Upload folder using huggingface_hub
Browse files- letters-v3/letters.onnx +3 -0
- letters-v3/metrics.json +276 -0
- letters-v3/test-outcomes.jsonl +0 -0
letters-v3/letters.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84dac682f96385c73485d0d00e975d19ee4e340e17493eb1408b2cf16c1bee61
|
| 3 |
+
size 6051774
|
letters-v3/metrics.json
ADDED
|
@@ -0,0 +1,276 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"args": {
|
| 3 |
+
"repo": "cfahlgren1/tinyfix-realword",
|
| 4 |
+
"data_repo": "cfahlgren1/tinyfix-train-data",
|
| 5 |
+
"data_dir": null,
|
| 6 |
+
"run_name": "letters-v3",
|
| 7 |
+
"fineweb_tokens": 70000000,
|
| 8 |
+
"smoltalk_tokens": 50000000,
|
| 9 |
+
"oasst_tokens": 1000000000000,
|
| 10 |
+
"vocabulary": 20000,
|
| 11 |
+
"embedding": 64,
|
| 12 |
+
"hidden": 256,
|
| 13 |
+
"steps": 40000,
|
| 14 |
+
"batch": 4096,
|
| 15 |
+
"learning_rate": 0.002,
|
| 16 |
+
"misspellings_per_word": 16,
|
| 17 |
+
"min_count": 5,
|
| 18 |
+
"workers": 48,
|
| 19 |
+
"tuning_pairs": 6000,
|
| 20 |
+
"output": "/tmp/tinyfix-letters",
|
| 21 |
+
"no_upload": false
|
| 22 |
+
},
|
| 23 |
+
"parameters": 1960537,
|
| 24 |
+
"train_minutes": 37.8,
|
| 25 |
+
"validation": {
|
| 26 |
+
"accuracy": 0.9548,
|
| 27 |
+
"corrupted_accuracy": 0.9294
|
| 28 |
+
},
|
| 29 |
+
"real_typos": {
|
| 30 |
+
"torch": {
|
| 31 |
+
"careful": {
|
| 32 |
+
"thresholds": {
|
| 33 |
+
"known_threshold": 8.0,
|
| 34 |
+
"unknown_threshold": 10.5,
|
| 35 |
+
"dev_gain": 2498
|
| 36 |
+
},
|
| 37 |
+
"dev": {
|
| 38 |
+
"misspelling": {
|
| 39 |
+
"right": 0.5953,
|
| 40 |
+
"wrong": 0.0266,
|
| 41 |
+
"total": 4243
|
| 42 |
+
},
|
| 43 |
+
"real word": {
|
| 44 |
+
"right": 0.0632,
|
| 45 |
+
"wrong": 0.004,
|
| 46 |
+
"total": 1757
|
| 47 |
+
},
|
| 48 |
+
"all right": 0.4395,
|
| 49 |
+
"all wrong": 0.02,
|
| 50 |
+
"email changed per 1000": 1.166,
|
| 51 |
+
"github changed per 1000": 1.466
|
| 52 |
+
},
|
| 53 |
+
"test": {
|
| 54 |
+
"misspelling": {
|
| 55 |
+
"right": 0.6082,
|
| 56 |
+
"wrong": 0.0255,
|
| 57 |
+
"total": 2938
|
| 58 |
+
},
|
| 59 |
+
"real word": {
|
| 60 |
+
"right": 0.0546,
|
| 61 |
+
"wrong": 0.0038,
|
| 62 |
+
"total": 1062
|
| 63 |
+
},
|
| 64 |
+
"all right": 0.4612,
|
| 65 |
+
"all wrong": 0.0198,
|
| 66 |
+
"email changed per 1000": 0.766,
|
| 67 |
+
"github changed per 1000": 0.733
|
| 68 |
+
}
|
| 69 |
+
},
|
| 70 |
+
"eager": {
|
| 71 |
+
"thresholds": {
|
| 72 |
+
"known_threshold": 6.0,
|
| 73 |
+
"unknown_threshold": 8.5,
|
| 74 |
+
"dev_gain": 2980
|
| 75 |
+
},
|
| 76 |
+
"dev": {
|
| 77 |
+
"misspelling": {
|
| 78 |
+
"right": 0.6967,
|
| 79 |
+
"wrong": 0.0391,
|
| 80 |
+
"total": 4243
|
| 81 |
+
},
|
| 82 |
+
