Upload sigma_lambda_proto_v4/history.json with huggingface_hub
Browse files
sigma_lambda_proto_v4/history.json
CHANGED
|
@@ -14,7 +14,8 @@
|
|
| 14 |
12,
|
| 15 |
13,
|
| 16 |
14,
|
| 17 |
-
15
|
|
|
|
| 18 |
],
|
| 19 |
"gen_loss": [
|
| 20 |
18.370433433850607,
|
|
@@ -31,7 +32,8 @@
|
|
| 31 |
11.003662894566855,
|
| 32 |
10.940823611346158,
|
| 33 |
10.90196396784349,
|
| 34 |
-
11.473219166524483
|
|
|
|
| 35 |
],
|
| 36 |
"ret_loss": [
|
| 37 |
0.0,
|
|
@@ -48,7 +50,8 @@
|
|
| 48 |
0.0,
|
| 49 |
0.0,
|
| 50 |
0.0,
|
| 51 |
-
297885048197.7794
|
|
|
|
| 52 |
],
|
| 53 |
"ent_loss": [
|
| 54 |
0.6078802872065342,
|
|
@@ -65,7 +68,8 @@
|
|
| 65 |
0.5932163899956328,
|
| 66 |
0.5945930913000396,
|
| 67 |
0.5991491057656028,
|
| 68 |
-
0.6038948570959496
|
|
|
|
| 69 |
],
|
| 70 |
"ppl": [
|
| 71 |
40729.60576844635,
|
|
@@ -82,7 +86,8 @@
|
|
| 82 |
49.970068452218335,
|
| 83 |
48.41525029954994,
|
| 84 |
47.256544770774525,
|
| 85 |
-
57.93225257466221
|
|
|
|
| 86 |
],
|
| 87 |
"tf_acc": [
|
| 88 |
9.249682110189831e-06,
|
|
@@ -99,7 +104,8 @@
|
|
| 99 |
0.47160701256487697,
|
| 100 |
0.4732986161931464,
|
| 101 |
0.47500545459195026,
|
| 102 |
-
0.46299773491608087
|
|
|
|
| 103 |
],
|
| 104 |
"fr_acc": [
|
| 105 |
0.0,
|
|
@@ -116,7 +122,8 @@
|
|
| 116 |
0.17273634607377905,
|
| 117 |
0.17980458818168835,
|
| 118 |
0.14786793252599717,
|
| 119 |
-
0.18805116826841511
|
|
|
|
| 120 |
],
|
| 121 |
"def_rate": [
|
| 122 |
0.0,
|
|
@@ -133,6 +140,7 @@
|
|
| 133 |
1.0,
|
| 134 |
1.0,
|
| 135 |
0.95,
|
|
|
|
| 136 |
1.0
|
| 137 |
],
|
| 138 |
"mrr": [
|
|
@@ -150,7 +158,8 @@
|
|
| 150 |
0.004974262323230505,
|
| 151 |
0.004978574346750975,
|
| 152 |
0.004983400925993919,
|
| 153 |
-
0.00775557104498148
|
|
|
|
| 154 |
],
|
| 155 |
"r1": [
|
| 156 |
0.0006249999860301614,
|
|
@@ -167,6 +176,7 @@
|
|
| 167 |
0.0006249999860301614,
|
| 168 |
0.0006249999860301614,
|
| 169 |
0.0006249999860301614,
|
|
|
|
| 170 |
0.0010416667209938169
|
| 171 |
],
|
| 172 |
"ret_ok_pct": [
|
|
@@ -184,6 +194,7 @@
|
|
| 184 |
0.0,
|
| 185 |
0.0,
|
| 186 |
0.0,
|
|
|
|
| 187 |
100.0
|
| 188 |
],
|
| 189 |
"grad_norm": [
|
|
@@ -201,6 +212,7 @@
|
|
| 201 |
995.1202453779451,
|
| 202 |
1172.9177006188306,
|
| 203 |
1544.1169985122392,
|
|
|
|
| 204 |
Infinity
|
| 205 |
],
|
| 206 |
"step_var": [
|
|
@@ -218,7 +230,8 @@
|
|
| 218 |
0.23860350251197815,
|
| 219 |
0.2396431863307953,
|
| 220 |
0.24049368500709534,
|
| 221 |
-
0.2347075343132019
|
|
|
|
| 222 |
],
|
| 223 |
"pos_slope": [
|
| 224 |
0.49990350008010864,
|
|
@@ -235,7 +248,8 @@
|
|
| 235 |
0.49595996737480164,
|
| 236 |
0.49562159180641174,
|
| 237 |
0.49534475803375244,
|
| 238 |
-
0.4971509277820587
|
|
|
|
| 239 |
],
|
| 240 |
"beta_mean": [
|
| 241 |
0.09998118132352829,
|
|
@@ -252,7 +266,8 @@
|
|
| 252 |
0.09918887913227081,
|
| 253 |
0.09912104904651642,
|
| 254 |
0.09906545281410217,
|
| 255 |
-
0.09942715615034103
|
|
|
|
| 256 |
],
|
| 257 |
"w_ret": [
|
| 258 |
0.0,
|
|
@@ -269,9 +284,11 @@
|
|
| 269 |
0.2,
|
| 270 |
0.2,
|
| 271 |
0.2,
|
|
|
|
| 272 |
0.2
|
| 273 |
],
|
| 274 |
"ret_raw": [
|
| 275 |
-
297885048197.7794
|
|
