Add trainer_state.json
Browse files- trainer_state.json +588 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,588 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 800,
|
| 3 |
+
"best_metric": 0.5327131748199463,
|
| 4 |
+
"best_model_checkpoint": "models/generation/query/checkpoint-800",
|
| 5 |
+
"epoch": 17.5,
|
| 6 |
+
"eval_steps": 400,
|
| 7 |
+
"global_step": 3500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.25,
|
| 14 |
+
"grad_norm": 4.102665424346924,
|
| 15 |
+
"learning_rate": 8.166666666666667e-05,
|
| 16 |
+
"loss": 9.566,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.5,
|
| 21 |
+
"grad_norm": 3.9690704345703125,
|
| 22 |
+
"learning_rate": 9.996828927498017e-05,
|
| 23 |
+
"loss": 2.8146,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.75,
|
| 28 |
+
"grad_norm": 3.130918502807617,
|
| 29 |
+
"learning_rate": 9.983493166277486e-05,
|
| 30 |
+
"loss": 2.4341,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 1.0,
|
| 35 |
+
"grad_norm": 2.8213088512420654,
|
| 36 |
+
"learning_rate": 9.959768225002347e-05,
|
| 37 |
+
"loss": 2.3782,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.25,
|
| 42 |
+
"grad_norm": 2.7424190044403076,
|
| 43 |
+
"learning_rate": 9.925703563494947e-05,
|
| 44 |
+
"loss": 2.2545,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 1.5,
|
| 49 |
+
"grad_norm": 2.545445442199707,
|
| 50 |
+
"learning_rate": 9.881370196982982e-05,
|
| 51 |
+
"loss": 2.2209,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 1.75,
|
| 56 |
+
"grad_norm": 2.717111587524414,
|
| 57 |
+
"learning_rate": 9.826860548052725e-05,
|
| 58 |
+
"loss": 2.2043,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 2.0,
|
| 63 |
+
"grad_norm": 2.9653303623199463,
|
| 64 |
+
"learning_rate": 9.76228825397397e-05,
|
| 65 |
+
"loss": 2.1926,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 2.0,
|
| 70 |
+
"eval_loss": 0.5487204194068909,
|
| 71 |
+
"eval_runtime": 18.2132,
|
| 72 |
+
"eval_samples_per_second": 21.962,
|
| 73 |
+
"eval_steps_per_second": 5.491,
|
| 74 |
+
"step": 400
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 2.25,
|
| 78 |
+
"grad_norm": 2.5305309295654297,
|
| 79 |
+
"learning_rate": 9.687787929798317e-05,
|
| 80 |
+
"loss": 2.0739,
|
| 81 |
+
"step": 450
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 2.5,
|
| 85 |
+
"grad_norm": 2.5425493717193604,
|
| 86 |
+
"learning_rate": 9.603514887724691e-05,
|
| 87 |
+
"loss": 2.0889,
|
| 88 |
+
"step": 500
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 2.75,
|
| 92 |
+
"grad_norm": 2.6191768646240234,
|
| 93 |
+
"learning_rate": 9.509644813317144e-05,
|
| 94 |
+
"loss": 2.0782,
|
| 95 |
+
"step": 550
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 3.0,
|
| 99 |
+
"grad_norm": 2.3347673416137695,
|
| 100 |
+
"learning_rate": 9.406373399249911e-05,
|
| 101 |
+
"loss": 2.0794,
|
| 102 |
+
"step": 600
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 3.25,
|
| 106 |
+
"grad_norm": 2.630434274673462,
|
| 107 |
+
"learning_rate": 9.293915937343299e-05,
|
| 108 |
+
"loss": 1.971,
|
| 109 |
+
"step": 650
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 3.5,
|
| 113 |
+
"grad_norm": 2.6370699405670166,
|
| 114 |
+
"learning_rate": 9.172506869740849e-05,
|
| 115 |
+
"loss": 1.964,
|
| 116 |
+
"step": 700
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 3.75,
|
| 120 |
+
"grad_norm": 2.5607361793518066,
|
| 121 |
+
"learning_rate": 9.042399300163484e-05,
|
| 122 |
+
"loss": 1.9781,
|
| 123 |
+
"step": 750
