Upload logs/checkpoint-800/trainer_state.json with huggingface_hub
Browse files
logs/checkpoint-800/trainer_state.json
ADDED
|
@@ -0,0 +1,731 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 550,
|
| 3 |
+
"best_metric": 0.027798546478152275,
|
| 4 |
+
"best_model_checkpoint": "/workspace/lampAI/finetuning/outputs/lamp-qwen-1.5b-full/checkpoint-550",
|
| 5 |
+
"epoch": 5.634920634920634,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 800,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.07054673721340388,
|
| 14 |
+
"grad_norm": 4.5,
|
| 15 |
+
"learning_rate": 1.267605633802817e-05,
|
| 16 |
+
"loss": 1.1558,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.14109347442680775,
|
| 21 |
+
"grad_norm": 1.1953125,
|
| 22 |
+
"learning_rate": 2.676056338028169e-05,
|
| 23 |
+
"loss": 0.4274,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.21164021164021163,
|
| 28 |
+
"grad_norm": 0.302734375,
|
| 29 |
+
"learning_rate": 4.0845070422535214e-05,
|
| 30 |
+
"loss": 0.0549,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.2821869488536155,
|
| 35 |
+
"grad_norm": 1.21875,
|
| 36 |
+
"learning_rate": 5.492957746478874e-05,
|
| 37 |
+
"loss": 0.0471,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.3527336860670194,
|
| 42 |
+
"grad_norm": 0.2119140625,
|
| 43 |
+
"learning_rate": 6.901408450704226e-05,
|
| 44 |
+
"loss": 0.0402,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.3527336860670194,
|
| 49 |
+
"eval_loss": 0.03969533368945122,
|
| 50 |
+
"eval_runtime": 6.3012,
|
| 51 |
+
"eval_samples_per_second": 40.151,
|
| 52 |
+
"eval_steps_per_second": 10.157,
|
| 53 |
+
"step": 50
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.42328042328042326,
|
| 57 |
+
"grad_norm": 0.193359375,
|
| 58 |
+
"learning_rate": 8.309859154929578e-05,
|
| 59 |
+
"loss": 0.0377,
|
| 60 |
+
"step": 60
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.49382716049382713,
|
| 64 |
+
"grad_norm": 0.1787109375,
|
| 65 |
+
"learning_rate": 9.718309859154931e-05,
|
| 66 |
+
"loss": 0.0343,
|
| 67 |
+
"step": 70
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.564373897707231,
|
| 71 |
+
"grad_norm": 0.19921875,
|
| 72 |
+
"learning_rate": 0.00011126760563380282,
|
| 73 |
+
"loss": 0.0356,
|
| 74 |
+
"step": 80
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.6349206349206349,
|
| 78 |
+
"grad_norm": 3.1875,
|
| 79 |
+
"learning_rate": 0.00012535211267605635,
|
| 80 |
+
"loss": 0.0855,
|
| 81 |
+
"step": 90
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.7054673721340388,
|
| 85 |
+
"grad_norm": 0.66796875,
|
| 86 |
+
"learning_rate": 0.00013943661971830987,
|
| 87 |
+
"loss": 0.0868,
|
| 88 |
+
"step": 100
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.7054673721340388,
|
| 92 |
+
"eval_loss": 0.05627487599849701,
|
| 93 |
+
"eval_runtime": 6.311,
|
| 94 |
+
"eval_samples_per_second": 40.089,
|
| 95 |
+
"eval_steps_per_second": 10.141,
|
| 96 |
+
"step": 100
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.7760141093474426,
|
| 100 |
+
"grad_norm": 0.8046875,
|
| 101 |
+
"learning_rate": 0.00015352112676056339,
|
| 102 |
+
"loss": 0.0538,
|
| 103 |
+
"step": 110
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.8465608465608465,
|
| 107 |
+
"grad_norm": 0.5234375,
