Upload g2_scalar_s0.json with huggingface_hub
Browse files- g2_scalar_s0.json +36 -208
g2_scalar_s0.json
CHANGED
|
@@ -21,242 +21,70 @@
|
|
| 21 |
"lr": 0.0003,
|
| 22 |
"clip": 1.0,
|
| 23 |
"warmup": 100,
|
| 24 |
-
"eval_events": 24
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
},
|
| 26 |
"off_law": {},
|
| 27 |
"device": "cuda",
|
| 28 |
"runs": [
|
| 29 |
{
|
| 30 |
-
"params":
|
| 31 |
-
"state_elems":
|
| 32 |
"history": [
|
| 33 |
{
|
| 34 |
-
"
|
| 35 |
-
"recall_exact_d32": 0.0,
|
| 36 |
-
"value_acc_d128": 0.0039,
|
| 37 |
-
"recall_exact_d128": 0.0,
|
| 38 |
-
"value_acc_d512": 0.0052,
|
| 39 |
-
"recall_exact_d512": 0.0,
|
| 40 |
-
"value_acc_d1024": 0.0078,
|
| 41 |
-
"recall_exact_d1024": 0.0,
|
| 42 |
-
"step": 0,
|
| 43 |
-
"train_bpb": 8.0515,
|
| 44 |
-
"grad_norm": 0.9682792425,
|
| 45 |
-
"pairs": 4,
|
| 46 |
-
"lr": 0.0003,
|
| 47 |
-
"s_per_step": 2.1819
|
| 48 |
-
},
|
| 49 |
-
{
|
| 50 |
-
"value_acc_d32": 0.0,
|
| 51 |
-
"recall_exact_d32": 0.0,
|
| 52 |
-
"value_acc_d128": 0.0,
|
| 53 |
-
"recall_exact_d128": 0.0,
|
| 54 |
-
"value_acc_d512": 0.0,
|
| 55 |
-
"recall_exact_d512": 0.0,
|
| 56 |
-
"value_acc_d1024": 0.0,
|
| 57 |
-
"recall_exact_d1024": 0.0,
|
| 58 |
-
"step": 500,
|
| 59 |
-
"train_bpb": 7.5286,
|
| 60 |
-
"grad_norm": 1.142816424,
|
| 61 |
-
"pairs": 4,
|
| 62 |
-
"lr": 0.0003,
|
| 63 |
-
"s_per_step": 0.5334
|
| 64 |
-
},
|
| 65 |
-
{
|
| 66 |
-
"value_acc_d32": 0.0013,
|
| 67 |
-
"recall_exact_d32": 0.0,
|
| 68 |
-
"value_acc_d128": 0.0013,
|
| 69 |
-
"recall_exact_d128": 0.0,
|
| 70 |
-
"value_acc_d512": 0.0,
|
| 71 |
-
"recall_exact_d512": 0.0,
|
| 72 |
-
"value_acc_d1024": 0.0065,
|
| 73 |
-
"recall_exact_d1024": 0.0,
|
| 74 |
-
"step": 1000,
|
| 75 |
-
"train_bpb": 7.2798,
|
| 76 |
-
"grad_norm": 0.9159350395,
|
| 77 |
-
"pairs": 4,
|
| 78 |
-
"lr": 0.0003,
|
| 79 |
-
"s_per_step": 0.5306
|
| 80 |
-
},
|
| 81 |
-
{
|
| 82 |
-
"value_acc_d32": 0.0039,
|
| 83 |
-
"recall_exact_d32": 0.0,
|
| 84 |
-
"value_acc_d128": 0.0026,
|
| 85 |
-
"recall_exact_d128": 0.0,
|
| 86 |
-
"value_acc_d512": 0.0039,
|
| 87 |
-
"recall_exact_d512": 0.0,
|
| 88 |
-
"value_acc_d1024": 0.0052,
|
| 89 |
-
"recall_exact_d1024": 0.0,
|
| 90 |
-
"step": 1500,
|
| 91 |
-
"train_bpb": 7.2412,
|
| 92 |
-
"grad_norm": 0.6914440989,
|
| 93 |
-
"pairs": 4,
|
| 94 |
-
"lr": 0.0003,
|
| 95 |
-
"s_per_step": 0.5298
