
|
@@ -194,8 +194,8 @@
|
|
| 194 |
"id": 50276,
|
| 195 |
"content": "<extra_id_18>",
|
| 196 |
"single_word": false,
|
| 197 |
-
"lstrip":
|
| 198 |
-
"rstrip":
|
| 199 |
"normalized": false,
|
| 200 |
"special": true
|
| 201 |
},
|
|
@@ -203,8 +203,8 @@
|
|
| 203 |
"id": 50277,
|
| 204 |
"content": "<extra_id_19>",
|
| 205 |
"single_word": false,
|
| 206 |
-
"lstrip":
|
| 207 |
-
"rstrip":
|
| 208 |
"normalized": false,
|
| 209 |
"special": true
|
| 210 |
},
|
|
@@ -212,8 +212,8 @@
|
|
| 212 |
"id": 50278,
|
| 213 |
"content": "<extra_id_20>",
|
| 214 |
"single_word": false,
|
| 215 |
-
"lstrip":
|
| 216 |
-
"rstrip":
|
| 217 |
"normalized": false,
|
| 218 |
"special": true
|
| 219 |
},
|
|
@@ -221,8 +221,8 @@
|
|
| 221 |
"id": 50279,
|
| 222 |
"content": "<extra_id_21>",
|
| 223 |
"single_word": false,
|
| 224 |
-
"lstrip":
|
| 225 |
-
"rstrip":
|
| 226 |
"normalized": false,
|
| 227 |
"special": true
|
| 228 |
},
|
|
@@ -230,8 +230,8 @@
|
|
| 230 |
"id": 50280,
|
| 231 |
"content": "<extra_id_22>",
|
| 232 |
"single_word": false,
|
| 233 |
-
"lstrip":
|
| 234 |
-
"rstrip":
|
| 235 |
"normalized": false,
|
| 236 |
"special": true
|
| 237 |
},
|
|
@@ -239,8 +239,8 @@
|
|
| 239 |
"id": 50281,
|
| 240 |
"content": "<extra_id_23>",
|
| 241 |
"single_word": false,
|
| 242 |
-
"lstrip":
|
| 243 |
-
"rstrip":
|
| 244 |
"normalized": false,
|
| 245 |
"special": true
|
| 246 |
},
|
|
@@ -248,8 +248,8 @@
|
|
| 248 |
"id": 50282,
|
| 249 |
"content": "<extra_id_24>",
|
| 250 |
"single_word": false,
|
| 251 |
-
"lstrip":
|
| 252 |
-
"rstrip":
|
| 253 |
"normalized": false,
|
| 254 |
"special": true
|
| 255 |
},
|
|
@@ -257,8 +257,8 @@
|
|
| 257 |
"id": 50283,
|
| 258 |
"content": "<extra_id_25>",
|
| 259 |
"single_word": false,
|
| 260 |
-
"lstrip":
|
| 261 |
-
"rstrip":
|
| 262 |
"normalized": false,
|
| 263 |
"special": true
|
| 264 |
},
|
|
@@ -266,8 +266,8 @@
|
|
| 266 |
"id": 50284,
|
| 267 |
"content": "<extra_id_26>",
|
| 268 |
"single_word": false,
|
| 269 |
-
"lstrip":
|
| 270 |
-
"rstrip":
|
| 271 |
"normalized": false,
|
| 272 |
"special": true
|
| 273 |
},
|
|
@@ -275,8 +275,8 @@
|
|
| 275 |
"id": 50285,
|
| 276 |
"content": "<extra_id_27>",
|
| 277 |
"single_word": false,
|
| 278 |
-
"lstrip":
|
| 279 |
-
"rstrip":
|
| 280 |
"normalized": false,
|
| 281 |
"special": true
|
| 282 |
},
|
|
@@ -284,8 +284,8 @@
|
|
| 284 |
"id": 50286,
|
| 285 |
"content": "<extra_id_28>",
|
| 286 |
"single_word": false,
|
| 287 |
-
"lstrip":
|
| 288 |
-
"rstrip":
|
| 289 |
"normalized": false,
|
| 290 |
"special": true
|
| 291 |
},
|
|
@@ -293,8 +293,8 @@
|
|
| 293 |
"id": 50287,
|
| 294 |
"content": "<extra_id_29>",
|
| 295 |
"single_word": false,
|
| 296 |
-
"lstrip":
|
| 297 |
-
"rstrip":
|
| 298 |
"normalized": false,
|
| 299 |
"special": true
|
| 300 |
},
|
|
@@ -302,8 +302,8 @@
|
|
| 302 |
"id": 50288,
|
| 303 |
"content": "<extra_id_30>",
|
| 304 |
"single_word": false,
|
| 305 |
-
"lstrip":
|
| 306 |
-
"rstrip":
|
| 307 |
"normalized": false,
|
| 308 |
"special": true
|
| 309 |
},
|
|
@@ -311,8 +311,8 @@
|
|
| 311 |
"id": 50289,
|
| 312 |
"content": "<extra_id_31>",
|
| 313 |
"single_word": false,
|
| 314 |
-
"lstrip":
|
| 315 |
-
"rstrip":
|
| 316 |
"normalized": false,
|
| 317 |
"special": true
|
| 318 |
},
|
|
@@ -320,8 +320,8 @@
|
|
| 320 |
"id": 50290,
|
| 321 |
"content": "<extra_id_32>",
|
| 322 |
"single_word": false,
|
| 323 |
-
"lstrip":
|
| 324 |
-
"rstrip":
|
| 325 |
"normalized": false,
|
| 326 |
"special": true
|
| 327 |
},
|
|
@@ -329,8 +329,8 @@
|
|
| 329 |
"id": 50291,
|
| 330 |
"content": "<extra_id_33>",
|
| 331 |
"single_word": false,
|
| 332 |
-
"lstrip":
|
| 333 |
-
"rstrip":
|
| 334 |
"normalized": false,
|
| 335 |
"special": true
|
| 336 |
},
|
|
@@ -338,8 +338,8 @@
|
|
| 338 |
"id": 50292,
|
| 339 |
"content": "<extra_id_34>",
|
| 340 |
"single_word": false,
|
| 341 |
-
"lstrip":
|
| 342 |
-
"rstrip":
|
| 343 |
"normalized": false,
|
| 344 |
"special": true
|
| 345 |
},
|
|
@@ -347,8 +347,8 @@
|
|
| 347 |
"id": 50293,
|
| 348 |
"content": "<extra_id_35>",
|
| 349 |
"single_word": false,
|
| 350 |
-
"lstrip":
|
| 351 |
-
"rstrip":
|
| 352 |
"normalized": false,
|
| 353 |
"special": true
|
| 354 |
},
|
|
@@ -356,8 +356,8 @@
|
|
| 356 |
"id": 50294,
|
| 357 |
"content": "<extra_id_36>",
|
| 358 |
"single_word": false,
|
| 359 |
-
"lstrip":
|
| 360 |
-
"rstrip":
|
| 361 |
"normalized": false,
|
| 362 |
"special": true
|
| 363 |
},
|
|
@@ -365,8 +365,8 @@
|
|
| 365 |
"id": 50295,
|
| 366 |
"content": "<extra_id_37>",
|
| 367 |
"single_word": false,
|
| 368 |
-
"lstrip":
|
| 369 |
-
"rstrip":
|
| 370 |
"normalized": false,
|
| 371 |
"special": true
|
| 372 |
},
|
|
@@ -374,8 +374,8 @@
|
|
| 374 |
"id": 50296,
|
| 375 |
"content": "<extra_id_38>",
|
| 376 |
"single_word": false,
|
| 377 |
-
"lstrip":
|
| 378 |
-
"rstrip":
|
| 379 |
"normalized": false,
|
| 380 |
"special": true
|
| 381 |
},
|
|
@@ -383,8 +383,8 @@
|
|
| 383 |
"id": 50297,
|
| 384 |
"content": "<extra_id_39>",
|
| 385 |
"single_word": false,
|
| 386 |
-
"lstrip":
|
| 387 |
-
"rstrip":
|
| 388 |
"normalized": false,
|
| 389 |
"special": true
|
| 390 |
},
|
|
@@ -392,8 +392,8 @@
|
|
| 392 |
"id": 50298,
|
| 393 |
"content": "<extra_id_40>",
|
| 394 |
"single_word": false,
|
| 395 |