|
| 108 |
+
"learning_rate": 0.0001676056338028169,
|
| 109 |
+
"loss": 0.0561,
|
| 110 |
+
"step": 120
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.9171075837742504,
|
| 114 |
+
"grad_norm": 1.8828125,
|
| 115 |
+
"learning_rate": 0.00018169014084507045,
|
| 116 |
+
"loss": 0.0641,
|
| 117 |
+
"step": 130
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.9876543209876543,
|
| 121 |
+
"grad_norm": 37.0,
|
| 122 |
+
"learning_rate": 0.00019577464788732396,
|
| 123 |
+
"loss": 0.0863,
|
| 124 |
+
"step": 140
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 1.056437389770723,
|
| 128 |
+
"grad_norm": 1.0546875,
|
| 129 |
+
"learning_rate": 0.00019999667815369528,
|
| 130 |
+
"loss": 0.3147,
|
| 131 |
+
"step": 150
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.056437389770723,
|
| 135 |
+
"eval_loss": 0.10409189015626907,
|
| 136 |
+
"eval_runtime": 6.0488,
|
| 137 |
+
"eval_samples_per_second": 41.826,
|
| 138 |
+
"eval_steps_per_second": 10.581,
|
| 139 |
+
"step": 150
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 1.126984126984127,
|
| 143 |
+
"grad_norm": 0.8671875,
|
| 144 |
+
"learning_rate": 0.00019998040841735952,
|
| 145 |
+
"loss": 0.0868,
|
| 146 |
+
"step": 160
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.1975308641975309,
|
| 150 |
+
"grad_norm": 0.2109375,
|
| 151 |
+
"learning_rate": 0.00019995058285912832,
|
| 152 |
+
"loss": 0.0527,
|
| 153 |
+
"step": 170
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 1.2680776014109347,
|
| 157 |
+
"grad_norm": 0.1435546875,
|
| 158 |
+
"learning_rate": 0.00019990720552289347,
|
| 159 |
+
"loss": 0.0418,
|
| 160 |
+
"step": 180
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 1.3386243386243386,
|
| 164 |
+
"grad_norm": 0.09912109375,
|
| 165 |
+
"learning_rate": 0.00019985028228996173,
|
| 166 |
+
"loss": 0.0328,
|
| 167 |
+
"step": 190
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.4091710758377425,
|
| 171 |
+
"grad_norm": 0.12890625,
|
| 172 |
+
"learning_rate": 0.00019977982087825713,
|
| 173 |
+
"loss": 0.0389,
|
| 174 |
+
"step": 200
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 1.4091710758377425,
|
| 178 |
+
"eval_loss": 0.03491974622011185,
|
| 179 |
+
"eval_runtime": 6.042,
|
| 180 |
+
"eval_samples_per_second": 41.874,
|
| 181 |
+
"eval_steps_per_second": 10.593,
|
| 182 |
+
"step": 200
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.4797178130511464,
|
| 186 |
+
"grad_norm": 0.10498046875,
|
| 187 |
+
"learning_rate": 0.00019969583084127485,
|
| 188 |
+
"loss": 0.0338,
|
| 189 |
+
"step": 210
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 1.5502645502645502,
|
| 193 |
+
"grad_norm": 0.0859375,
|
| 194 |
+
"learning_rate": 0.00019959832356678583,
|
| 195 |
+
"loss": 0.033,
|
| 196 |
+
"step": 220
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 1.620811287477954,
|
| 200 |
+
"grad_norm": 0.12255859375,
|
| 201 |
+
"learning_rate": 0.00019948731227529258,
|
| 202 |
+
"loss": 0.0313,
|
| 203 |
+
"step": 230
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.691358024691358,
|
| 207 |
+
"grad_norm": 0.10400390625,
|
| 208 |
+
"learning_rate": 0.00019936281201823688,
|
| 209 |
+
"loss": 0.0288,
|
| 210 |
+
"step": 240
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 1.7619047619047619,