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 4.0,
|
| 127 |
+
"grad_norm": 2.5376696586608887,
|
| 128 |
+
"learning_rate": 8.90386446625952e-05,
|
| 129 |
+
"loss": 1.9771,
|
| 130 |
+
"step": 800
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 4.0,
|
| 134 |
+
"eval_loss": 0.5327131748199463,
|
| 135 |
+
"eval_runtime": 18.1107,
|
| 136 |
+
"eval_samples_per_second": 22.086,
|
| 137 |
+
"eval_steps_per_second": 5.522,
|
| 138 |
+
"step": 800
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 4.25,
|
| 142 |
+
"grad_norm": 2.8353099822998047,
|
| 143 |
+
"learning_rate": 8.757191174150532e-05,
|
| 144 |
+
"loss": 1.8681,
|
| 145 |
+
"step": 850
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 4.5,
|
| 149 |
+
"grad_norm": 2.8943753242492676,
|
| 150 |
+
"learning_rate": 8.60268519635192e-05,
|
| 151 |
+
"loss": 1.8682,
|
| 152 |
+
"step": 900
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 4.75,
|
| 156 |
+
"grad_norm": 2.9507288932800293,
|
| 157 |
+
"learning_rate": 8.440668634323305e-05,
|
| 158 |
+
"loss": 1.8892,
|
| 159 |
+
"step": 950
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 5.0,
|
| 163 |
+
"grad_norm": 2.7886505126953125,
|
| 164 |
+
"learning_rate": 8.271479246977678e-05,
|
| 165 |
+
"loss": 1.8694,
|
| 166 |
+
"step": 1000
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 5.25,
|
| 170 |
+
"grad_norm": 3.3199687004089355,
|
| 171 |
+
"learning_rate": 8.095469746549172e-05,
|
| 172 |
+
"loss": 1.7663,
|
| 173 |
+
"step": 1050
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 5.5,
|
| 177 |
+
"grad_norm": 3.080099105834961,
|
| 178 |
+
"learning_rate": 7.913007063287361e-05,
|
| 179 |
+
"loss": 1.7902,
|
| 180 |
+
"step": 1100
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 5.75,
|
| 184 |
+
"grad_norm": 2.8006722927093506,
|
| 185 |
+
"learning_rate": 7.724471580511021e-05,
|
| 186 |
+
"loss": 1.7655,
|
| 187 |
+
"step": 1150
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 6.0,
|
| 191 |
+
"grad_norm": 3.0085508823394775,
|
| 192 |
+
"learning_rate": 7.530256341615994e-05,
|
| 193 |
+
"loss": 1.7722,
|
| 194 |
+
"step": 1200
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 6.0,
|
| 198 |
+
"eval_loss": 0.5345460176467896,
|
| 199 |
+
"eval_runtime": 18.3119,
|
| 200 |
+
"eval_samples_per_second": 21.844,
|
| 201 |
+
"eval_steps_per_second": 5.461,
|
| 202 |
+
"step": 1200
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 6.25,
|
| 206 |
+
"grad_norm": 3.3692119121551514,
|
| 207 |
+
"learning_rate": 7.33076623069039e-05,
|
| 208 |
+
"loss": 1.6605,
|
| 209 |
+
"step": 1250
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 6.5,
|
| 213 |
+
"grad_norm": 3.402806282043457,
|
| 214 |
+
"learning_rate": 7.126417128445263e-05,
|
| 215 |
+
"loss": 1.662,
|
| 216 |
+
"step": 1300
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 6.75,
|
| 220 |
+
"grad_norm": 3.5842807292938232,
|
| 221 |
+
"learning_rate": 6.917635045220425e-05,
|
| 222 |
+
"loss": 1.6779,
|
| 223 |
+
"step": 1350
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 7.0,
|
| 227 |
+
"grad_norm": 3.314953565597534,
|
| 228 |
+
"learning_rate": 6.704855232872843e-05,
|
| 229 |
+
"loss": 1.6842,
|
| 230 |
+
"step": 1400
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 7.25,
|
| 234 |
+
"grad_norm": 4.195573329925537,
|
| 235 |
+
"learning_rate": 6.488521277399067e-05,
|
| 236 |
+
"loss": 1.5449,
|
| 237 |
+
"step": 1450
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 7.5,
|