-
"lstrip":
|
| 396 |
-
"rstrip":
|
| 397 |
"normalized": false,
|
| 398 |
"special": true
|
| 399 |
},
|
|
@@ -401,8 +401,8 @@
|
|
| 401 |
"id": 50299,
|
| 402 |
"content": "<extra_id_41>",
|
| 403 |
"single_word": false,
|
| 404 |
-
"lstrip":
|
| 405 |
-
"rstrip":
|
| 406 |
"normalized": false,
|
| 407 |
"special": true
|
| 408 |
},
|
|
@@ -410,8 +410,8 @@
|
|
| 410 |
"id": 50300,
|
| 411 |
"content": "<extra_id_42>",
|
| 412 |
"single_word": false,
|
| 413 |
-
"lstrip":
|
| 414 |
-
"rstrip":
|
| 415 |
"normalized": false,
|
| 416 |
"special": true
|
| 417 |
},
|
|
@@ -419,8 +419,8 @@
|
|
| 419 |
"id": 50301,
|
| 420 |
"content": "<extra_id_43>",
|
| 421 |
"single_word": false,
|
| 422 |
-
"lstrip":
|
| 423 |
-
"rstrip":
|
| 424 |
"normalized": false,
|
| 425 |
"special": true
|
| 426 |
},
|
|
@@ -428,8 +428,8 @@
|
|
| 428 |
"id": 50302,
|
| 429 |
"content": "<extra_id_44>",
|
| 430 |
"single_word": false,
|
| 431 |
-
"lstrip":
|
| 432 |
-
"rstrip":
|
| 433 |
"normalized": false,
|
| 434 |
"special": true
|
| 435 |
},
|
|
@@ -437,8 +437,8 @@
|
|
| 437 |
"id": 50303,
|
| 438 |
"content": "<extra_id_45>",
|
| 439 |
"single_word": false,
|
| 440 |
-
"lstrip":
|
| 441 |
-
"rstrip":
|
| 442 |
"normalized": false,
|
| 443 |
"special": true
|
| 444 |
},
|
|
@@ -446,8 +446,8 @@
|
|
| 446 |
"id": 50304,
|
| 447 |
"content": "<extra_id_46>",
|
| 448 |
"single_word": false,
|
| 449 |
-
"lstrip":
|
| 450 |
-
"rstrip":
|
| 451 |
"normalized": false,
|
| 452 |
"special": true
|
| 453 |
},
|
|
@@ -455,8 +455,8 @@
|
|
| 455 |
"id": 50305,
|
| 456 |
"content": "<extra_id_47>",
|
| 457 |
"single_word": false,
|
| 458 |
-
"lstrip":
|
| 459 |
-
"rstrip":
|
| 460 |
"normalized": false,
|
| 461 |
"special": true
|
| 462 |
},
|
|
@@ -464,8 +464,8 @@
|
|
| 464 |
"id": 50306,
|
| 465 |
"content": "<extra_id_48>",
|
| 466 |
"single_word": false,
|
| 467 |
-
"lstrip":
|
| 468 |
-
"rstrip":
|
| 469 |
"normalized": false,
|
| 470 |
"special": true
|
| 471 |
},
|
|
@@ -473,8 +473,8 @@
|
|
| 473 |
"id": 50307,
|
| 474 |
"content": "<extra_id_49>",
|
| 475 |
"single_word": false,
|
| 476 |
-
"lstrip":
|
| 477 |
-
"rstrip":
|
| 478 |
"normalized": false,
|
| 479 |
"special": true
|
| 480 |
},
|
|
@@ -482,8 +482,8 @@
|
|
| 482 |
"id": 50308,
|
| 483 |
"content": "<extra_id_50>",
|
| 484 |
"single_word": false,
|
| 485 |
-
"lstrip":
|
| 486 |
-
"rstrip":
|
| 487 |
"normalized": false,
|
| 488 |
"special": true
|
| 489 |
},
|
|
@@ -491,8 +491,8 @@
|
|
| 491 |
"id": 50309,
|
| 492 |
"content": "<extra_id_51>",
|
| 493 |
"single_word": false,
|
| 494 |
-
"lstrip":
|
| 495 |
-
"rstrip":
|
| 496 |
"normalized": false,
|
| 497 |
"special": true
|
| 498 |
},
|
|
@@ -500,8 +500,8 @@
|
|
| 500 |
"id": 50310,
|
| 501 |
"content": "<extra_id_52>",
|
| 502 |
"single_word": false,
|
| 503 |
-
"lstrip":
|
| 504 |
-
"rstrip":
|
| 505 |
"normalized": false,
|
| 506 |
"special": true
|
| 507 |
},
|
|
@@ -509,8 +509,8 @@
|
|
| 509 |
"id": 50311,
|
| 510 |
"content": "<extra_id_53>",
|
| 511 |
"single_word": false,
|
| 512 |
-
"lstrip":
|
| 513 |
-
"rstrip":
|
| 514 |
"normalized": false,
|
| 515 |
"special": true
|
| 516 |
},
|
|
@@ -518,8 +518,8 @@
|
|
| 518 |
"id": 50312,
|
| 519 |
"content": "<extra_id_54>",
|
| 520 |
"single_word": false,
|
| 521 |
-
"lstrip":
|
| 522 |
-
"rstrip":
|
| 523 |
"normalized": false,
|
| 524 |
"special": true
|
| 525 |
},
|
|
@@ -527,8 +527,8 @@
|
|
| 527 |
"id": 50313,
|
| 528 |
"content": "<extra_id_55>",
|
| 529 |
"single_word": false,
|
| 530 |
-
"lstrip":
|
| 531 |
-
"rstrip":
|
| 532 |
"normalized": false,
|
| 533 |
"special": true
|
| 534 |
},
|
|
@@ -536,8 +536,8 @@
|
|
| 536 |
"id": 50314,
|
| 537 |
"content": "<extra_id_56>",
|
| 538 |
"single_word": false,
|
| 539 |
-
"lstrip":
|
| 540 |
-
"rstrip":
|
| 541 |
"normalized": false,
|
| 542 |
"special": true
|
| 543 |
},
|
|
@@ -545,8 +545,8 @@
|
|
| 545 |
"id": 50315,
|
| 546 |
"content": "<extra_id_57>",
|
| 547 |
"single_word": false,
|
| 548 |
-
"lstrip":
|
| 549 |
-
"rstrip":
|
| 550 |
"normalized": false,
|
| 551 |
"special": true
|
| 552 |
},
|
|
@@ -554,8 +554,8 @@
|
|
| 554 |
"id": 50316,
|
| 555 |
"content": "<extra_id_58>",
|
| 556 |
"single_word": false,
|
| 557 |
-
"lstrip":
|
| 558 |
-
"rstrip":
|
| 559 |
"normalized": false,
|
| 560 |
"special": true
|
| 561 |
},
|
|
@@ -563,8 +563,8 @@
|
|
| 563 |
"id": 50317,
|
| 564 |
"content": "<extra_id_59>",
|
| 565 |
"single_word": false,
|
| 566 |
-
"lstrip":
|
| 567 |
-
"rstrip":
|
| 568 |
"normalized": false,
|
| 569 |
"special": true
|
| 570 |
},
|
|
@@ -572,8 +572,8 @@
|
|
| 572 |
"id": 50318,
|
| 573 |
"content": "<extra_id_60>",
|
| 574 |
"single_word": false,
|
| 575 |
-
"lstrip":
|
| 576 |
-
"rstrip":
|
| 577 |
"normalized": false,
|
| 578 |
"special": true
|
| 579 |
},
|
|
@@ -581,8 +581,8 @@
|
|
| 581 |
"id": 50319,
|
| 582 |
"content": "<extra_id_61>",
|
| 583 |
"single_word": false,
|
| 584 |
-
"lstrip":
|
| 585 |
-
"rstrip":
|
| 586 |
"normalized": false,
|
| 587 |
"special": true
|
| 588 |
},
|
|
@@ -590,8 +590,8 @@
|
|
| 590 |
"id": 50320,
|
| 591 |
"content": "<extra_id_62>",
|
| 592 |
"single_word": false,
|
| 593 |
-
"lstrip":
|
| 594 |
-
"rstrip":
|
| 595 |
"normalized": false,
|
| 596 |
"special": true
|
| 597 |
},
|
|
@@ -599,8 +599,8 @@