|
| 214 |
+
"grad_norm": 0.1396484375,
|
| 215 |
+
"learning_rate": 0.00019922483967595893,
|
| 216 |
+
"loss": 0.0322,
|
| 217 |
+
"step": 250
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 1.7619047619047619,
|
| 221 |
+
"eval_loss": 0.03154641017317772,
|
| 222 |
+
"eval_runtime": 6.0077,
|
| 223 |
+
"eval_samples_per_second": 42.113,
|
| 224 |
+
"eval_steps_per_second": 10.653,
|
| 225 |
+
"step": 250
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 1.8324514991181657,
|
| 229 |
+
"grad_norm": 0.095703125,
|
| 230 |
+
"learning_rate": 0.00019907341395540877,
|
| 231 |
+
"loss": 0.0279,
|
| 232 |
+
"step": 260
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 1.9029982363315696,
|
| 236 |
+
"grad_norm": 0.10107421875,
|
| 237 |
+
"learning_rate": 0.00019890855538760974,
|
| 238 |
+
"loss": 0.0293,
|
| 239 |
+
"step": 270
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.9735449735449735,
|
| 243 |
+
"grad_norm": 0.1015625,
|
| 244 |
+
"learning_rate": 0.00019873028632487474,
|
| 245 |
+
"loss": 0.029,
|
| 246 |
+
"step": 280
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 2.0423280423280423,
|
| 250 |
+
"grad_norm": 0.1025390625,
|
| 251 |
+
"learning_rate": 0.0001985386309377759,
|
| 252 |
+
"loss": 0.0265,
|
| 253 |
+
"step": 290
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 2.112874779541446,
|
| 257 |
+
"grad_norm": 0.11962890625,
|
| 258 |
+
"learning_rate": 0.0001983336152118671,
|
| 259 |
+
"loss": 0.0256,
|
| 260 |
+
"step": 300
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.112874779541446,
|
| 264 |
+
"eval_loss": 0.029803331941366196,
|
| 265 |
+
"eval_runtime": 5.9968,
|
| 266 |
+
"eval_samples_per_second": 42.189,
|
| 267 |
+
"eval_steps_per_second": 10.672,
|
| 268 |
+
"step": 300
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 2.18342151675485,
|
| 272 |
+
"grad_norm": 0.09130859375,
|
| 273 |
+
"learning_rate": 0.0001981152669441609,
|
| 274 |
+
"loss": 0.0247,
|
| 275 |
+
"step": 310
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 2.253968253968254,
|
| 279 |
+
"grad_norm": 0.10009765625,
|
| 280 |
+
"learning_rate": 0.00019788361573935958,
|
| 281 |
+
"loss": 0.0257,
|
| 282 |
+
"step": 320
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 2.324514991181658,
|
| 286 |
+
"grad_norm": 0.10498046875,
|
| 287 |
+
"learning_rate": 0.00019763869300584128,
|
| 288 |
+
"loss": 0.0254,
|
| 289 |
+
"step": 330
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 2.3950617283950617,
|
| 293 |
+
"grad_norm": 0.0986328125,
|
| 294 |
+
"learning_rate": 0.00019738053195140148,
|
| 295 |
+
"loss": 0.023,
|
| 296 |
+
"step": 340
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.4656084656084656,
|
| 300 |
+
"grad_norm": 0.0888671875,
|
| 301 |
+
"learning_rate": 0.00019710916757875052,
|
| 302 |
+
"loss": 0.0243,
|
| 303 |
+
"step": 350
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 2.4656084656084656,
|
| 307 |
+
"eval_loss": 0.029007520526647568,
|
| 308 |
+
"eval_runtime": 6.0663,
|
| 309 |
+
"eval_samples_per_second": 41.706,
|
| 310 |
+
"eval_steps_per_second": 10.55,
|
| 311 |
+
"step": 350
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 2.5361552028218695,
|
| 315 |
+
"grad_norm": 0.099609375,
