Upload folder using huggingface_hub
Browse files- config.json +15 -0
- cubic_strided_step_3000.pt +3 -0
- tokenizer.json +256 -0
config.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 125,
|
| 3 |
+
"dim": 16,
|
| 4 |
+
"n_layers": 7,
|
| 5 |
+
"n_heads": 4,
|
| 6 |
+
"n_kv_heads": 1,
|
| 7 |
+
"intermediate_size": 44,
|
| 8 |
+
"max_seq_len": 1024,
|
| 9 |
+
"rope_theta": 10000.0,
|
| 10 |
+
"rms_norm_eps": 1e-06,
|
| 11 |
+
"dropout": 0.1,
|
| 12 |
+
"tie_weights": true,
|
| 13 |
+
"attention_type": "cubic-strided",
|
| 14 |
+
"mod_stride": 16
|
| 15 |
+
}
|
cubic_strided_step_3000.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:531a50842f6c5d4b42408c023a5ea02fd0d5a7d30633b41d6ead713b2fb6c379
|
| 3 |
+
size 177478
|
tokenizer.json
ADDED
|
@@ -0,0 +1,256 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab": [
|
| 3 |
+
"<pad>",
|
| 4 |
+
"<unk>",
|
| 5 |
+
"<bos>",
|
| 6 |
+
"<eos>",
|
| 7 |
+
"a",
|
| 8 |
+
"b",
|
| 9 |
+
"c",
|
| 10 |
+
"d",
|
| 11 |
+
"e",
|
| 12 |
+
"f",
|
| 13 |
+
"g",
|
| 14 |
+
"h",
|
| 15 |
+
"i",
|
| 16 |
+
"j",
|
| 17 |
+
"k",
|
| 18 |
+
"l",
|
| 19 |
+
"m",
|
| 20 |
+
"n",
|
| 21 |
+
"o",
|
| 22 |
+
"p",
|
| 23 |
+
"q",
|
| 24 |
+
"r",
|
| 25 |
+
"s",
|
| 26 |
+
"t",
|
| 27 |
+
"u",
|
| 28 |
+
"v",
|
| 29 |
+
"w",
|
| 30 |
+
"x",
|
| 31 |
+
"y",
|
| 32 |
+
"z",
|
| 33 |
+
"es",
|
| 34 |
+
"er",
|
| 35 |
+
"ed",
|
| 36 |
+
"'s",
|
| 37 |
+
"ng",
|
| 38 |
+
"ing",
|
| 39 |
+
"ma",
|
| 40 |
+
"le",
|
| 41 |
+
"al",
|
| 42 |
+
"re",
|
| 43 |
+
"en",
|
| 44 |
+
"co",
|
| 45 |
+
"ne",
|
| 46 |
+
"ch",
|
| 47 |
+
"st",
|
| 48 |
+
"te",
|
| 49 |
+
"the",
|
| 50 |
+
"and",
|
| 51 |
+
"of",
|
| 52 |
+
"in",
|
| 53 |
+
"to",
|
| 54 |
+
"is",
|
| 55 |
+
"was",
|
| 56 |
+
"it",
|
| 57 |
+
"he",
|
| 58 |
+
"for",
|
| 59 |
+
"on",
|
| 60 |
+
"as",
|
| 61 |
+
"by",
|
| 62 |
+
"that",
|
| 63 |
+
"with",
|
| 64 |
+
"at",
|
| 65 |
+
"be",
|
| 66 |
+
"this",
|
| 67 |
+
"have",
|
| 68 |
+
"water",
|
| 69 |
+
"sun",
|
| 70 |
+
"earth",
|
| 71 |
+
"tree",
|
| 72 |
+
"plant",
|
| 73 |
+
"animal",
|
| 74 |
+
"bird",
|
| 75 |
+
"fish",
|
| 76 |
+
"dog",
|
| 77 |
+
"cat",
|
| 78 |
+
"fire",
|
| 79 |
+
"air",
|
| 80 |
+
"head",
|
| 81 |
+
"hand",
|
| 82 |
+
"eye",
|
| 83 |
+
"face",
|
| 84 |
+
"body",
|
| 85 |
+
"eat",
|
| 86 |
+
"walk",
|
| 87 |
+
"play",
|
| 88 |
+
"work",
|
| 89 |
+
"see",
|
| 90 |
+
"make",
|
| 91 |
+
"take",
|
| 92 |
+
"sleep",
|
| 93 |
+
"talk",
|
| 94 |
+
"help",
|
| 95 |
+
"know",
|
| 96 |
+
"house",
|
| 97 |
+
"home",
|
| 98 |
+
"room",
|
| 99 |
+
"door",
|
| 100 |
+
"car",
|
| 101 |
+
"book",
|
| 102 |
+
"bed",
|
| 103 |
+
"school",
|
| 104 |
+
"big",
|
| 105 |
+
"small",
|
| 106 |
+
"good",
|
| 107 |
+
"bad",
|
| 108 |
+
"new",
|
| 109 |
+
"old",
|
| 110 |
+
"hot",
|
| 111 |
+
"cold",
|
| 112 |
+
"time",
|
| 113 |
+
"day",
|
| 114 |
+
"night",
|
| 115 |
+
"year",
|
| 116 |
+
"world",
|
| 117 |
+
"up",
|
| 118 |
+
"man",
|
| 119 |
+
"woman",
|
| 120 |
+
"child",
|
| 121 |
+
"boy",
|
| 122 |
+
"girl",
|
| 123 |
+
"people",
|
