Upload 4 files
Browse files- merges.txt +0 -0
- token_frequencies.pkl +3 -0
- tokenizer_config.json +107 -0
- vocab.json +0 -0
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
token_frequencies.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a7ae0615c05920ca290b769650eda0552e116598548e4cfb8398eab2452c1e7
|
| 3 |
+
size 371919
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,107 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 32000,
|
| 3 |
+
"min_freq": 2,
|
| 4 |
+
"special_tokens": {
|
| 5 |
+
"<pad>": 0,
|
| 6 |
+
"<unk>": 1,
|
| 7 |
+
"<bos>": 2,
|
| 8 |
+
"<eos>": 3,
|
| 9 |
+
"<system>": 4,
|
| 10 |
+
"<user>": 5,
|
| 11 |
+
"<assistant>": 6,
|
| 12 |
+
"<|endoftext|>": 7,
|
| 13 |
+
"<|newline|>": 8,
|
| 14 |
+
"<|tab|>": 9,
|
| 15 |
+
"<|code|>": 10,
|
| 16 |
+
"<|/code|>": 11,
|
| 17 |
+
"<|math|>": 12,
|
| 18 |
+
"<|/math|>": 13
|
| 19 |
+
},
|
| 20 |
+
"technical_terms": [
|
| 21 |
+
"protocol",
|
| 22 |
+
"backpropagation",
|
| 23 |
+
"object",
|
| 24 |
+
"async",
|
| 25 |
+
"validation",
|
| 26 |
+
"test",
|
| 27 |
+
"await",
|
| 28 |
+
"boolean",
|
| 29 |
+
"hypothesis",
|
| 30 |
+
"f1score",
|
| 31 |
+
"string",
|
| 32 |
+
"html",
|
| 33 |
+
"polynomial",
|
| 34 |
+
"neuron",
|
| 35 |
+
"class",
|
| 36 |
+
"loss",
|
| 37 |
+
"import",
|
| 38 |
+
"promise",
|
| 39 |
+
"graphql",
|
| 40 |
+
"export",
|
| 41 |
+
"authorization",
|
| 42 |
+
"proof",
|
| 43 |
+
"theorem",
|
| 44 |
+
"variable",
|
| 45 |
+
"authentication",
|
| 46 |
+
"https",
|
| 47 |
+
"regression",
|
| 48 |
+
"algorithm",
|
| 49 |
+
"security",
|
| 50 |
+
"architecture",
|
| 51 |
+
"function",
|
| 52 |
+
"accuracy",
|
| 53 |
+
"statistics",
|
| 54 |
+
"precision",
|
| 55 |
+
"css",
|
| 56 |
+
"weight",
|
| 57 |
+
"python",
|
| 58 |
+
"sql",
|
| 59 |
+
"attention",
|
| 60 |
+
"float",
|
| 61 |
+
"encryption",
|
| 62 |
+
"docker",
|
| 63 |
+
"gradient",
|
| 64 |
+
"datatype",
|
| 65 |
+
"transformer",
|
| 66 |
+
"java",
|
| 67 |
+
"formula",
|
| 68 |
+
"equation",
|
| 69 |
+
"network",
|
| 70 |
+
"vector",
|
| 71 |
+
"client",
|
| 72 |
+
"callback",
|
| 73 |
+
"integer",
|
| 74 |
+
"http",
|
| 75 |
+
"database",
|
| 76 |
+
"layer",
|
| 77 |
+
"javascript",
|
| 78 |
+
"bias",
|
| 79 |
+
"cpp",
|
| 80 |
+
"matrix",
|
| 81 |
+
"embedding",
|
| 82 |
+
"microservice",
|
| 83 |
+
"rest",
|
| 84 |
+
"derivative",
|
| 85 |
+
"recall",
|
| 86 |
+
"tokenization",
|
| 87 |
+
"xml",
|
| 88 |
+
"scalability",
|
| 89 |
+
"kubernetes",
|
| 90 |
+
"model",
|
| 91 |
+
"return",
|
| 92 |
+
"neural",
|
| 93 |
+
"parameter",
|
| 94 |
+
"server",
|
| 95 |
+
"probability",
|
| 96 |
+
"deployment",
|
| 97 |
+
"array",
|
| 98 |
+
"api",
|
| 99 |
+
"integral",
|
| 100 |
+
"training",
|
| 101 |
+
"forward",
|
| 102 |
+
"correlation",
|
| 103 |
+
"method",
|
| 104 |
+
"json",
|
| 105 |
+
"performance"
|
| 106 |
+
]
|
| 107 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|