|
| 316 |
+
"learning_rate": 0.0001968246366807677,
|
| 317 |
+
"loss": 0.0232,
|
| 318 |
+
"step": 360
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 2.6067019400352733,
|
| 322 |
+
"grad_norm": 0.10595703125,
|
| 323 |
+
"learning_rate": 0.0001965269778355129,
|
| 324 |
+
"loss": 0.0253,
|
| 325 |
+
"step": 370
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.677248677248677,
|
| 329 |
+
"grad_norm": 0.0927734375,
|
| 330 |
+
"learning_rate": 0.00019621623140099578,
|
| 331 |
+
"loss": 0.0253,
|
| 332 |
+
"step": 380
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.747795414462081,
|
| 336 |
+
"grad_norm": 0.099609375,
|
| 337 |
+
"learning_rate": 0.00019589243950970402,
|
| 338 |
+
"loss": 0.0231,
|
| 339 |
+
"step": 390
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 2.818342151675485,
|
| 343 |
+
"grad_norm": 0.08984375,
|
| 344 |
+
"learning_rate": 0.0001955556460628906,
|
| 345 |
+
"loss": 0.0237,
|
| 346 |
+
"step": 400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 2.818342151675485,
|
| 350 |
+
"eval_loss": 0.027823707088828087,
|
| 351 |
+
"eval_runtime": 6.0126,
|
| 352 |
+
"eval_samples_per_second": 42.078,
|
| 353 |
+
"eval_steps_per_second": 10.644,
|
| 354 |
+
"step": 400
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 2.888888888888889,
|
| 358 |
+
"grad_norm": 0.10009765625,
|
| 359 |
+
"learning_rate": 0.0001952058967246217,
|
| 360 |
+
"loss": 0.025,
|
| 361 |
+
"step": 410
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 2.9594356261022927,
|
| 365 |
+
"grad_norm": 0.11572265625,
|
| 366 |
+
"learning_rate": 0.00019484323891558506,
|
| 367 |
+
"loss": 0.0238,
|
| 368 |
+
"step": 420
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 3.0282186948853616,
|
| 372 |
+
"grad_norm": 0.083984375,
|
| 373 |
+
"learning_rate": 0.00019446772180666084,
|
| 374 |
+
"loss": 0.0206,
|
| 375 |
+
"step": 430
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 3.0987654320987654,
|
| 379 |
+
"grad_norm": 0.0986328125,
|
| 380 |
+
"learning_rate": 0.00019407939631225439,
|
| 381 |
+
"loss": 0.019,
|
| 382 |
+
"step": 440
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 3.1693121693121693,
|
| 386 |
+
"grad_norm": 0.10400390625,
|
| 387 |
+
"learning_rate": 0.00019367831508339327,
|
| 388 |
+
"loss": 0.0186,
|
| 389 |
+
"step": 450
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 3.1693121693121693,
|
| 393 |
+
"eval_loss": 0.029222065582871437,
|
| 394 |
+
"eval_runtime": 5.9948,
|
| 395 |
+
"eval_samples_per_second": 42.203,
|
| 396 |
+
"eval_steps_per_second": 10.676,
|
| 397 |
+
"step": 450
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 3.239858906525573,
|
| 401 |
+
"grad_norm": 0.09130859375,
|
| 402 |
+
"learning_rate": 0.0001932645325005885,
|
| 403 |
+
"loss": 0.0167,
|
| 404 |
+
"step": 460
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 3.310405643738977,
|
| 408 |
+
"grad_norm": 0.10107421875,
|
| 409 |
+
"learning_rate": 0.0001928381046664615,
|
| 410 |
+
"loss": 0.0208,
|
| 411 |
+
"step": 470
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 3.380952380952381,
|
| 415 |
+
"grad_norm": 0.119140625,
|
| 416 |
+
"learning_rate": 0.00019239908939813722,
|
| 417 |
+
"loss": 0.0197,
|
| 418 |
+
"step": 480