"real word": {
|
| 83 |
+
"right": 0.1246,
|
| 84 |
+
"wrong": 0.0057,
|
| 85 |
+
"total": 1757
|
| 86 |
+
},
|
| 87 |
+
"all right": 0.5292,
|
| 88 |
+
"all wrong": 0.0293,
|
| 89 |
+
"email changed per 1000": 1.866,
|
| 90 |
+
"github changed per 1000": 2.132
|
| 91 |
+
},
|
| 92 |
+
"test": {
|
| 93 |
+
"misspelling": {
|
| 94 |
+
"right": 0.7103,
|
| 95 |
+
"wrong": 0.0425,
|
| 96 |
+
"total": 2938
|
| 97 |
+
},
|
| 98 |
+
"real word": {
|
| 99 |
+
"right": 0.1177,
|
| 100 |
+
"wrong": 0.0075,
|
| 101 |
+
"total": 1062
|
| 102 |
+
},
|
| 103 |
+
"all right": 0.553,
|
| 104 |
+
"all wrong": 0.0333,
|
| 105 |
+
"email changed per 1000": 1.033,
|
| 106 |
+
"github changed per 1000": 1.266
|
| 107 |
+
}
|
| 108 |
+
}
|
| 109 |
+
},
|
| 110 |
+
"matmul-gather": {
|
| 111 |
+
"careful": {
|
| 112 |
+
"thresholds": {
|
| 113 |
+
"known_threshold": 16.5,
|
| 114 |
+
"unknown_threshold": 13.0,
|
| 115 |
+
"dev_gain": 2129
|
| 116 |
+
},
|
| 117 |
+
"dev": {
|
| 118 |
+
"misspelling": {
|
| 119 |
+
"right": 0.5263,
|
| 120 |
+
"wrong": 0.0259,
|
| 121 |
+
"total": 4243
|
| 122 |
+
},
|
| 123 |
+
"real word": {
|
| 124 |
+
"right": 0.0159,
|
| 125 |
+
"wrong": 0.0017,
|
| 126 |
+
"total": 1757
|
| 127 |
+
},
|
| 128 |
+
"all right": 0.3768,
|
| 129 |
+
"all wrong": 0.0188,
|
| 130 |
+
"email changed per 1000": 1.166,
|
| 131 |
+
"github changed per 1000": 1.199
|
| 132 |
+
},
|
| 133 |
+
"test": {
|
| 134 |
+
"misspelling": {
|
| 135 |
+
"right": 0.5296,
|
| 136 |
+
"wrong": 0.0248,
|
| 137 |
+
"total": 2938
|
| 138 |
+
},
|
| 139 |
+
"real word": {
|
| 140 |
+
"right": 0.0122,
|
| 141 |
+
"wrong": 0.0,
|
| 142 |
+
"total": 1062
|
| 143 |
+
},
|
| 144 |
+
"all right": 0.3922,
|
| 145 |
+
"all wrong": 0.0182,
|
| 146 |
+
"email changed per 1000": 0.6,
|
| 147 |
+
"github changed per 1000": 0.6
|
| 148 |
+
}
|
| 149 |
+
},
|
| 150 |
+
"eager": {
|
| 151 |
+
"thresholds": {
|
| 152 |
+
"known_threshold": 12.0,
|
| 153 |
+
"unknown_threshold": 9.5,
|
| 154 |
+
"dev_gain": 2814
|
| 155 |
+
},
|
| 156 |
+
"dev": {
|
| 157 |
+
"misspelling": {
|
| 158 |
+
"right": 0.6825,
|
| 159 |
+
"wrong": 0.0455,
|
| 160 |
+
"total": 4243
|
| 161 |
+
},
|
| 162 |
+
"real word": {
|
| 163 |
+
"right": 0.0774,
|
| 164 |
+
"wrong": 0.0034,
|
| 165 |
+
"total": 1757
|
| 166 |
+
},
|
| 167 |
+
"all right": 0.5053,
|
| 168 |
+
"all wrong": 0.0332,
|
| 169 |
+
"email changed per 1000": 1.833,
|
| 170 |
+
"github changed per 1000": 2.732
|
| 171 |
+
},
|
| 172 |
+
"test": {
|
| 173 |
+
"misspelling": {
|
| 174 |
+
"right": 0.7022,
|
| 175 |
+
"wrong": 0.0425,
|
| 176 |
+
"total": 2938
|
| 177 |
+
},
|
| 178 |
+
"real word": {
|
| 179 |
+
"right": 0.0631,
|
| 180 |
+
