Spaces:
Runtime error
Runtime error
Commit
·
28237f0
1
Parent(s):
5cdc0fd
Upload checkpoint-11500/trainer_state.json
Browse files
checkpoint-11500/trainer_state.json
ADDED
|
@@ -0,0 +1,752 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.122239351272583,
|
| 3 |
+
"best_model_checkpoint": "./res_1/checkpoint-8500",
|
| 4 |
+
"epoch": 3.72168284789644,
|
| 5 |
+
"global_step": 11500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.08,
|
| 12 |
+
"learning_rate": 2.5e-05,
|
| 13 |
+
"loss": 4.4764,
|
| 14 |
+
"step": 250
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.08,
|
| 18 |
+
"eval_accuracy": 0.2770019218449712,
|
| 19 |
+
"eval_f1_score": 0.15899206409534755,
|
| 20 |
+
"eval_loss": 3.3759119510650635,
|
| 21 |
+
"eval_runtime": 78.6208,
|
| 22 |
+
"eval_samples_per_second": 198.548,
|
| 23 |
+
"eval_steps_per_second": 2.073,
|
| 24 |
+
"step": 250
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.16,
|
| 28 |
+
"learning_rate": 5e-05,
|
| 29 |
+
"loss": 2.7621,
|
| 30 |
+
"step": 500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.16,
|
| 34 |
+
"eval_accuracy": 0.45624599615631006,
|
| 35 |
+
"eval_f1_score": 0.37938588394970946,
|
| 36 |
+
"eval_loss": 2.22213077545166,
|
| 37 |
+
"eval_runtime": 79.1544,
|
| 38 |
+
"eval_samples_per_second": 197.21,
|
| 39 |
+
"eval_steps_per_second": 2.059,
|
| 40 |
+
"step": 500
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.24,
|
| 44 |
+
"learning_rate": 4.89460370994941e-05,
|
| 45 |
+
"loss": 2.0369,
|
| 46 |
+
"step": 750
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.24,
|
| 50 |
+
"eval_accuracy": 0.5180653427290198,
|
| 51 |
+
"eval_f1_score": 0.4651383493082148,
|
| 52 |
+
"eval_loss": 1.8033452033996582,
|
| 53 |
+
"eval_runtime": 78.4993,
|
| 54 |
+
"eval_samples_per_second": 198.855,
|
| 55 |
+
"eval_steps_per_second": 2.076,
|
| 56 |
+
"step": 750
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.32,
|
| 60 |
+
"learning_rate": 4.7892074198988196e-05,
|
| 61 |
+
"loss": 1.7846,
|
| 62 |
+
"step": 1000
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.32,
|
| 66 |
+
"eval_accuracy": 0.5547085201793722,
|
| 67 |
+
"eval_f1_score": 0.5145917984904441,
|
| 68 |
+
"eval_loss": 1.6128593683242798,
|
| 69 |
+
"eval_runtime": 79.1904,
|
| 70 |
+
"eval_samples_per_second": 197.12,
|
| 71 |
+
"eval_steps_per_second": 2.058,
|
| 72 |
+
"step": 1000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.4,
|
| 76 |
+
"learning_rate": 4.683811129848229e-05,
|
| 77 |
+
"loss": 1.5885,
|
| 78 |
+
"step": 1250
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 0.4,
|
| 82 |
+
"eval_accuracy": 0.5733504163997437,
|
| 83 |
+
"eval_f1_score": 0.5403077333842975,
|
| 84 |
+
"eval_loss": 1.5020724534988403,
|
| 85 |
+
"eval_runtime": 79.2228,
|
| 86 |
+
"eval_samples_per_second": 197.039,
|
| 87 |
+
"eval_steps_per_second": 2.057,
|
| 88 |
+
"step": 1250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.49,
|
| 92 |
+
"learning_rate": 4.5784148397976396e-05,
|
| 93 |
+
"loss": 1.5474,
|
| 94 |
+
"step": 1500
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.49,
|
| 98 |
+
"eval_accuracy": 0.5885970531710442,
|
| 99 |
+
"eval_f1_score": 0.5598433530383174,
|
| 100 |
+
"eval_loss": 1.4353328943252563,
|
| 101 |
+
"eval_runtime": 79.3289,
|
| 102 |
+
"eval_samples_per_second": 196.776,
|
| 103 |
+
"eval_steps_per_second": 2.055,
|
| 104 |
+
"step": 1500
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.57,
|
| 108 |
+
"learning_rate": 4.4730185497470486e-05,
|
| 109 |
+
"loss": 1.4754,
|
| 110 |
+
"step": 1750
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.57,
|
| 114 |
+
"eval_accuracy": 0.5910954516335682,