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 3.451499118165785,
|
| 422 |
+
"grad_norm": 0.10009765625,
|
| 423 |
+
"learning_rate": 0.0001919475462194052,
|
| 424 |
+
"loss": 0.0199,
|
| 425 |
+
"step": 490
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 3.5220458553791887,
|
| 429 |
+
"grad_norm": 0.0908203125,
|
| 430 |
+
"learning_rate": 0.00019148353635264895,
|
| 431 |
+
"loss": 0.0209,
|
| 432 |
+
"step": 500
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 3.5220458553791887,
|
| 436 |
+
"eval_loss": 0.028565241023898125,
|
| 437 |
+
"eval_runtime": 6.5453,
|
| 438 |
+
"eval_samples_per_second": 38.654,
|
| 439 |
+
"eval_steps_per_second": 9.778,
|
| 440 |
+
"step": 500
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 3.5925925925925926,
|
| 444 |
+
"grad_norm": 0.1103515625,
|
| 445 |
+
"learning_rate": 0.00019100712271054516,
|
| 446 |
+
"loss": 0.0215,
|
| 447 |
+
"step": 510
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 3.6631393298059964,
|
| 451 |
+
"grad_norm": 0.10400390625,
|
| 452 |
+
"learning_rate": 0.00019051836988753372,
|
| 453 |
+
"loss": 0.022,
|
| 454 |
+
"step": 520
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 3.7336860670194003,
|
| 458 |
+
"grad_norm": 0.10791015625,
|
| 459 |
+
"learning_rate": 0.0001900173441510597,
|
| 460 |
+
"loss": 0.02,
|
| 461 |
+
"step": 530
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 3.804232804232804,
|
| 465 |
+
"grad_norm": 0.0732421875,
|
| 466 |
+
"learning_rate": 0.00018950411343258842,
|
| 467 |
+
"loss": 0.0201,
|
| 468 |
+
"step": 540
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 3.874779541446208,
|
| 472 |
+
"grad_norm": 0.0849609375,
|
| 473 |
+
"learning_rate": 0.00018897874731839504,
|
| 474 |
+
"loss": 0.0208,
|
| 475 |
+
"step": 550
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 3.874779541446208,
|
| 479 |
+
"eval_loss": 0.027798546478152275,
|
| 480 |
+
"eval_runtime": 6.1662,
|
| 481 |
+
"eval_samples_per_second": 41.03,
|
| 482 |
+
"eval_steps_per_second": 10.379,
|
| 483 |
+
"step": 550
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 3.945326278659612,
|
| 487 |
+
"grad_norm": 0.10302734375,
|
| 488 |
+
"learning_rate": 0.00018844131704012968,
|
| 489 |
+
"loss": 0.0202,
|
| 490 |
+
"step": 560
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 4.01410934744268,
|
| 494 |
+
"grad_norm": 0.08984375,
|
| 495 |
+
"learning_rate": 0.00018789189546515958,
|
| 496 |
+
"loss": 0.0199,
|
| 497 |
+
"step": 570
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 4.084656084656085,
|
| 501 |
+
"grad_norm": 0.09912109375,
|
| 502 |
+
"learning_rate": 0.00018733055708668926,
|
| 503 |
+
"loss": 0.0143,
|
| 504 |
+
"step": 580
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 4.155202821869489,
|
| 508 |
+
"grad_norm": 0.0927734375,
|
| 509 |
+
"learning_rate": 0.00018675737801366056,
|
| 510 |
+
"loss": 0.0141,
|
| 511 |
+
"step": 590
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 4.225749559082892,
|
| 515 |
+
"grad_norm": 0.09912109375,
|
| 516 |
+
"learning_rate": 0.00018617243596043314,
|
| 517 |
+
"loss": 0.0158,
|
| 518 |
+
"step": 600
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 4.225749559082892,
|
| 522 |
+
"eval_loss": 0.02914673648774624,
|