|
| 96 |
-
},
|
| 97 |
-
{
|
| 98 |
"value_acc_d32": 0.0026,
|
| 99 |
"recall_exact_d32": 0.0,
|
|
|
|
| 100 |
"value_acc_d128": 0.0026,
|
| 101 |
"recall_exact_d128": 0.0,
|
| 102 |
-
"
|
| 103 |
-
"recall_exact_d512": 0.0,
|
| 104 |
-
"value_acc_d1024": 0.0026,
|
| 105 |
-
"recall_exact_d1024": 0.0,
|
| 106 |
-
"step": 2000,
|
| 107 |
-
"train_bpb": 7.2441,
|
| 108 |
-
"grad_norm": 1.549390197,
|
| 109 |
-
"pairs": 4,
|
| 110 |
-
"lr": 0.0003,
|
| 111 |
-
"s_per_step": 0.5293
|
| 112 |
-
},
|
| 113 |
-
{
|
| 114 |
-
"value_acc_d32": 0.0065,
|
| 115 |
-
"recall_exact_d32": 0.0,
|
| 116 |
-
"value_acc_d128": 0.0039,
|
| 117 |
-
"recall_exact_d128": 0.0,
|
| 118 |
-
"value_acc_d512": 0.0078,
|
| 119 |
-
"recall_exact_d512": 0.0,
|
| 120 |
-
"value_acc_d1024": 0.0026,
|
| 121 |
-
"recall_exact_d1024": 0.0,
|
| 122 |
-
"step": 2500,
|
| 123 |
-
"train_bpb": 7.2385,
|
| 124 |
-
"grad_norm": 1.329554915,
|
| 125 |
-
"pairs": 4,
|
| 126 |
-
"lr": 0.0003,
|
| 127 |
-
"s_per_step": 0.5289
|
| 128 |
-
},
|
| 129 |
-
{
|
| 130 |
-
"value_acc_d32": 0.0052,
|
| 131 |
-
"recall_exact_d32": 0.0,
|
| 132 |
-
"value_acc_d128": 0.0013,
|
| 133 |
-
"recall_exact_d128": 0.0,
|
| 134 |
"value_acc_d512": 0.0052,
|
| 135 |
"recall_exact_d512": 0.0,
|
|
|
|
| 136 |
"value_acc_d1024": 0.0039,
|
| 137 |
"recall_exact_d1024": 0.0,
|
| 138 |
-
"
|
| 139 |
-
"
|
| 140 |
-
"
|
| 141 |
-
"
|
| 142 |
-
"
|
| 143 |
-
"
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"value_acc_d32": 0.0026,
|
| 147 |
-
"recall_exact_d32": 0.0,
|
| 148 |
-
"value_acc_d128": 0.0052,
|
| 149 |
-
"recall_exact_d128": 0.0,
|
| 150 |
-
"value_acc_d512": 0.0078,
|
| 151 |
-
"recall_exact_d512": 0.0,
|
| 152 |
-
"value_acc_d1024": 0.0091,
|
| 153 |
-
"recall_exact_d1024": 0.0,
|
| 154 |
-
"step": 3500,
|
| 155 |
-
"train_bpb": 7.2283,
|
| 156 |
-
"grad_norm": 0.8794643283,
|
| 157 |
-
"pairs": 4,
|
| 158 |
-
"lr": 0.0003,
|
| 159 |
-
"s_per_step": 0.5285
|
| 160 |
-
},
|
| 161 |
-
{
|
| 162 |
-
"value_acc_d32": 0.0026,
|
| 163 |
-
"recall_exact_d32": 0.0,
|
| 164 |
-
"value_acc_d128": 0.0013,
|
| 165 |
-
"recall_exact_d128": 0.0,
|
| 166 |
-
"value_acc_d512": 0.0052,
|
| 167 |
-
"recall_exact_d512": 0.0,
|
| 168 |
-
"value_acc_d1024": 0.0026,
|
| 169 |
-
"recall_exact_d1024": 0.0,
|
| 170 |
-
"step": 4000,
|
| 171 |
-
"train_bpb": 7.2264,
|
| 172 |
-
"grad_norm": 1.533946037,