|
|
| 599 |
"id": 50321,
|
| 600 |
"content": "<extra_id_63>",
|
| 601 |
"single_word": false,
|
| 602 |
-
"lstrip":
|
| 603 |
-
"rstrip":
|
| 604 |
"normalized": false,
|
| 605 |
"special": true
|
| 606 |
},
|
|
@@ -608,8 +608,8 @@
|
|
| 608 |
"id": 50322,
|
| 609 |
"content": "<extra_id_64>",
|
| 610 |
"single_word": false,
|
| 611 |
-
"lstrip":
|
| 612 |
-
"rstrip":
|
| 613 |
"normalized": false,
|
| 614 |
"special": true
|
| 615 |
},
|
|
@@ -617,8 +617,8 @@
|
|
| 617 |
"id": 50323,
|
| 618 |
"content": "<extra_id_65>",
|
| 619 |
"single_word": false,
|
| 620 |
-
"lstrip":
|
| 621 |
-
"rstrip":
|
| 622 |
"normalized": false,
|
| 623 |
"special": true
|
| 624 |
},
|
|
@@ -626,8 +626,8 @@
|
|
| 626 |
"id": 50324,
|
| 627 |
"content": "<extra_id_66>",
|
| 628 |
"single_word": false,
|
| 629 |
-
"lstrip":
|
| 630 |
-
"rstrip":
|
| 631 |
"normalized": false,
|
| 632 |
"special": true
|
| 633 |
},
|
|
@@ -635,8 +635,8 @@
|
|
| 635 |
"id": 50325,
|
| 636 |
"content": "<extra_id_67>",
|
| 637 |
"single_word": false,
|
| 638 |
-
"lstrip":
|
| 639 |
-
"rstrip":
|
| 640 |
"normalized": false,
|
| 641 |
"special": true
|
| 642 |
},
|
|
@@ -644,8 +644,8 @@
|
|
| 644 |
"id": 50326,
|
| 645 |
"content": "<extra_id_68>",
|
| 646 |
"single_word": false,
|
| 647 |
-
"lstrip":
|
| 648 |
-
"rstrip":
|
| 649 |
"normalized": false,
|
| 650 |
"special": true
|
| 651 |
},
|
|
@@ -653,8 +653,8 @@
|
|
| 653 |
"id": 50327,
|
| 654 |
"content": "<extra_id_69>",
|
| 655 |
"single_word": false,
|
| 656 |
-
"lstrip":
|
| 657 |
-
"rstrip":
|
| 658 |
"normalized": false,
|
| 659 |
"special": true
|
| 660 |
},
|
|
@@ -662,8 +662,8 @@
|
|
| 662 |
"id": 50328,
|
| 663 |
"content": "<extra_id_70>",
|
| 664 |
"single_word": false,
|
| 665 |
-
"lstrip":
|
| 666 |
-
"rstrip":
|
| 667 |
"normalized": false,
|
| 668 |
"special": true
|
| 669 |
},
|
|
@@ -671,8 +671,8 @@
|
|
| 671 |
"id": 50329,
|
| 672 |
"content": "<extra_id_71>",
|
| 673 |
"single_word": false,
|
| 674 |
-
"lstrip":
|
| 675 |
-
"rstrip":
|
| 676 |
"normalized": false,
|
| 677 |
"special": true
|
| 678 |
},
|
|
@@ -680,8 +680,8 @@
|
|
| 680 |
"id": 50330,
|
| 681 |
"content": "<extra_id_72>",
|
| 682 |
"single_word": false,
|
| 683 |
-
"lstrip":
|
| 684 |
-
"rstrip":
|
| 685 |
"normalized": false,
|
| 686 |
"special": true
|
| 687 |
},
|
|
@@ -689,8 +689,8 @@
|
|
| 689 |
"id": 50331,
|
| 690 |
"content": "<extra_id_73>",
|
| 691 |
"single_word": false,
|
| 692 |
-
"lstrip":
|
| 693 |
-
"rstrip":
|
| 694 |
"normalized": false,
|
| 695 |
"special": true
|
| 696 |
},
|
|
@@ -698,8 +698,8 @@
|
|
| 698 |
"id": 50332,
|
| 699 |
"content": "<extra_id_74>",
|
| 700 |
"single_word": false,
|
| 701 |
-
"lstrip":
|
| 702 |
-
"rstrip":
|
| 703 |
"normalized": false,
|
| 704 |
"special": true
|
| 705 |
},
|
|
@@ -707,8 +707,8 @@
|
|
| 707 |
"id": 50333,
|
| 708 |
"content": "<extra_id_75>",
|
| 709 |
"single_word": false,
|
| 710 |
-
"lstrip":
|
| 711 |
-
"rstrip":
|
| 712 |
"normalized": false,
|
| 713 |
"special": true
|
| 714 |
},
|
|
@@ -716,8 +716,8 @@
|
|
| 716 |
"id": 50334,
|
| 717 |
"content": "<extra_id_76>",
|
| 718 |
"single_word": false,
|
| 719 |
-
"lstrip":
|
| 720 |
-
"rstrip":
|
| 721 |
"normalized": false,
|
| 722 |
"special": true
|
| 723 |
},
|
|
@@ -725,8 +725,8 @@
|
|
| 725 |
"id": 50335,
|
| 726 |
"content": "<extra_id_77>",
|
| 727 |
"single_word": false,
|
| 728 |
-
"lstrip":
|
| 729 |
-
"rstrip":
|
| 730 |
"normalized": false,
|
| 731 |
"special": true
|
| 732 |
},
|
|
@@ -734,8 +734,8 @@
|
|
| 734 |
"id": 50336,
|
| 735 |
"content": "<extra_id_78>",
|
| 736 |
"single_word": false,
|
| 737 |
-
"lstrip":
|
| 738 |
-
"rstrip":
|
| 739 |
"normalized": false,
|
| 740 |
"special": true
|
| 741 |
},
|
|
@@ -743,8 +743,8 @@
|
|
| 743 |
"id": 50337,
|
| 744 |
"content": "<extra_id_79>",
|
| 745 |
"single_word": false,
|
| 746 |
-
"lstrip":
|
| 747 |
-
"rstrip":
|
| 748 |
"normalized": false,
|
| 749 |
"special": true
|
| 750 |
},
|
|
@@ -752,8 +752,8 @@
|
|
| 752 |
"id": 50338,
|
| 753 |
"content": "<extra_id_80>",
|
| 754 |
"single_word": false,
|
| 755 |
-
"lstrip":
|
| 756 |
-
"rstrip":
|
| 757 |
"normalized": false,
|
| 758 |
"special": true
|
| 759 |
},
|
|
@@ -761,8 +761,8 @@
|
|
| 761 |
"id": 50339,
|
| 762 |
"content": "<extra_id_81>",
|
| 763 |
"single_word": false,
|
| 764 |
-
"lstrip":
|
| 765 |
-
"rstrip":
|
| 766 |
"normalized": false,
|
| 767 |
"special": true
|
| 768 |
},
|
|
@@ -770,8 +770,8 @@
|
|
| 770 |
"id": 50340,
|
| 771 |
"content": "<extra_id_82>",
|
| 772 |
"single_word": false,
|
| 773 |
-
"lstrip":
|
| 774 |
-
"rstrip":
|
| 775 |
"normalized": false,
|
| 776 |
"special": true
|
| 777 |
},
|
|
@@ -779,8 +779,8 @@
|
|
| 779 |
"id": 50341,
|
| 780 |
"content": "<extra_id_83>",
|
| 781 |
"single_word": false,
|
| 782 |
-
"lstrip":
|
| 783 |
-
"rstrip":
|
| 784 |
"normalized": false,
|
| 785 |
"special": true
|
| 786 |
},
|
|
@@ -788,8 +788,8 @@
|
|
| 788 |
"id": 50342,
|
| 789 |
"content": "<extra_id_84>",
|
| 790 |
"single_word": false,
|
| 791 |
-
"lstrip":
|
| 792 |
-
"rstrip":
|
| 793 |
"normalized": false,
|
| 794 |
"special": true
|
| 795 |
},
|
|
@@ -797,8 +797,8 @@
|
|
| 797 |
"id": 50343,
|
| 798 |
"content": "<extra_id_85>",
|
| 799 |
"single_word": false,
|
| 800 |
-
"lstrip":