| 523 |
+
"eval_runtime": 6.1782,
|
| 524 |
+
"eval_samples_per_second": 40.95,
|
| 525 |
+
"eval_steps_per_second": 10.359,
|
| 526 |
+
"step": 600
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 4.296296296296296,
|
| 530 |
+
"grad_norm": 0.11279296875,
|
| 531 |
+
"learning_rate": 0.00018557581023624788,
|
| 532 |
+
"loss": 0.0158,
|
| 533 |
+
"step": 610
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 4.3668430335097,
|
| 537 |
+
"grad_norm": 0.11083984375,
|
| 538 |
+
"learning_rate": 0.00018496758173447368,
|
| 539 |
+
"loss": 0.0166,
|
| 540 |
+
"step": 620
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 4.4373897707231045,
|
| 544 |
+
"grad_norm": 0.09521484375,
|
| 545 |
+
"learning_rate": 0.0001843478329216394,
|
| 546 |
+
"loss": 0.0171,
|
| 547 |
+
"step": 630
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 4.507936507936508,
|
| 551 |
+
"grad_norm": 0.0859375,
|
| 552 |
+
"learning_rate": 0.00018371664782625287,
|
| 553 |
+
"loss": 0.0164,
|
| 554 |
+
"step": 640
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.578483245149911,
|
| 558 |
+
"grad_norm": 0.09130859375,
|
| 559 |
+
"learning_rate": 0.00018307411202740773,
|
| 560 |
+
"loss": 0.0175,
|
| 561 |
+
"step": 650
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 4.578483245149911,
|
| 565 |
+
"eval_loss": 0.02929055318236351,
|
| 566 |
+
"eval_runtime": 6.0557,
|
| 567 |
+
"eval_samples_per_second": 41.779,
|
| 568 |
+
"eval_steps_per_second": 10.568,
|
| 569 |
+
"step": 650
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 4.649029982363316,
|
| 573 |
+
"grad_norm": 0.1083984375,
|
| 574 |
+
"learning_rate": 0.00018242031264318026,
|
| 575 |
+
"loss": 0.0179,
|
| 576 |
+
"step": 660
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 4.71957671957672,
|
| 580 |
+
"grad_norm": 0.1015625,
|
| 581 |
+
"learning_rate": 0.00018175533831881757,
|
| 582 |
+
"loss": 0.0175,
|
| 583 |
+
"step": 670
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 4.790123456790123,
|
| 587 |
+
"grad_norm": 0.1015625,
|
| 588 |
+
"learning_rate": 0.0001810792792147186,
|
| 589 |
+
"loss": 0.0177,
|
| 590 |
+
"step": 680
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 4.860670194003527,
|
| 594 |
+
"grad_norm": 0.09033203125,
|
| 595 |
+
"learning_rate": 0.00018039222699420965,
|
| 596 |
+
"loss": 0.0177,
|
| 597 |
+
"step": 690
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 4.931216931216931,
|
| 601 |
+
"grad_norm": 0.11328125,
|
| 602 |
+
"learning_rate": 0.00017969427481111643,
|
| 603 |
+
"loss": 0.0184,
|
| 604 |
+
"step": 700
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 4.931216931216931,
|
| 608 |
+
"eval_loss": 0.028933366760611534,
|
| 609 |
+
"eval_runtime": 6.028,
|
| 610 |
+
"eval_samples_per_second": 41.971,
|
| 611 |
+
"eval_steps_per_second": 10.617,
|
| 612 |
+
"step": 700
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 5.0,
|
| 616 |
+
"grad_norm": 0.126953125,
|
| 617 |
+
"learning_rate": 0.00017898551729713362,
|
| 618 |
+
"loss": 0.0166,
|
| 619 |
+
"step": 710
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 5.070546737213403,
|
| 623 |
+
"grad_norm": 0.10302734375,
|
| 624 |
+
"learning_rate": 0.00017826605054899433,
|
| 625 |
+
"loss": 0.0125,
|
| 626 |
+
"step": 720