|
| 173 |
-
"pairs": 4,
|
| 174 |
-
"lr": 0.0003,
|
| 175 |
-
"s_per_step": 0.5284
|
| 176 |
-
},
|
| 177 |
-
{
|
| 178 |
-
"value_acc_d32": 0.0039,
|
| 179 |
-
"recall_exact_d32": 0.0,
|
| 180 |
-
"value_acc_d128": 0.0078,
|
| 181 |
-
"recall_exact_d128": 0.0,
|
| 182 |
-
"value_acc_d512": 0.0026,
|
| 183 |
-
"recall_exact_d512": 0.0,
|
| 184 |
-
"value_acc_d1024": 0.0026,
|
| 185 |
-
"recall_exact_d1024": 0.0,
|
| 186 |
-
"step": 4500,
|
| 187 |
-
"train_bpb": 7.2225,
|
| 188 |
-
"grad_norm": 0.8560543656,
|
| 189 |
-
"pairs": 4,
|
| 190 |
-
"lr": 0.0003,
|
| 191 |
-
"s_per_step": 0.5283
|
| 192 |
-
},
|
| 193 |
-
{
|
| 194 |
-
"value_acc_d32": 0.0052,
|
| 195 |
-
"recall_exact_d32": 0.0,
|
| 196 |
-
"value_acc_d128": 0.0,
|
| 197 |
-
"recall_exact_d128": 0.0,
|
| 198 |
-
"value_acc_d512": 0.0026,
|
| 199 |
-
"recall_exact_d512": 0.0,
|
| 200 |
-
"value_acc_d1024": 0.0013,
|
| 201 |
-
"recall_exact_d1024": 0.0,
|
| 202 |
-
"step": 5000,
|
| 203 |
-
"train_bpb": 7.2207,
|
| 204 |
-
"grad_norm": 1.546509385,
|
| 205 |
-
"pairs": 4,
|
| 206 |
-
"lr": 0.0003,
|
| 207 |
-
"s_per_step": 0.5283
|
| 208 |
-
},
|
| 209 |
-
{
|
| 210 |
-
"value_acc_d32": 0.0026,
|
| 211 |
-
"recall_exact_d32": 0.0,
|
| 212 |
-
"value_acc_d128": 0.0013,
|
| 213 |
-
"recall_exact_d128": 0.0,
|
| 214 |
-
"value_acc_d512": 0.0026,
|
| 215 |
-
"recall_exact_d512": 0.0,
|
| 216 |
-
"value_acc_d1024": 0.0052,
|
| 217 |
-
"recall_exact_d1024": 0.0,
|
| 218 |
-
"step": 5500,
|
| 219 |
-
"train_bpb": 7.2413,
|
| 220 |
-
"grad_norm": 5.208947182,
|
| 221 |
-
"pairs": 4,
|
| 222 |
-
"lr": 0.0003,
|
| 223 |
-
"s_per_step": 0.5282
|
| 224 |
-
},
|
| 225 |
-
{
|
| 226 |
-
"value_acc_d32": 0.0052,
|
| 227 |
-
"recall_exact_d32": 0.0,
|
| 228 |
-
"value_acc_d128": 0.0065,
|
| 229 |
-
"recall_exact_d128": 0.0,
|
| 230 |
-
"value_acc_d512": 0.0039,
|
| 231 |
-
"recall_exact_d512": 0.0,
|
| 232 |
-
"value_acc_d1024": 0.0091,
|
| 233 |
-
"recall_exact_d1024": 0.0,
|
| 234 |
-
"step": 5999,
|
| 235 |
-
"train_bpb": 7.8621,
|
| 236 |
-
"grad_norm": 0.1858101487,
|
| 237 |
"pairs": 4,
|
| 238 |
"lr": 0.0003,
|
| 239 |
-
"
|
|
|
|
| 240 |
}
|
| 241 |
],
|
| 242 |
"final": {
|
| 243 |
-
"
|
|
|
|
| 244 |
"recall_exact_d32": 0.0,
|
| 245 |
-
"
|
|
|
|
| 246 |
"recall_exact_d128": 0.0,
|
| 247 |
-
"
|
|
|
|
| 248 |
"recall_exact_d512": 0.0,
|
| 249 |
-
"
|
|
|
|
| 250 |