|
| 801 |
-
"rstrip":
|
| 802 |
"normalized": false,
|
| 803 |
"special": true
|
| 804 |
},
|
|
@@ -806,8 +806,8 @@
|
|
| 806 |
"id": 50344,
|
| 807 |
"content": "<extra_id_86>",
|
| 808 |
"single_word": false,
|
| 809 |
-
"lstrip":
|
| 810 |
-
"rstrip":
|
| 811 |
"normalized": false,
|
| 812 |
"special": true
|
| 813 |
},
|
|
@@ -815,8 +815,8 @@
|
|
| 815 |
"id": 50345,
|
| 816 |
"content": "<extra_id_87>",
|
| 817 |
"single_word": false,
|
| 818 |
-
"lstrip":
|
| 819 |
-
"rstrip":
|
| 820 |
"normalized": false,
|
| 821 |
"special": true
|
| 822 |
},
|
|
@@ -824,8 +824,8 @@
|
|
| 824 |
"id": 50346,
|
| 825 |
"content": "<extra_id_88>",
|
| 826 |
"single_word": false,
|
| 827 |
-
"lstrip":
|
| 828 |
-
"rstrip":
|
| 829 |
"normalized": false,
|
| 830 |
"special": true
|
| 831 |
},
|
|
@@ -833,8 +833,8 @@
|
|
| 833 |
"id": 50347,
|
| 834 |
"content": "<extra_id_89>",
|
| 835 |
"single_word": false,
|
| 836 |
-
"lstrip":
|
| 837 |
-
"rstrip":
|
| 838 |
"normalized": false,
|
| 839 |
"special": true
|
| 840 |
},
|
|
@@ -842,8 +842,8 @@
|
|
| 842 |
"id": 50348,
|
| 843 |
"content": "<extra_id_90>",
|
| 844 |
"single_word": false,
|
| 845 |
-
"lstrip":
|
| 846 |
-
"rstrip":
|
| 847 |
"normalized": false,
|
| 848 |
"special": true
|
| 849 |
},
|
|
@@ -851,8 +851,8 @@
|
|
| 851 |
"id": 50349,
|
| 852 |
"content": "<extra_id_91>",
|
| 853 |
"single_word": false,
|
| 854 |
-
"lstrip":
|
| 855 |
-
"rstrip":
|
| 856 |
"normalized": false,
|
| 857 |
"special": true
|
| 858 |
},
|
|
@@ -860,8 +860,8 @@
|
|
| 860 |
"id": 50350,
|
| 861 |
"content": "<extra_id_92>",
|
| 862 |
"single_word": false,
|
| 863 |
-
"lstrip":
|
| 864 |
-
"rstrip":
|
| 865 |
"normalized": false,
|
| 866 |
"special": true
|
| 867 |
},
|
|
@@ -869,8 +869,8 @@
|
|
| 869 |
"id": 50351,
|
| 870 |
"content": "<extra_id_93>",
|
| 871 |
"single_word": false,
|
| 872 |
-
"lstrip":
|
| 873 |
-
"rstrip":
|
| 874 |
"normalized": false,
|
| 875 |
"special": true
|
| 876 |
},
|
|
@@ -878,8 +878,8 @@
|
|
| 878 |
"id": 50352,
|
| 879 |
"content": "<extra_id_94>",
|
| 880 |
"single_word": false,
|
| 881 |
-
"lstrip":
|
| 882 |
-
"rstrip":
|
| 883 |
"normalized": false,
|
| 884 |
"special": true
|
| 885 |
},
|
|
@@ -887,8 +887,8 @@
|
|
| 887 |
"id": 50353,
|
| 888 |
"content": "<extra_id_95>",
|
| 889 |
"single_word": false,
|
| 890 |
-
"lstrip":
|
| 891 |
-
"rstrip":
|
| 892 |
"normalized": false,
|
| 893 |
"special": true
|
| 894 |
},
|
|
@@ -896,8 +896,8 @@
|
|
| 896 |
"id": 50354,
|
| 897 |
"content": "<extra_id_96>",
|
| 898 |
"single_word": false,
|
| 899 |
-
"lstrip":
|
| 900 |
-
"rstrip":
|
| 901 |
"normalized": false,
|
| 902 |
"special": true
|
| 903 |
},
|
|
@@ -905,8 +905,8 @@
|
|
| 905 |
"id": 50355,
|
| 906 |
"content": "<extra_id_97>",
|
| 907 |
"single_word": false,
|
| 908 |
-
"lstrip":
|
| 909 |
-
"rstrip":
|
| 910 |
"normalized": false,
|
| 911 |
"special": true
|
| 912 |
},
|
|
@@ -914,8 +914,8 @@
|
|
| 914 |
"id": 50356,
|
| 915 |
"content": "<extra_id_98>",
|
| 916 |
"single_word": false,
|
| 917 |
-
"lstrip":
|
| 918 |
-
"rstrip":
|
| 919 |
"normalized": false,
|
| 920 |
"special": true
|
| 921 |
},
|
|
@@ -923,8 +923,8 @@
|
|
| 923 |
"id": 50357,
|
| 924 |
"content": "<extra_id_99>",
|
| 925 |
"single_word": false,
|
| 926 |
-
"lstrip":
|
| 927 |
-
"rstrip":
|
| 928 |
"normalized": false,
|
| 929 |
"special": true
|
| 930 |
}
|
|
@@ -999,6 +999,8 @@
|
|
| 999 |
"continuing_subword_prefix": "",
|
| 1000 |
"end_of_word_suffix": "",
|
| 1001 |
"fuse_unk": false,
|
|
|
|
|
|
|
| 1002 |
"vocab": {
|
| 1003 |
"<pad>": 0,
|
| 1004 |
"</s>": 1,
|
|
|
|
| 32 |
"id": 50258,
|
| 33 |
"content": "<extra_id_0>",
|
| 34 |
"single_word": false,
|
| 35 |
+
"lstrip": true,
|
| 36 |
+
"rstrip": true,
|
| 37 |
"normalized": false,
|
| 38 |
"special": true
|
| 39 |
},
|
|
|
|
| 41 |
"id": 50259,
|
| 42 |
"content": "<extra_id_1>",
|
| 43 |
"single_word": false,
|
| 44 |
+
"lstrip": true,
|
| 45 |
+
"rstrip": true,
|
| 46 |
"normalized": false,
|
| 47 |
"special": true
|
| 48 |
},
|
|
|
|
| 50 |
"id": 50260,
|
| 51 |
"content": "<extra_id_2>",
|
| 52 |
"single_word": false,
|
| 53 |
+
"lstrip": true,
|
| 54 |
+
"rstrip": true,
|
| 55 |
"normalized": false,
|
| 56 |
"special": true
|
| 57 |
},
|
|
|
|
| 59 |
"id": 50261,
|
| 60 |
"content": "<extra_id_3>",
|
| 61 |
"single_word": false,
|
| 62 |
+
"lstrip": true,
|
| 63 |
+
"rstrip": true,
|
| 64 |
"normalized": false,
|
| 65 |
"special": true
|
| 66 |
},
|
|
|
|
| 68 |
"id": 50262,
|
| 69 |
"content": "<extra_id_4>",
|
| 70 |
"single_word": false,
|
| 71 |
+
"lstrip": true,
|
| 72 |
+
"rstrip": true,
|
| 73 |
"normalized": false,
|
| 74 |
"special": true
|
| 75 |
},
|
|
|
|
| 77 |
"id": 50263,
|
| 78 |
"content": "<extra_id_5>",
|
| 79 |
"single_word": false,
|
| 80 |
+
"lstrip": true,
|
| 81 |
+
"rstrip": true,
|
| 82 |
"normalized": false,
|
| 83 |
"special": true
|
| 84 |
},
|
|
|
|
| 86 |
"id": 50264,
|
| 87 |
"content": "<extra_id_6>",
|
| 88 |
"single_word": false,
|
| 89 |
+
"lstrip": true,
|
| 90 |
+
"rstrip": true,
|
| 91 |
"normalized": false,
|
| 92 |
"special": true
|
| 93 |
},
|
|
|
|
| 95 |
"id": 50265,
|
| 96 |
"content": "<extra_id_7>",
|
| 97 |
"single_word": false,
|
| 98 |
+
"lstrip": true,
|
| 99 |
+
"rstrip": true,
|
| 100 |
"normalized": false,