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 5.141093474426808,
|
| 630 |
+
"grad_norm": 0.09228515625,
|
| 631 |
+
"learning_rate": 0.00017753597211544092,
|
| 632 |
+
"loss": 0.0139,
|
| 633 |
+
"step": 730
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 5.211640211640212,
|
| 637 |
+
"grad_norm": 0.09521484375,
|
| 638 |
+
"learning_rate": 0.0001767953809839987,
|
| 639 |
+
"loss": 0.0136,
|
| 640 |
+
"step": 740
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 5.2821869488536155,
|
| 644 |
+
"grad_norm": 0.103515625,
|
| 645 |
+
"learning_rate": 0.00017604437756755498,
|
| 646 |
+
"loss": 0.014,
|
| 647 |
+
"step": 750
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 5.2821869488536155,
|
| 651 |
+
"eval_loss": 0.030887732282280922,
|
| 652 |
+
"eval_runtime": 6.0316,
|
| 653 |
+
"eval_samples_per_second": 41.946,
|
| 654 |
+
"eval_steps_per_second": 10.611,
|
| 655 |
+
"step": 750
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 5.352733686067019,
|
| 659 |
+
"grad_norm": 0.087890625,
|
| 660 |
+
"learning_rate": 0.0001752830636907443,
|
| 661 |
+
"loss": 0.0139,
|
| 662 |
+
"step": 760
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 5.423280423280423,
|
| 666 |
+
"grad_norm": 0.1201171875,
|
| 667 |
+
"learning_rate": 0.00017451154257614287,
|
| 668 |
+
"loss": 0.0142,
|
| 669 |
+
"step": 770
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 5.493827160493828,
|
| 673 |
+
"grad_norm": 0.1025390625,
|
| 674 |
+
"learning_rate": 0.00017372991883027287,
|
| 675 |
+
"loss": 0.0142,
|
| 676 |
+
"step": 780
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 5.564373897707231,
|
| 680 |
+
"grad_norm": 0.0947265625,
|
| 681 |
+
"learning_rate": 0.00017293829842941972,
|
| 682 |
+
"loss": 0.0146,
|
| 683 |
+
"step": 790
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 5.634920634920634,
|
| 687 |
+
"grad_norm": 0.1025390625,
|
| 688 |
+
"learning_rate": 0.00017213678870526292,
|
| 689 |
+
"loss": 0.0147,
|
| 690 |
+
"step": 800
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 5.634920634920634,
|
| 694 |
+
"eval_loss": 0.02992367185652256,
|
| 695 |
+
"eval_runtime": 6.0048,
|
| 696 |
+
"eval_samples_per_second": 42.133,
|
| 697 |
+
"eval_steps_per_second": 10.658,
|
| 698 |
+
"step": 800
|
| 699 |
+
}
|
| 700 |
+
],
|
| 701 |
+
"logging_steps": 10,
|
| 702 |
+
"max_steps": 2840,
|
| 703 |
+
"num_input_tokens_seen": 0,
|
| 704 |
+
"num_train_epochs": 20,
|
| 705 |
+
"save_steps": 50,
|
| 706 |
+
"stateful_callbacks": {
|
| 707 |
+
"EarlyStoppingCallback": {
|
| 708 |
+
"args": {
|
| 709 |
+
"early_stopping_patience": 5,
|
| 710 |
+
"early_stopping_threshold": 0.0
|
| 711 |
+
},
|
| 712 |
+
"attributes": {
|
| 713 |
+
"early_stopping_patience_counter": 5
|
| 714 |
+
}
|
| 715 |
+
},
|
| 716 |
+
"TrainerControl": {
|
| 717 |
+
"args": {
|
| 718 |
+
"should_epoch_stop": false,
|
| 719 |
+
"should_evaluate": false,
|
| 720 |
+
"should_log": false,
|
| 721 |
+
"should_save": true,
|
| 722 |
+
"should_training_stop": true
|
| 723 |
+
},
|
| 724 |
+
"attributes": {}
|
| 725 |
+
}
|
| 726 |
+
},
|
| 727 |
+
"total_flos": 1.8482800577851085e+17,
|
| 728 |
+
"train_batch_size": 4,
|
| 729 |
+
"trial_name": null,
|
| 730 |
+
"trial_params": null
|
| 731 |
+
}
|