"wrong": 0.0019,
|
| 181 |
+
"total": 1062
|
| 182 |
+
},
|
| 183 |
+
"all right": 0.5325,
|
| 184 |
+
"all wrong": 0.0318,
|
| 185 |
+
"email changed per 1000": 1.033,
|
| 186 |
+
"github changed per 1000": 2.066
|
| 187 |
+
}
|
| 188 |
+
}
|
| 189 |
+
},
|
| 190 |
+
"matmul": {
|
| 191 |
+
"careful": {
|
| 192 |
+
"thresholds": {
|
| 193 |
+
"known_threshold": 15.5,
|
| 194 |
+
"unknown_threshold": 12.0,
|
| 195 |
+
"dev_gain": 2339
|
| 196 |
+
},
|
| 197 |
+
"dev": {
|
| 198 |
+
"misspelling": {
|
| 199 |
+
"right": 0.5798,
|
| 200 |
+
"wrong": 0.0321,
|
| 201 |
+
"total": 4243
|
| 202 |
+
},
|
| 203 |
+
"real word": {
|
| 204 |
+
"right": 0.0216,
|
| 205 |
+
"wrong": 0.0023,
|
| 206 |
+
"total": 1757
|
| 207 |
+
},
|
| 208 |
+
"all right": 0.4163,
|
| 209 |
+
"all wrong": 0.0233,
|
| 210 |
+
"email changed per 1000": 1.166,
|
| 211 |
+
"github changed per 1000": 1.799
|
| 212 |
+
},
|
| 213 |
+
"test": {
|
| 214 |
+
"misspelling": {
|
| 215 |
+
"right": 0.6045,
|
| 216 |
+
"wrong": 0.0289,
|
| 217 |
+
"total": 2938
|
| 218 |
+
},
|
| 219 |
+
"real word": {
|
| 220 |
+
"right": 0.0151,
|
| 221 |
+
"wrong": 0.0,
|
| 222 |
+
"total": 1062
|
| 223 |
+
},
|
| 224 |
+
"all right": 0.448,
|
| 225 |
+
"all wrong": 0.0213,
|
| 226 |
+
"email changed per 1000": 0.766,
|
| 227 |
+
"github changed per 1000": 1.066
|
| 228 |
+
}
|
| 229 |
+
},
|
| 230 |
+
"eager": {
|
| 231 |
+
"thresholds": {
|
| 232 |
+
"known_threshold": 10.5,
|
| 233 |
+
"unknown_threshold": 10.0,
|
| 234 |
+
"dev_gain": 2823
|
| 235 |
+
},
|
| 236 |
+
"dev": {
|
| 237 |
+
"misspelling": {
|
| 238 |
+
"right": 0.667,
|
| 239 |
+
"wrong": 0.0431,
|
| 240 |
+
"total": 4243
|
| 241 |
+
},
|
| 242 |
+
"real word": {
|
| 243 |
+
"right": 0.1155,
|
| 244 |
+
"wrong": 0.0046,
|
| 245 |
+
"total": 1757
|
| 246 |
+
},
|
| 247 |
+
"all right": 0.5055,
|
| 248 |
+
"all wrong": 0.0318,
|
| 249 |
+
"email changed per 1000": 1.866,
|
| 250 |
+
"github changed per 1000": 2.598
|
| 251 |
+
},
|
| 252 |
+
"test": {
|
| 253 |
+
"misspelling": {
|
| 254 |
+
"right": 0.6903,
|
| 255 |
+
"wrong": 0.0408,
|
| 256 |
+
"total": 2938
|
| 257 |
+
},
|
| 258 |
+
"real word": {
|
| 259 |
+
"right": 0.0989,
|
| 260 |
+
"wrong": 0.0038,
|
| 261 |
+
"total": 1062
|
| 262 |
+
},
|
| 263 |
+
"all right": 0.5333,
|
| 264 |
+
"all wrong": 0.031,
|
| 265 |
+
"email changed per 1000": 0.966,
|
| 266 |
+
"github changed per 1000": 1.999
|
| 267 |
+
}
|
| 268 |
+
}
|
| 269 |
+
}
|
| 270 |
+
},
|
| 271 |
+
"sizes_mb": {
|
| 272 |
+
"matmul-gather": 2.15,
|
| 273 |
+
"matmul": 6.05
|
| 274 |
+
},
|
| 275 |
+
"shipped": "matmul"
|
| 276 |
+
}
|
letters-v3/test-outcomes.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|