|
| 101 |
"special": true
|
| 102 |
},
|
|
|
|
| 104 |
"id": 50266,
|
| 105 |
"content": "<extra_id_8>",
|
| 106 |
"single_word": false,
|
| 107 |
+
"lstrip": true,
|
| 108 |
+
"rstrip": true,
|
| 109 |
"normalized": false,
|
| 110 |
"special": true
|
| 111 |
},
|
|
|
|
| 113 |
"id": 50267,
|
| 114 |
"content": "<extra_id_9>",
|
| 115 |
"single_word": false,
|
| 116 |
+
"lstrip": true,
|
| 117 |
+
"rstrip": true,
|
| 118 |
"normalized": false,
|
| 119 |
"special": true
|
| 120 |
},
|
|
|
|
| 122 |
"id": 50268,
|
| 123 |
"content": "<extra_id_10>",
|
| 124 |
"single_word": false,
|
| 125 |
+
"lstrip": true,
|
| 126 |
+
"rstrip": true,
|
| 127 |
"normalized": false,
|
| 128 |
"special": true
|
| 129 |
},
|
|
|
|
| 131 |
"id": 50269,
|
| 132 |
"content": "<extra_id_11>",
|
| 133 |
"single_word": false,
|
| 134 |
+
"lstrip": true,
|
| 135 |
+
"rstrip": true,
|
| 136 |
"normalized": false,
|
| 137 |
"special": true
|
| 138 |
},
|
|
|
|
| 140 |
"id": 50270,
|
| 141 |
"content": "<extra_id_12>",
|
| 142 |
"single_word": false,
|
| 143 |
+
"lstrip": true,
|
| 144 |
+
"rstrip": true,
|
| 145 |
"normalized": false,
|
| 146 |
"special": true
|
| 147 |
},
|
|
|
|
| 149 |
"id": 50271,
|
| 150 |
"content": "<extra_id_13>",
|
| 151 |
"single_word": false,
|
| 152 |
+
"lstrip": true,
|
| 153 |
+
"rstrip": true,
|
| 154 |
"normalized": false,
|
| 155 |
"special": true
|
| 156 |
},
|
|
|
|
| 158 |
"id": 50272,
|
| 159 |
"content": "<extra_id_14>",
|
| 160 |
"single_word": false,
|
| 161 |
+
"lstrip": true,
|
| 162 |
+
"rstrip": true,
|
| 163 |
"normalized": false,
|
| 164 |
"special": true
|
| 165 |
},
|
|
|
|
| 167 |
"id": 50273,
|
| 168 |
"content": "<extra_id_15>",
|
| 169 |
"single_word": false,
|
| 170 |
+
"lstrip": true,
|
| 171 |
+
"rstrip": true,
|
| 172 |
"normalized": false,
|
| 173 |
"special": true
|
| 174 |
},
|
|
|
|
| 176 |
"id": 50274,
|
| 177 |
"content": "<extra_id_16>",
|
| 178 |
"single_word": false,
|
| 179 |
+
"lstrip": true,
|
| 180 |
+
"rstrip": true,
|
| 181 |
"normalized": false,
|
| 182 |
"special": true
|
| 183 |
},
|
|
|
|
| 185 |
"id": 50275,
|
| 186 |
"content": "<extra_id_17>",
|
| 187 |
"single_word": false,
|
| 188 |
+
"lstrip": true,
|
| 189 |
+
"rstrip": true,
|
| 190 |
"normalized": false,
|
| 191 |
"special": true
|
| 192 |
},
|
|
|
|
| 194 |
"id": 50276,
|
| 195 |
"content": "<extra_id_18>",
|
| 196 |
"single_word": false,
|
| 197 |
+
"lstrip": true,
|
| 198 |
+
"rstrip": true,
|
| 199 |
"normalized": false,
|
| 200 |
"special": true
|
| 201 |
},
|
|
|
|
| 203 |
"id": 50277,
|
| 204 |
"content": "<extra_id_19>",
|
| 205 |
"single_word": false,
|
| 206 |
+
"lstrip": true,
|
| 207 |
+
"rstrip": true,
|
| 208 |
"normalized": false,
|
| 209 |
"special": true
|
| 210 |
},
|
|
|
|
| 212 |
"id": 50278,
|
| 213 |
"content": "<extra_id_20>",
|
| 214 |
"single_word": false,
|
| 215 |
+
"lstrip": true,
|
| 216 |
+
"rstrip": true,
|
| 217 |
"normalized": false,
|
| 218 |
"special": true
|
| 219 |
},
|
|
|
|
| 221 |
"id": 50279,
|
| 222 |
"content": "<extra_id_21>",
|
| 223 |
"single_word": false,
|
| 224 |
+
"lstrip": true,
|
| 225 |
+
"rstrip": true,
|
| 226 |
"normalized": false,
|
| 227 |
"special": true
|
| 228 |
},
|
|
|
|
| 230 |
"id": 50280,
|
| 231 |
"content": "<extra_id_22>",
|
| 232 |
"single_word": false,
|
| 233 |
+
"lstrip": true,
|
| 234 |
+
"rstrip": true,
|
| 235 |
"normalized": false,
|
| 236 |
"special": true
|
| 237 |
},
|
|
|
|
| 239 |
"id": 50281,
|
| 240 |
"content": "<extra_id_23>",
|
| 241 |
"single_word": false,
|
| 242 |
+
"lstrip": true,
|
| 243 |
+
"rstrip": true,
|
| 244 |
"normalized": false,
|
| 245 |
"special": true
|
| 246 |
},
|
|
|
|
| 248 |
"id": 50282,
|
| 249 |
"content": "<extra_id_24>",
|
| 250 |
"single_word": false,
|
| 251 |
+
"lstrip": true,
|
| 252 |
+
"rstrip": true,
|
| 253 |
"normalized": false,
|
| 254 |
"special": true
|
| 255 |
},
|
|
|
|
| 257 |
"id": 50283,
|
| 258 |
"content": "<extra_id_25>",
|
| 259 |
"single_word": false,
|
| 260 |
+
"lstrip": true,
|
| 261 |
+
"rstrip": true,
|
| 262 |
"normalized": false,
|
| 263 |
"special": true
|
| 264 |
},
|
|
|
|
| 266 |
"id": 50284,
|
| 267 |
"content": "<extra_id_26>",
|
| 268 |
"single_word": false,
|
| 269 |
+
"lstrip": true,
|
| 270 |
+
"rstrip": true,
|
| 271 |
"normalized": false,
|
| 272 |
"special": true
|
| 273 |
},
|
|
|
|
| 275 |
"id": 50285,
|
| 276 |
"content": "<extra_id_27>",
|
| 277 |
"single_word": false,
|
| 278 |
+
"lstrip": true,
|
| 279 |
+
"rstrip": true,
|
| 280 |
"normalized": false,
|
| 281 |
"special": true
|
| 282 |
},
|
|
|
|
| 284 |
"id": 50286,
|
| 285 |
"content": "<extra_id_28>",
|
| 286 |
"single_word": false,
|
| 287 |
+
"lstrip": true,
|
| 288 |
+
"rstrip": true,
|
| 289 |
"normalized": false,
|
| 290 |
"special": true
|
| 291 |
},
|
|
|
|
| 293 |
"id": 50287,
|
| 294 |
"content": "<extra_id_29>",
|
| 295 |
"single_word": false,
|
| 296 |
+
"lstrip": true,
|
| 297 |
+
"rstrip": true,
|
| 298 |
"normalized": false,
|
| 299 |
"special": true
|
| 300 |
},
|
|
|
|
| 302 |
"id": 50288,
|
| 303 |
"content": "<extra_id_30>",
|
| 304 |
"single_word": false,
|
| 305 |
+
"lstrip": true,
|
| 306 |
+
"rstrip": true,
|
| 307 |
"normalized": false,
|
| 308 |
"special": true
|
| 309 |
},
|
|
|
|
| 311 |
"id": 50289,
|