| 241 |
+
"grad_norm": 3.56549334526062,
|
| 242 |
+
"learning_rate": 6.26908417418333e-05,
|
| 243 |
+
"loss": 1.564,
|
| 244 |
+
"step": 1500
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 7.75,
|
| 248 |
+
"grad_norm": 3.6120212078094482,
|
| 249 |
+
"learning_rate": 6.0470013877991525e-05,
|
| 250 |
+
"loss": 1.5727,
|
| 251 |
+
"step": 1550
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 8.0,
|
| 255 |
+
"grad_norm": 3.4173295497894287,
|
| 256 |
+
"learning_rate": 5.8227358983245274e-05,
|
| 257 |
+
"loss": 1.5821,
|
| 258 |
+
"step": 1600
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 8.0,
|
| 262 |
+
"eval_loss": 0.5576403737068176,
|
| 263 |
+
"eval_runtime": 18.1898,
|
| 264 |
+
"eval_samples_per_second": 21.99,
|
| 265 |
+
"eval_steps_per_second": 5.498,
|
| 266 |
+
"step": 1600
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 8.25,
|
| 270 |
+
"grad_norm": 4.163740158081055,
|
| 271 |
+
"learning_rate": 5.5967552361588e-05,
|
| 272 |
+
"loss": 1.4497,
|
| 273 |
+
"step": 1650
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 8.5,
|
| 277 |
+
"grad_norm": 4.105224132537842,
|
| 278 |
+
"learning_rate": 5.3695305073534455e-05,
|
| 279 |
+
"loss": 1.4493,
|
| 280 |
+
"step": 1700
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 8.75,
|
| 284 |
+
"grad_norm": 4.076621055603027,
|
| 285 |
+
"learning_rate": 5.141535411488584e-05,
|
| 286 |
+
"loss": 1.4613,
|
| 287 |
+
"step": 1750
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 9.0,
|
| 291 |
+
"grad_norm": 4.06991720199585,
|
| 292 |
+
"learning_rate": 4.913245254142751e-05,
|
| 293 |
+
"loss": 1.475,
|
| 294 |
+
"step": 1800
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 9.25,
|
| 298 |
+
"grad_norm": 4.908267021179199,
|
| 299 |
+
"learning_rate": 4.685135956014587e-05,
|
| 300 |
+
"loss": 1.3275,
|
| 301 |
+
"step": 1850
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 9.5,
|
| 305 |
+
"grad_norm": 4.249743938446045,
|
| 306 |
+
"learning_rate": 4.4576830607621834e-05,
|
| 307 |
+
"loss": 1.3592,
|
| 308 |
+
"step": 1900
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 9.75,
|
| 312 |
+
"grad_norm": 4.3078932762146,
|
| 313 |
+
"learning_rate": 4.231360743628464e-05,
|
| 314 |
+
"loss": 1.3674,
|
| 315 |
+
"step": 1950
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 10.0,
|
| 319 |
+
"grad_norm": 4.63353157043457,
|
| 320 |
+
"learning_rate": 4.00664082291931e-05,
|
| 321 |
+
"loss": 1.3652,
|
| 322 |
+
"step": 2000
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 10.0,
|
| 326 |
+
"eval_loss": 0.6032015681266785,
|
| 327 |
+
"eval_runtime": 18.102,
|
| 328 |
+
"eval_samples_per_second": 22.097,
|
| 329 |
+
"eval_steps_per_second": 5.524,
|
| 330 |
+
"step": 2000
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 10.25,
|
| 334 |
+
"grad_norm": 4.48530912399292,
|
| 335 |
+
"learning_rate": 3.78399177639524e-05,
|
| 336 |
+
"loss": 1.247,
|
| 337 |
+
"step": 2050
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 10.5,
|
| 341 |
+
"grad_norm": 4.907229423522949,
|
| 342 |
+
"learning_rate": 3.563877764627195e-05,
|
| 343 |
+
"loss": 1.2537,
|
| 344 |
+
"step": 2100
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 10.75,
|
| 348 |
+
"grad_norm": 5.136667728424072,
|
| 349 |
+
"learning_rate": 3.34675766335243e-05,
|
| 350 |
+
"loss": 1.2556,
|
| 351 |
+
"step": 2150
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 11.0,
|
| 355 |
+
"grad_norm": 4.633574962615967,
|
| 356 |
+
"learning_rate": 3.13308410684782e-05,