"recall_exact_d1024": 0.0,
|
| 251 |
-
"
|
| 252 |
-
"
|
| 253 |
-
"
|
|
|
|
|
|
|
|
|
|
| 254 |
"pairs": 4,
|
| 255 |
"lr": 0.0003,
|
| 256 |
-
"
|
|
|
|
| 257 |
}
|
| 258 |
}
|
| 259 |
],
|
| 260 |
-
"partial":
|
| 261 |
"result_file": "g2_scalar_s0.json"
|
| 262 |
}
|
|
|
|
| 21 |
"lr": 0.0003,
|
| 22 |
"clip": 1.0,
|
| 23 |
"warmup": 100,
|
| 24 |
+
"eval_events": 24,
|
| 25 |
+
"law_rev": "candm-v1",
|
| 26 |
+
"kappa": 8.0,
|
| 27 |
+
"echo_phase": 1500,
|
| 28 |
+
"echo_delay": 32
|
| 29 |
},
|
| 30 |
"off_law": {},
|
| 31 |
"device": "cuda",
|
| 32 |
"runs": [
|
| 33 |
{
|
| 34 |
+
"params": 552962,
|
| 35 |
+
"state_elems": 10176,
|
| 36 |
"history": [
|
| 37 |
{
|
| 38 |
+
"law_rev": "candm-v1",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
"value_acc_d32": 0.0026,
|
| 40 |
"recall_exact_d32": 0.0,
|
| 41 |
+
"key_acc_d32": 0.0104,
|
| 42 |
"value_acc_d128": 0.0026,
|
| 43 |
"recall_exact_d128": 0.0,
|
| 44 |
+
"key_acc_d128": 0.0052,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
"value_acc_d512": 0.0052,
|
| 46 |
"recall_exact_d512": 0.0,
|
| 47 |
+
"key_acc_d512": 0.0,
|
| 48 |
"value_acc_d1024": 0.0039,
|
| 49 |
"recall_exact_d1024": 0.0,
|
| 50 |
+
"key_acc_d1024": 0.0026,
|
| 51 |
+
"m_fro": 0.142,
|
| 52 |
+
"p_fro": 4.899,
|
| 53 |
+
"step": 0,
|
| 54 |
+
"train_bpb": 15.5027,
|
| 55 |
+
"grad_norm": 16.17715836,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
"pairs": 4,
|
| 57 |
"lr": 0.0003,
|
| 58 |
+
"eta": 1.0,
|
| 59 |
+
"s_per_step": 3.7454
|
| 60 |
}
|
| 61 |
],
|
| 62 |
"final": {
|
| 63 |
+
"law_rev": "candm-v1",
|
| 64 |
+
"value_acc_d32": 0.0026,
|
| 65 |
"recall_exact_d32": 0.0,
|
| 66 |
+
"key_acc_d32": 0.0104,
|
| 67 |
+
"value_acc_d128": 0.0026,
|
| 68 |
"recall_exact_d128": 0.0,
|
| 69 |
+
"key_acc_d128": 0.0052,
|
| 70 |
+
"value_acc_d512": 0.0052,
|
| 71 |
"recall_exact_d512": 0.0,
|
| 72 |
+
"key_acc_d512": 0.0,
|
| 73 |
+
"value_acc_d1024": 0.0039,
|
| 74 |
"recall_exact_d1024": 0.0,
|
| 75 |
+
"key_acc_d1024": 0.0026,
|
| 76 |
+
"m_fro": 0.142,
|
| 77 |
+
"p_fro": 4.899,
|
| 78 |
+
"step": 0,
|
| 79 |
+
"train_bpb": 15.5027,
|
| 80 |
+
"grad_norm": 16.17715836,
|
| 81 |
"pairs": 4,
|
| 82 |
"lr": 0.0003,
|
| 83 |
+
"eta": 1.0,
|
| 84 |
+
"s_per_step": 3.7454
|
| 85 |
}
|
| 86 |
}
|
| 87 |
],
|
| 88 |
+
"partial": true,
|
| 89 |
"result_file": "g2_scalar_s0.json"
|
| 90 |
}
|