| 312 |
"content": "<extra_id_31>",
|
| 313 |
"single_word": false,
|
| 314 |
+
"lstrip": true,
|
| 315 |
+
"rstrip": true,
|
| 316 |
"normalized": false,
|
| 317 |
"special": true
|
| 318 |
},
|
|
|
|
| 320 |
"id": 50290,
|
| 321 |
"content": "<extra_id_32>",
|
| 322 |
"single_word": false,
|
| 323 |
+
"lstrip": true,
|
| 324 |
+
"rstrip": true,
|
| 325 |
"normalized": false,
|
| 326 |
"special": true
|
| 327 |
},
|
|
|
|
| 329 |
"id": 50291,
|
| 330 |
"content": "<extra_id_33>",
|
| 331 |
"single_word": false,
|
| 332 |
+
"lstrip": true,
|
| 333 |
+
"rstrip": true,
|
| 334 |
"normalized": false,
|
| 335 |
"special": true
|
| 336 |
},
|
|
|
|
| 338 |
"id": 50292,
|
| 339 |
"content": "<extra_id_34>",
|
| 340 |
"single_word": false,
|
| 341 |
+
"lstrip": true,
|
| 342 |
+
"rstrip": true,
|
| 343 |
"normalized": false,
|
| 344 |
"special": true
|
| 345 |
},
|
|
|
|
| 347 |
"id": 50293,
|
| 348 |
"content": "<extra_id_35>",
|
| 349 |
"single_word": false,
|
| 350 |
+
"lstrip": true,
|
| 351 |
+
"rstrip": true,
|
| 352 |
"normalized": false,
|
| 353 |
"special": true
|
| 354 |
},
|
|
|
|
| 356 |
"id": 50294,
|
| 357 |
"content": "<extra_id_36>",
|
| 358 |
"single_word": false,
|
| 359 |
+
"lstrip": true,
|
| 360 |
+
"rstrip": true,
|
| 361 |
"normalized": false,
|
| 362 |
"special": true
|
| 363 |
},
|
|
|
|
| 365 |
"id": 50295,
|
| 366 |
"content": "<extra_id_37>",
|
| 367 |
"single_word": false,
|
| 368 |
+
"lstrip": true,
|
| 369 |
+
"rstrip": true,
|
| 370 |
"normalized": false,
|
| 371 |
"special": true
|
| 372 |
},
|
|
|
|
| 374 |
"id": 50296,
|
| 375 |
"content": "<extra_id_38>",
|
| 376 |
"single_word": false,
|
| 377 |
+
"lstrip": true,
|
| 378 |
+
"rstrip": true,
|
| 379 |
"normalized": false,
|
| 380 |
"special": true
|
| 381 |
},
|
|
|
|
| 383 |
"id": 50297,
|
| 384 |
"content": "<extra_id_39>",
|
| 385 |
"single_word": false,
|
| 386 |
+
"lstrip": true,
|
| 387 |
+
"rstrip": true,
|
| 388 |
"normalized": false,
|
| 389 |
"special": true
|
| 390 |
},
|
|
|
|
| 392 |
"id": 50298,
|
| 393 |
"content": "<extra_id_40>",
|
| 394 |
"single_word": false,
|
| 395 |
+
"lstrip": true,
|
| 396 |
+
"rstrip": true,
|
| 397 |
"normalized": false,
|
| 398 |
"special": true
|
| 399 |
},
|
|
|
|
| 401 |
"id": 50299,
|
| 402 |
"content": "<extra_id_41>",
|
| 403 |
"single_word": false,
|
| 404 |
+
"lstrip": true,
|
| 405 |
+
"rstrip": true,
|
| 406 |
"normalized": false,
|
| 407 |
"special": true
|
| 408 |
},
|
|
|
|
| 410 |
"id": 50300,
|
| 411 |
"content": "<extra_id_42>",
|
| 412 |
"single_word": false,
|
| 413 |
+
"lstrip": true,
|
| 414 |
+
"rstrip": true,
|
| 415 |
"normalized": false,
|
| 416 |
"special": true
|
| 417 |
},
|
|
|
|
| 419 |
"id": 50301,
|
| 420 |
"content": "<extra_id_43>",
|
| 421 |
"single_word": false,
|
| 422 |
+
"lstrip": true,
|
| 423 |
+
"rstrip": true,
|
| 424 |
"normalized": false,
|
| 425 |
"special": true
|
| 426 |
},
|
|
|
|
| 428 |
"id": 50302,
|
| 429 |
"content": "<extra_id_44>",
|
| 430 |
"single_word": false,
|
| 431 |
+
"lstrip": true,
|
| 432 |
+
"rstrip": true,
|
| 433 |
"normalized": false,
|
| 434 |
"special": true
|
| 435 |
},
|
|
|
|
| 437 |
"id": 50303,
|
| 438 |
"content": "<extra_id_45>",
|
| 439 |
"single_word": false,
|
| 440 |
+
"lstrip": true,
|
| 441 |
+
"rstrip": true,
|
| 442 |
"normalized": false,
|
| 443 |
"special": true
|
| 444 |
},
|
|
|
|
| 446 |
"id": 50304,
|
| 447 |
"content": "<extra_id_46>",
|
| 448 |
"single_word": false,
|
| 449 |
+
"lstrip": true,
|
| 450 |
+
"rstrip": true,
|
| 451 |
"normalized": false,
|
| 452 |
"special": true
|
| 453 |
},
|
|
|
|
| 455 |
"id": 50305,
|
| 456 |
"content": "<extra_id_47>",
|
| 457 |
"single_word": false,
|
| 458 |
+
"lstrip": true,
|
| 459 |
+
"rstrip": true,
|
| 460 |
"normalized": false,
|
| 461 |
"special": true
|
| 462 |
},
|
|
|
|
| 464 |
"id": 50306,
|
| 465 |
"content": "<extra_id_48>",
|
| 466 |
"single_word": false,
|
| 467 |
+
"lstrip": true,
|
| 468 |
+
"rstrip": true,
|
| 469 |
"normalized": false,
|
| 470 |
"special": true
|
| 471 |
},
|
|
|
|
| 473 |
"id": 50307,
|
| 474 |
"content": "<extra_id_49>",
|
| 475 |
"single_word": false,
|
| 476 |
+
"lstrip": true,
|
| 477 |
+
"rstrip": true,
|
| 478 |
"normalized": false,
|
| 479 |
"special": true
|
| 480 |
},
|
|
|
|
| 482 |
"id": 50308,
|
| 483 |
"content": "<extra_id_50>",
|
| 484 |
"single_word": false,
|
| 485 |
+
"lstrip": true,
|
| 486 |
+
"rstrip": true,
|
| 487 |
"normalized": false,
|
| 488 |
"special": true
|
| 489 |
},
|
|
|
|
| 491 |
"id": 50309,
|
| 492 |
"content": "<extra_id_51>",
|
| 493 |
"single_word": false,
|
| 494 |
+
"lstrip": true,
|
| 495 |
+
"rstrip": true,
|
| 496 |
"normalized": false,
|
| 497 |
"special": true
|
| 498 |
},
|
|
|
|
| 500 |
"id": 50310,
|
| 501 |
"content": "<extra_id_52>",
|
| 502 |
"single_word": false,
|
| 503 |
+
"lstrip": true,
|
| 504 |
+
"rstrip": true,
|
| 505 |
"normalized": false,
|
| 506 |
"special": true
|
| 507 |
},
|
|
|
|
| 509 |
"id": 50311,
|
| 510 |
"content": "<extra_id_53>",
|
| 511 |
"single_word": false,
|
| 512 |
+
"lstrip": true,
|
| 513 |
+
"rstrip": true,
|
| 514 |
"normalized": false,
|
| 515 |
"special": true
|
| 516 |
},
|
|
|
|
| 518 |
"id": 50312,
|
| 519 |
"content": "<extra_id_54>",
|
| 520 |
"single_word": false,
|
| 521 |