|
| 357 |
+
"loss": 1.2624,
|
| 358 |
+
"step": 2200
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 11.25,
|
| 362 |
+
"grad_norm": 4.862861156463623,
|
| 363 |
+
"learning_rate": 2.9233025443148317e-05,
|
| 364 |
+
"loss": 1.1407,
|
| 365 |
+
"step": 2250
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 11.5,
|
| 369 |
+
"grad_norm": 5.060881614685059,
|
| 370 |
+
"learning_rate": 2.7178503112433672e-05,
|
| 371 |
+
"loss": 1.1681,
|
| 372 |
+
"step": 2300
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 11.75,
|
| 376 |
+
"grad_norm": 5.76928186416626,
|
| 377 |
+
"learning_rate": 2.517155717690404e-05,
|
| 378 |
+
"loss": 1.1701,
|
| 379 |
+
"step": 2350
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 12.0,
|
| 383 |
+
"grad_norm": 5.6520094871521,
|
| 384 |
+
"learning_rate": 2.3216371553741295e-05,
|
| 385 |
+
"loss": 1.1729,
|
| 386 |
+
"step": 2400
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 12.0,
|
| 390 |
+
"eval_loss": 0.6796687245368958,
|
| 391 |
+
"eval_runtime": 18.1622,
|
| 392 |
+
"eval_samples_per_second": 22.024,
|
| 393 |
+
"eval_steps_per_second": 5.506,
|
| 394 |
+
"step": 2400
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 12.25,
|
| 398 |
+
"grad_norm": 5.579859256744385,
|
| 399 |
+
"learning_rate": 2.131702225445008e-05,
|
| 400 |
+
"loss": 1.0688,
|
| 401 |
+
"step": 2450
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 12.5,
|
| 405 |
+
"grad_norm": 5.693879127502441,
|
| 406 |
+
"learning_rate": 1.9477468887521627e-05,
|
| 407 |
+
"loss": 1.0746,
|
| 408 |
+
"step": 2500
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 12.75,
|
| 412 |
+
"grad_norm": 5.642828941345215,
|
| 413 |
+
"learning_rate": 1.770154640376479e-05,
|
| 414 |
+
"loss": 1.0922,
|
| 415 |
+
"step": 2550
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 13.0,
|
| 419 |
+
"grad_norm": 5.785156726837158,
|
| 420 |
+
"learning_rate": 1.5992957101513524e-05,
|
| 421 |
+
"loss": 1.0954,
|
| 422 |
+
"step": 2600
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 13.25,
|
| 426 |
+
"grad_norm": 5.720382213592529,
|
| 427 |
+
"learning_rate": 1.4355262908377271e-05,
|
| 428 |
+
"loss": 1.0076,
|
| 429 |
+
"step": 2650
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 13.5,
|
| 433 |
+
"grad_norm": 5.837440490722656,
|
| 434 |
+
"learning_rate": 1.2791877955624859e-05,
|
| 435 |
+
"loss": 1.0198,
|
| 436 |
+
"step": 2700
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 13.75,
|
| 440 |
+
"grad_norm": 5.974554061889648,
|
| 441 |
+
"learning_rate": 1.1306061460682072e-05,
|
| 442 |
+
"loss": 1.0281,
|
| 443 |
+
"step": 2750
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 14.0,
|
| 447 |
+
"grad_norm": 6.06301736831665,
|
| 448 |
+
"learning_rate": 9.90091093258102e-06,
|
| 449 |
+
"loss": 1.0109,
|
| 450 |
+
"step": 2800
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 14.0,
|
| 454 |
+
"eval_loss": 0.7651693820953369,
|
| 455 |
+
"eval_runtime": 18.0723,
|
| 456 |
+
"eval_samples_per_second": 22.133,
|
| 457 |
+
"eval_steps_per_second": 5.533,
|
| 458 |
+
"step": 2800
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 14.25,
|
| 462 |
+
"grad_norm": 6.1065473556518555,
|
| 463 |
+
"learning_rate": 8.579355714525994e-06,
|
| 464 |
+
"loss": 0.959,
|
| 465 |
+
"step": 2850
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 14.5,
|
| 469 |
+
"grad_norm": 5.750560283660889,
|
| 470 |
+
"learning_rate": 7.3441508770376975e-06,
|
| 471 |
+
"loss": 0.9667,
|
| 472 |