| 124 |
+
"one",
|
| 125 |
+
"two",
|
| 126 |
+
"three",
|
| 127 |
+
"four"
|
| 128 |
+
],
|
| 129 |
+
"tok_to_id": {
|
| 130 |
+
"<pad>": 0,
|
| 131 |
+
"<unk>": 1,
|
| 132 |
+
"<bos>": 2,
|
| 133 |
+
"<eos>": 3,
|
| 134 |
+
"a": 4,
|
| 135 |
+
"b": 5,
|
| 136 |
+
"c": 6,
|
| 137 |
+
"d": 7,
|
| 138 |
+
"e": 8,
|
| 139 |
+
"f": 9,
|
| 140 |
+
"g": 10,
|
| 141 |
+
"h": 11,
|
| 142 |
+
"i": 12,
|
| 143 |
+
"j": 13,
|
| 144 |
+
"k": 14,
|
| 145 |
+
"l": 15,
|
| 146 |
+
"m": 16,
|
| 147 |
+
"n": 17,
|
| 148 |
+
"o": 18,
|
| 149 |
+
"p": 19,
|
| 150 |
+
"q": 20,
|
| 151 |
+
"r": 21,
|
| 152 |
+
"s": 22,
|
| 153 |
+
"t": 23,
|
| 154 |
+
"u": 24,
|
| 155 |
+
"v": 25,
|
| 156 |
+
"w": 26,
|
| 157 |
+
"x": 27,
|
| 158 |
+
"y": 28,
|
| 159 |
+
"z": 29,
|
| 160 |
+
"es": 30,
|
| 161 |
+
"er": 31,
|
| 162 |
+
"ed": 32,
|
| 163 |
+
"'s": 33,
|
| 164 |
+
"ng": 34,
|
| 165 |
+
"ing": 35,
|
| 166 |
+
"ma": 36,
|
| 167 |
+
"le": 37,
|
| 168 |
+
"al": 38,
|
| 169 |
+
"re": 39,
|
| 170 |
+
"en": 40,
|
| 171 |
+
"co": 41,
|
| 172 |
+
"ne": 42,
|
| 173 |
+
"ch": 43,
|
| 174 |
+
"st": 44,
|
| 175 |
+
"te": 45,
|
| 176 |
+
"the": 46,
|
| 177 |
+
"and": 47,
|
| 178 |
+
"of": 48,
|
| 179 |
+
"in": 49,
|
| 180 |
+
"to": 50,
|
| 181 |
+
"is": 51,
|
| 182 |
+
"was": 52,
|
| 183 |
+
"it": 53,
|
| 184 |
+
"he": 54,
|
| 185 |
+
"for": 55,
|
| 186 |
+
"on": 56,
|
| 187 |
+
"as": 57,
|
| 188 |
+
"by": 58,
|
| 189 |
+
"that": 59,
|
| 190 |
+
"with": 60,
|
| 191 |
+
"at": 61,
|
| 192 |
+
"be": 62,
|
| 193 |
+
"this": 63,
|
| 194 |
+
"have": 64,
|
| 195 |
+
"water": 65,
|
| 196 |
+
"sun": 66,
|
| 197 |
+
"earth": 67,
|
| 198 |
+
"tree": 68,
|
| 199 |
+
"plant": 69,
|
| 200 |
+
"animal": 70,
|
| 201 |
+
"bird": 71,
|
| 202 |
+
"fish": 72,
|
| 203 |
+
"dog": 73,
|
| 204 |
+
"cat": 74,
|
| 205 |
+
"fire": 75,
|
| 206 |
+
"air": 76,
|
| 207 |
+
"head": 77,
|
| 208 |
+
"hand": 78,
|
| 209 |
+
"eye": 79,
|
| 210 |
+
"face": 80,
|
| 211 |
+
"body": 81,
|
| 212 |
+
"eat": 82,
|
| 213 |
+
"walk": 83,
|
| 214 |
+
"play": 84,
|
| 215 |
+
"work": 85,
|
| 216 |
+
"see": 86,
|
| 217 |
+
"make": 87,
|
| 218 |
+
"take": 88,
|
| 219 |
+
"sleep": 89,
|
| 220 |
+
"talk": 90,
|
| 221 |
+
"help": 91,
|
| 222 |
+
"know": 92,
|
| 223 |
+
"house": 93,
|
| 224 |
+
"home": 94,
|
| 225 |
+
"room": 95,
|
| 226 |
+
"door": 96,
|
| 227 |
+
"car": 97,
|
| 228 |
+
"book": 98,
|
| 229 |
+
"bed": 99,
|
| 230 |
+
"school": 100,
|
| 231 |
+
"big": 101,
|
| 232 |
+
"small": 102,
|
| 233 |
+
"good": 103,
|
| 234 |
+
"bad": 104,
|
| 235 |
+
"new": 105,
|
| 236 |
+
"old": 106,
|
| 237 |
+
"hot": 107,
|
| 238 |
+
"cold": 108,
|
| 239 |
+
"time": 109,
|
| 240 |
+
"day": 110,
|
| 241 |
+
"night": 111,
|
| 242 |
+
"year": 112,
|
| 243 |
+
"world": 113,
|
| 244 |
+
"up": 114,
|
| 245 |
+
"man": 115,
|
| 246 |
+
"woman": 116,
|
| 247 |
+
"child": 117,
|
| 248 |
+
"boy": 118,
|
| 249 |
+
"girl": 119,
|
| 250 |
+
"people": 120,
|
| 251 |
+
"one": 121,
|
| 252 |
+
"two": 122,
|
| 253 |
+
"three": 123,
|
| 254 |
+
"four": 124
|
| 255 |
+
}
|
| 256 |
+
}
|