+
"lstrip": true,
|
| 522 |
+
"rstrip": true,
|
| 523 |
"normalized": false,
|
| 524 |
"special": true
|
| 525 |
},
|
|
|
|
| 527 |
"id": 50313,
|
| 528 |
"content": "<extra_id_55>",
|
| 529 |
"single_word": false,
|
| 530 |
+
"lstrip": true,
|
| 531 |
+
"rstrip": true,
|
| 532 |
"normalized": false,
|
| 533 |
"special": true
|
| 534 |
},
|
|
|
|
| 536 |
"id": 50314,
|
| 537 |
"content": "<extra_id_56>",
|
| 538 |
"single_word": false,
|
| 539 |
+
"lstrip": true,
|
| 540 |
+
"rstrip": true,
|
| 541 |
"normalized": false,
|
| 542 |
"special": true
|
| 543 |
},
|
|
|
|
| 545 |
"id": 50315,
|
| 546 |
"content": "<extra_id_57>",
|
| 547 |
"single_word": false,
|
| 548 |
+
"lstrip": true,
|
| 549 |
+
"rstrip": true,
|
| 550 |
"normalized": false,
|
| 551 |
"special": true
|
| 552 |
},
|
|
|
|
| 554 |
"id": 50316,
|
| 555 |
"content": "<extra_id_58>",
|
| 556 |
"single_word": false,
|
| 557 |
+
"lstrip": true,
|
| 558 |
+
"rstrip": true,
|
| 559 |
"normalized": false,
|
| 560 |
"special": true
|
| 561 |
},
|
|
|
|
| 563 |
"id": 50317,
|
| 564 |
"content": "<extra_id_59>",
|
| 565 |
"single_word": false,
|
| 566 |
+
"lstrip": true,
|
| 567 |
+
"rstrip": true,
|
| 568 |
"normalized": false,
|
| 569 |
"special": true
|
| 570 |
},
|
|
|
|
| 572 |
"id": 50318,
|
| 573 |
"content": "<extra_id_60>",
|
| 574 |
"single_word": false,
|
| 575 |
+
"lstrip": true,
|
| 576 |
+
"rstrip": true,
|
| 577 |
"normalized": false,
|
| 578 |
"special": true
|
| 579 |
},
|
|
|
|
| 581 |
"id": 50319,
|
| 582 |
"content": "<extra_id_61>",
|
| 583 |
"single_word": false,
|
| 584 |
+
"lstrip": true,
|
| 585 |
+
"rstrip": true,
|
| 586 |
"normalized": false,
|
| 587 |
"special": true
|
| 588 |
},
|
|
|
|
| 590 |
"id": 50320,
|
| 591 |
"content": "<extra_id_62>",
|
| 592 |
"single_word": false,
|
| 593 |
+
"lstrip": true,
|
| 594 |
+
"rstrip": true,
|
| 595 |
"normalized": false,
|
| 596 |
"special": true
|
| 597 |
},
|
|
|
|
| 599 |
"id": 50321,
|
| 600 |
"content": "<extra_id_63>",
|
| 601 |
"single_word": false,
|
| 602 |
+
"lstrip": true,
|
| 603 |
+
"rstrip": true,
|
| 604 |
"normalized": false,
|
| 605 |
"special": true
|
| 606 |
},
|
|
|
|
| 608 |
"id": 50322,
|
| 609 |
"content": "<extra_id_64>",
|
| 610 |
"single_word": false,
|
| 611 |
+
"lstrip": true,
|
| 612 |
+
"rstrip": true,
|
| 613 |
"normalized": false,
|
| 614 |
"special": true
|
| 615 |
},
|
|
|
|
| 617 |
"id": 50323,
|
| 618 |
"content": "<extra_id_65>",
|
| 619 |
"single_word": false,
|
| 620 |
+
"lstrip": true,
|
| 621 |
+
"rstrip": true,
|
| 622 |
"normalized": false,
|
| 623 |
"special": true
|
| 624 |
},
|
|
|
|
| 626 |
"id": 50324,
|
| 627 |
"content": "<extra_id_66>",
|
| 628 |
"single_word": false,
|
| 629 |
+
"lstrip": true,
|
| 630 |
+
"rstrip": true,
|
| 631 |
"normalized": false,
|
| 632 |
"special": true
|
| 633 |
},
|
|
|
|
| 635 |
"id": 50325,
|
| 636 |
"content": "<extra_id_67>",
|
| 637 |
"single_word": false,
|
| 638 |
+
"lstrip": true,
|
| 639 |
+
"rstrip": true,
|
| 640 |
"normalized": false,
|
| 641 |
"special": true
|
| 642 |
},
|
|
|
|
| 644 |
"id": 50326,
|
| 645 |
"content": "<extra_id_68>",
|
| 646 |
"single_word": false,
|
| 647 |
+
"lstrip": true,
|
| 648 |
+
"rstrip": true,
|
| 649 |
"normalized": false,
|
| 650 |
"special": true
|
| 651 |
},
|
|
|
|
| 653 |
"id": 50327,
|
| 654 |
"content": "<extra_id_69>",
|
| 655 |
"single_word": false,
|
| 656 |
+
"lstrip": true,
|
| 657 |
+
"rstrip": true,
|
| 658 |
"normalized": false,
|
| 659 |
"special": true
|
| 660 |
},
|
|
|
|
| 662 |
"id": 50328,
|
| 663 |
"content": "<extra_id_70>",
|
| 664 |
"single_word": false,
|
| 665 |
+
"lstrip": true,
|
| 666 |
+
"rstrip": true,
|
| 667 |
"normalized": false,
|
| 668 |
"special": true
|
| 669 |
},
|
|
|
|
| 671 |
"id": 50329,
|
| 672 |
"content": "<extra_id_71>",
|
| 673 |
"single_word": false,
|
| 674 |
+
"lstrip": true,
|
| 675 |
+
"rstrip": true,
|
| 676 |
"normalized": false,
|
| 677 |
"special": true
|
| 678 |
},
|
|
|
|
| 680 |
"id": 50330,
|
| 681 |
"content": "<extra_id_72>",
|
| 682 |
"single_word": false,
|
| 683 |
+
"lstrip": true,
|
| 684 |
+
"rstrip": true,
|
| 685 |
"normalized": false,
|
| 686 |
"special": true
|
| 687 |
},
|
|
|
|
| 689 |
"id": 50331,
|
| 690 |
"content": "<extra_id_73>",
|
| 691 |
"single_word": false,
|
| 692 |
+
"lstrip": true,
|
| 693 |
+
"rstrip": true,
|
| 694 |
"normalized": false,
|
| 695 |
"special": true
|
| 696 |
},
|
|
|
|
| 698 |
"id": 50332,
|
| 699 |
"content": "<extra_id_74>",
|
| 700 |
"single_word": false,
|
| 701 |
+
"lstrip": true,
|
| 702 |
+
"rstrip": true,
|
| 703 |
"normalized": false,
|
| 704 |
"special": true
|
| 705 |
},
|
|
|
|
| 707 |
"id": 50333,
|
| 708 |
"content": "<extra_id_75>",
|
| 709 |
"single_word": false,
|
| 710 |
+
"lstrip": true,
|
| 711 |
+
"rstrip": true,
|
| 712 |
"normalized": false,
|
| 713 |
"special": true
|
| 714 |
},
|
|
|
|
| 716 |
"id": 50334,
|
| 717 |
"content": "<extra_id_76>",
|
| 718 |
"single_word": false,
|
| 719 |
+
"lstrip": true,
|
| 720 |
+
"rstrip": true,
|
| 721 |
"normalized": false,
|
| 722 |
"special": true
|
| 723 |
},
|
|
|
|
| 725 |
"id": 50335,
|
| 726 |
"content": "<extra_id_77>",
|
| 727 |
"single_word": false,
|
| 728 |
+
"lstrip": true,
|
| 729 |
+
"rstrip": true,
|
| 730 |
"normalized": false,