|
|
| 276 |
]
|
| 277 |
}
|
|
|
|
| 14 |
12,
|
| 15 |
13,
|
| 16 |
14,
|
| 17 |
+
15,
|
| 18 |
+
16
|
| 19 |
],
|
| 20 |
"gen_loss": [
|
| 21 |
18.370433433850607,
|
|
|
|
| 32 |
11.003662894566855,
|
| 33 |
10.940823611346158,
|
| 34 |
10.90196396784349,
|
| 35 |
+
11.473219166524483,
|
| 36 |
+
11.477066975073381
|
| 37 |
],
|
| 38 |
"ret_loss": [
|
| 39 |
0.0,
|
|
|
|
| 50 |
0.0,
|
| 51 |
0.0,
|
| 52 |
0.0,
|
| 53 |
+
297885048197.7794,
|
| 54 |
+
185471042979.8594
|
| 55 |
],
|
| 56 |
"ent_loss": [
|
| 57 |
0.6078802872065342,
|
|
|
|
| 68 |
0.5932163899956328,
|
| 69 |
0.5945930913000396,
|
| 70 |
0.5991491057656028,
|
| 71 |
+
0.6038948570959496,
|
| 72 |
+
0.6058890042521737
|
| 73 |
],
|
| 74 |
"ppl": [
|
| 75 |
40729.60576844635,
|
|
|
|
| 86 |
49.970068452218335,
|
| 87 |
48.41525029954994,
|
| 88 |
47.256544770774525,
|
| 89 |
+
57.93225257466221,
|
| 90 |
+
58.00014256149768
|
| 91 |
],
|
| 92 |
"tf_acc": [
|
| 93 |
9.249682110189831e-06,
|
|
|
|
| 104 |
0.47160701256487697,
|
| 105 |
0.4732986161931464,
|
| 106 |
0.47500545459195026,
|
| 107 |
+
0.46299773491608087,
|
| 108 |
+
0.4629857647392324
|
| 109 |
],
|
| 110 |
"fr_acc": [
|
| 111 |
0.0,
|
|
|
|
| 122 |
0.17273634607377905,
|
| 123 |
0.17980458818168835,
|
| 124 |
0.14786793252599717,
|
| 125 |
+
0.18805116826841511,
|
| 126 |
+
0.18621015772268792
|
| 127 |
],
|
| 128 |
"def_rate": [
|
| 129 |
0.0,
|
|
|
|
| 140 |
1.0,
|
| 141 |
1.0,
|
| 142 |
0.95,
|
| 143 |
+
1.0,
|
| 144 |
1.0
|
| 145 |
],
|
| 146 |
"mrr": [
|
|
|
|
| 158 |
0.004974262323230505,
|
| 159 |
0.004978574346750975,
|
| 160 |
0.004983400925993919,
|
| 161 |
+
0.00775557104498148,
|
| 162 |
+
0.007759490050375462
|
| 163 |
],
|
| 164 |
"r1": [
|
| 165 |
0.0006249999860301614,
|
|
|
|
| 176 |
0.0006249999860301614,
|
| 177 |
0.0006249999860301614,
|
| 178 |
0.0006249999860301614,
|
| 179 |
+
0.0010416667209938169,
|
| 180 |
0.0010416667209938169
|
| 181 |
],
|
| 182 |
"ret_ok_pct": [
|
|
|
|
| 194 |
0.0,
|
| 195 |
0.0,
|
| 196 |
0.0,
|
| 197 |
+
100.0,
|
| 198 |
100.0
|
| 199 |
],
|
| 200 |
"grad_norm": [
|
|
|
|
| 212 |
995.1202453779451,
|
| 213 |
1172.9177006188306,
|
| 214 |
1544.1169985122392,
|
| 215 |
+
Infinity,
|
| 216 |
Infinity
|
| 217 |
],
|
| 218 |
"step_var": [
|
|
|
|
| 230 |
0.23860350251197815,
|
| 231 |
0.2396431863307953,
|
| 232 |
0.24049368500709534,
|
| 233 |
+
0.2347075343132019,
|
| 234 |
+
0.23476119339466095
|
| 235 |
],
|
| 236 |
"pos_slope": [
|
| 237 |
0.49990350008010864,
|
|
|
|
| 248 |
0.49595996737480164,
|
| 249 |
0.49562159180641174,
|
| 250 |
0.49534475803375244,
|
| 251 |
+
0.4971509277820587,
|
| 252 |
+
0.4969623386859894
|
| 253 |
],
|
| 254 |
"beta_mean": [
|
| 255 |
0.09998118132352829,
|
|
|
|
| 266 |
0.09918887913227081,
|
| 267 |
0.09912104904651642,
|
| 268 |
0.09906545281410217,
|
| 269 |
+
0.09942715615034103,
|
| 270 |
+
0.0993889719247818
|
| 271 |
],
|
| 272 |
"w_ret": [
|
| 273 |
0.0,
|
|
|
|
| 284 |
0.2,
|
| 285 |
0.2,
|
| 286 |
0.2,
|
| 287 |
+
0.2,
|
| 288 |
0.2
|
| 289 |
],
|
| 290 |
"ret_raw": [
|
| 291 |
+
297885048197.7794,
|
| 292 |
+
185471042979.8594
|
| 293 |
]
|
| 294 |
}
|