+
"step": 2900
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 14.75,
|
| 476 |
+
"grad_norm": 6.242477893829346,
|
| 477 |
+
"learning_rate": 6.197871474406936e-06,
|
| 478 |
+
"loss": 0.9692,
|
| 479 |
+
"step": 2950
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 15.0,
|
| 483 |
+
"grad_norm": 6.384776592254639,
|
| 484 |
+
"learning_rate": 5.142907176431455e-06,
|
| 485 |
+
"loss": 0.9755,
|
| 486 |
+
"step": 3000
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 15.25,
|
| 490 |
+
"grad_norm": 5.764701843261719,
|
| 491 |
+
"learning_rate": 4.181457286627316e-06,
|
| 492 |
+
"loss": 0.9159,
|
| 493 |
+
"step": 3050
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 15.5,
|
| 497 |
+
"grad_norm": 5.83530855178833,
|
| 498 |
+
"learning_rate": 3.3155261573003195e-06,
|
| 499 |
+
"loss": 0.9422,
|
| 500 |
+
"step": 3100
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 15.75,
|
| 504 |
+
"grad_norm": 6.114677429199219,
|
| 505 |
+
"learning_rate": 2.5469190110357475e-06,
|
| 506 |
+
"loss": 0.939,
|
| 507 |
+
"step": 3150
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 16.0,
|
| 511 |
+
"grad_norm": 6.1888628005981445,
|
| 512 |
+
"learning_rate": 1.8772381773176417e-06,
|
| 513 |
+
"loss": 0.9391,
|
| 514 |
+
"step": 3200
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 16.0,
|
| 518 |
+
"eval_loss": 0.8231673240661621,
|
| 519 |
+
"eval_runtime": 18.0722,
|
| 520 |
+
"eval_samples_per_second": 22.133,
|
| 521 |
+
"eval_steps_per_second": 5.533,
|
| 522 |
+
"step": 3200
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 16.25,
|
| 526 |
+
"grad_norm": 6.108009338378906,
|
| 527 |
+
"learning_rate": 1.307879752122948e-06,
|
| 528 |
+
"loss": 0.907,
|
| 529 |
+
"step": 3250
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 16.5,
|
| 533 |
+
"grad_norm": 6.366396903991699,
|
| 534 |
+
"learning_rate": 8.40030687454535e-07,
|
| 535 |
+
"loss": 0.9297,
|
| 536 |
+
"step": 3300
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 16.75,
|
| 540 |
+
"grad_norm": 6.0306901931762695,
|
| 541 |
+
"learning_rate": 4.746663168804566e-07,
|
| 542 |
+
"loss": 0.9071,
|
| 543 |
+
"step": 3350
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 17.0,
|
| 547 |
+
"grad_norm": 6.2290215492248535,
|
| 548 |
+
"learning_rate": 2.1254832223808196e-07,
|
| 549 |
+
"loss": 0.9209,
|
| 550 |
+
"step": 3400
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 17.25,
|
| 554 |
+
"grad_norm": 6.210936069488525,
|
| 555 |
+
"learning_rate": 5.4223145741943983e-08,
|
| 556 |
+
"loss": 0.913,
|
| 557 |
+
"step": 3450
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 17.5,
|
| 561 |
+
"grad_norm": 6.788633823394775,
|
| 562 |
+
"learning_rate": 2.08508055765666e-11,
|
| 563 |
+
"loss": 0.9034,
|
| 564 |
+
"step": 3500
|
| 565 |
+
}
|
| 566 |
+
],
|
| 567 |
+
"logging_steps": 50,
|
| 568 |
+
"max_steps": 3500,
|
| 569 |
+
"num_input_tokens_seen": 0,
|
| 570 |
+
"num_train_epochs": 18,
|
| 571 |
+
"save_steps": 400,
|
| 572 |
+
"stateful_callbacks": {
|
| 573 |
+
"TrainerControl": {
|
| 574 |
+
"args": {
|
| 575 |
+
"should_epoch_stop": false,
|
| 576 |
+
"should_evaluate": false,
|
| 577 |
+
"should_log": false,
|
| 578 |
+
"should_save": true,
|
| 579 |
+
"should_training_stop": true
|
| 580 |
+
},
|
| 581 |
+
"attributes": {}
|
| 582 |
+
}
|
| 583 |
+
},
|
| 584 |
+
"total_flos": 4.323066342987053e+17,
|
| 585 |
+
"train_batch_size": 4,
|
| 586 |
+
"trial_name": null,
|
| 587 |
+
"trial_params": null
|
| 588 |
+
}
|