|
| 731 |
"special": true
|
| 732 |
},
|
|
|
|
| 734 |
"id": 50336,
|
| 735 |
"content": "<extra_id_78>",
|
| 736 |
"single_word": false,
|
| 737 |
+
"lstrip": true,
|
| 738 |
+
"rstrip": true,
|
| 739 |
"normalized": false,
|
| 740 |
"special": true
|
| 741 |
},
|
|
|
|
| 743 |
"id": 50337,
|
| 744 |
"content": "<extra_id_79>",
|
| 745 |
"single_word": false,
|
| 746 |
+
"lstrip": true,
|
| 747 |
+
"rstrip": true,
|
| 748 |
"normalized": false,
|
| 749 |
"special": true
|
| 750 |
},
|
|
|
|
| 752 |
"id": 50338,
|
| 753 |
"content": "<extra_id_80>",
|
| 754 |
"single_word": false,
|
| 755 |
+
"lstrip": true,
|
| 756 |
+
"rstrip": true,
|
| 757 |
"normalized": false,
|
| 758 |
"special": true
|
| 759 |
},
|
|
|
|
| 761 |
"id": 50339,
|
| 762 |
"content": "<extra_id_81>",
|
| 763 |
"single_word": false,
|
| 764 |
+
"lstrip": true,
|
| 765 |
+
"rstrip": true,
|
| 766 |
"normalized": false,
|
| 767 |
"special": true
|
| 768 |
},
|
|
|
|
| 770 |
"id": 50340,
|
| 771 |
"content": "<extra_id_82>",
|
| 772 |
"single_word": false,
|
| 773 |
+
"lstrip": true,
|
| 774 |
+
"rstrip": true,
|
| 775 |
"normalized": false,
|
| 776 |
"special": true
|
| 777 |
},
|
|
|
|
| 779 |
"id": 50341,
|
| 780 |
"content": "<extra_id_83>",
|
| 781 |
"single_word": false,
|
| 782 |
+
"lstrip": true,
|
| 783 |
+
"rstrip": true,
|
| 784 |
"normalized": false,
|
| 785 |
"special": true
|
| 786 |
},
|
|
|
|
| 788 |
"id": 50342,
|
| 789 |
"content": "<extra_id_84>",
|
| 790 |
"single_word": false,
|
| 791 |
+
"lstrip": true,
|
| 792 |
+
"rstrip": true,
|
| 793 |
"normalized": false,
|
| 794 |
"special": true
|
| 795 |
},
|
|
|
|
| 797 |
"id": 50343,
|
| 798 |
"content": "<extra_id_85>",
|
| 799 |
"single_word": false,
|
| 800 |
+
"lstrip": true,
|
| 801 |
+
"rstrip": true,
|
| 802 |
"normalized": false,
|
| 803 |
"special": true
|
| 804 |
},
|
|
|
|
| 806 |
"id": 50344,
|
| 807 |
"content": "<extra_id_86>",
|
| 808 |
"single_word": false,
|
| 809 |
+
"lstrip": true,
|
| 810 |
+
"rstrip": true,
|
| 811 |
"normalized": false,
|
| 812 |
"special": true
|
| 813 |
},
|
|
|
|
| 815 |
"id": 50345,
|
| 816 |
"content": "<extra_id_87>",
|
| 817 |
"single_word": false,
|
| 818 |
+
"lstrip": true,
|
| 819 |
+
"rstrip": true,
|
| 820 |
"normalized": false,
|
| 821 |
"special": true
|
| 822 |
},
|
|
|
|
| 824 |
"id": 50346,
|
| 825 |
"content": "<extra_id_88>",
|
| 826 |
"single_word": false,
|
| 827 |
+
"lstrip": true,
|
| 828 |
+
"rstrip": true,
|
| 829 |
"normalized": false,
|
| 830 |
"special": true
|
| 831 |
},
|
|
|
|
| 833 |
"id": 50347,
|
| 834 |
"content": "<extra_id_89>",
|
| 835 |
"single_word": false,
|
| 836 |
+
"lstrip": true,
|
| 837 |
+
"rstrip": true,
|
| 838 |
"normalized": false,
|
| 839 |
"special": true
|
| 840 |
},
|
|
|
|
| 842 |
"id": 50348,
|
| 843 |
"content": "<extra_id_90>",
|
| 844 |
"single_word": false,
|
| 845 |
+
"lstrip": true,
|
| 846 |
+
"rstrip": true,
|
| 847 |
"normalized": false,
|
| 848 |
"special": true
|
| 849 |
},
|
|
|
|
| 851 |
"id": 50349,
|
| 852 |
"content": "<extra_id_91>",
|
| 853 |
"single_word": false,
|
| 854 |
+
"lstrip": true,
|
| 855 |
+
"rstrip": true,
|
| 856 |
"normalized": false,
|
| 857 |
"special": true
|
| 858 |
},
|
|
|
|
| 860 |
"id": 50350,
|
| 861 |
"content": "<extra_id_92>",
|
| 862 |
"single_word": false,
|
| 863 |
+
"lstrip": true,
|
| 864 |
+
"rstrip": true,
|
| 865 |
"normalized": false,
|
| 866 |
"special": true
|
| 867 |
},
|
|
|
|
| 869 |
"id": 50351,
|
| 870 |
"content": "<extra_id_93>",
|
| 871 |
"single_word": false,
|
| 872 |
+
"lstrip": true,
|
| 873 |
+
"rstrip": true,
|
| 874 |
"normalized": false,
|
| 875 |
"special": true
|
| 876 |
},
|
|
|
|
| 878 |
"id": 50352,
|
| 879 |
"content": "<extra_id_94>",
|
| 880 |
"single_word": false,
|
| 881 |
+
"lstrip": true,
|
| 882 |
+
"rstrip": true,
|
| 883 |
"normalized": false,
|
| 884 |
"special": true
|
| 885 |
},
|
|
|
|
| 887 |
"id": 50353,
|
| 888 |
"content": "<extra_id_95>",
|
| 889 |
"single_word": false,
|
| 890 |
+
"lstrip": true,
|
| 891 |
+
"rstrip": true,
|
| 892 |
"normalized": false,
|
| 893 |
"special": true
|
| 894 |
},
|
|
|
|
| 896 |
"id": 50354,
|
| 897 |
"content": "<extra_id_96>",
|
| 898 |
"single_word": false,
|
| 899 |
+
"lstrip": true,
|
| 900 |
+
"rstrip": true,
|
| 901 |
"normalized": false,
|
| 902 |
"special": true
|
| 903 |
},
|
|
|
|
| 905 |
"id": 50355,
|
| 906 |
"content": "<extra_id_97>",
|
| 907 |
"single_word": false,
|
| 908 |
+
"lstrip": true,
|
| 909 |
+
"rstrip": true,
|
| 910 |
"normalized": false,
|
| 911 |
"special": true
|
| 912 |
},
|
|
|
|
| 914 |
"id": 50356,
|
| 915 |
"content": "<extra_id_98>",
|
| 916 |
"single_word": false,
|
| 917 |
+
"lstrip": true,
|
| 918 |
+
"rstrip": true,
|
| 919 |
"normalized": false,
|
| 920 |
"special": true
|
| 921 |
},
|
|
|
|
| 923 |
"id": 50357,
|
| 924 |
"content": "<extra_id_99>",
|
| 925 |
"single_word": false,
|
| 926 |
+
"lstrip": true,
|
| 927 |
+
"rstrip": true,
|
| 928 |
"normalized": false,
|
| 929 |
"special": true
|
| 930 |
}
|
|
|
|
| 999 |
"continuing_subword_prefix": "",
|
| 1000 |
"end_of_word_suffix": "",
|
| 1001 |
"fuse_unk": false,
|
| 1002 |
+
"byte_fallback": false,
|
| 1003 |
+
"ignore_merges": false,
|
| 1004 |
"vocab": {
|
| 1005 |
"<pad>": 0,
|
| 1006 |
"</s>": 1,
|