|
| 115 |
+
"eval_f1_score": 0.567613249026739,
|
| 116 |
+
"eval_loss": 1.4085466861724854,
|
| 117 |
+
"eval_runtime": 79.2487,
|
| 118 |
+
"eval_samples_per_second": 196.975,
|
| 119 |
+
"eval_steps_per_second": 2.057,
|
| 120 |
+
"step": 1750
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.65,
|
| 124 |
+
"learning_rate": 4.367622259696459e-05,
|
| 125 |
+
"loss": 1.3951,
|
| 126 |
+
"step": 2000
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.65,
|
| 130 |
+
"eval_accuracy": 0.6080717488789238,
|
| 131 |
+
"eval_f1_score": 0.5881168924607102,
|
| 132 |
+
"eval_loss": 1.3425586223602295,
|
| 133 |
+
"eval_runtime": 79.2258,
|
| 134 |
+
"eval_samples_per_second": 197.032,
|
| 135 |
+
"eval_steps_per_second": 2.057,
|
| 136 |
+
"step": 2000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.73,
|
| 140 |
+
"learning_rate": 4.262225969645869e-05,
|
| 141 |
+
"loss": 1.3711,
|
| 142 |
+
"step": 2250
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.73,
|
| 146 |
+
"eval_accuracy": 0.6153747597693786,
|
| 147 |
+
"eval_f1_score": 0.5932326664510665,
|
| 148 |
+
"eval_loss": 1.3140783309936523,
|
| 149 |
+
"eval_runtime": 79.1766,
|
| 150 |
+
"eval_samples_per_second": 197.154,
|
| 151 |
+
"eval_steps_per_second": 2.059,
|
| 152 |
+
"step": 2250
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.81,
|
| 156 |
+
"learning_rate": 4.1568296795952783e-05,
|
| 157 |
+
"loss": 1.3487,
|
| 158 |
+
"step": 2500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.81,
|
| 162 |
+
"eval_accuracy": 0.6163997437540039,
|
| 163 |
+
"eval_f1_score": 0.5956180167874238,
|
| 164 |
+
"eval_loss": 1.3029619455337524,
|
| 165 |
+
"eval_runtime": 79.2987,
|
| 166 |
+
"eval_samples_per_second": 196.851,
|
| 167 |
+
"eval_steps_per_second": 2.056,
|
| 168 |
+
"step": 2500
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.89,
|
| 172 |
+
"learning_rate": 4.051433389544688e-05,
|
| 173 |
+
"loss": 1.325,
|
| 174 |
+
"step": 2750
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.89,
|
| 178 |
+
"eval_accuracy": 0.6240230621396541,
|
| 179 |
+
"eval_f1_score": 0.6027158431466506,
|
| 180 |
+
"eval_loss": 1.2634259462356567,
|
| 181 |
+
"eval_runtime": 79.2608,
|
| 182 |
+
"eval_samples_per_second": 196.945,
|
| 183 |
+
"eval_steps_per_second": 2.057,
|
| 184 |
+
"step": 2750
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.97,
|
| 188 |
+
"learning_rate": 3.9460370994940984e-05,
|
| 189 |
+
"loss": 1.307,
|
| 190 |
+
"step": 3000
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.97,
|
| 194 |
+
"eval_accuracy": 0.6235105701473415,
|
| 195 |
+
"eval_f1_score": 0.6034162405543351,
|
| 196 |
+
"eval_loss": 1.2556581497192383,
|
| 197 |
+
"eval_runtime": 79.2153,
|
| 198 |
+
"eval_samples_per_second": 197.058,
|
| 199 |
+
"eval_steps_per_second": 2.058,
|
| 200 |
+
"step": 3000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.05,
|
| 204 |
+
"learning_rate": 3.8406408094435074e-05,
|
| 205 |
+
"loss": 1.2019,
|
| 206 |
+
"step": 3250
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.05,
|
| 210 |
+
"eval_accuracy": 0.6322869955156951,
|
| 211 |
+
"eval_f1_score": 0.6147432260064974,
|
| 212 |
+
"eval_loss": 1.227844476699829,
|
| 213 |
+
"eval_runtime": 79.4547,
|
| 214 |
+
"eval_samples_per_second": 196.464,
|
| 215 |
+
"eval_steps_per_second": 2.051,
|
| 216 |
+
"step": 3250
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 1.13,
|
| 220 |
+
"learning_rate": 3.735244519392918e-05,
|
| 221 |
+
"loss": 1.1366,
|
| 222 |
+
"step": 3500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 1.13,
|
| 226 |
+
"eval_accuracy": 0.6326713645099296,
|
| 227 |
+
"eval_f1_score": 0.6163567005428549,
|
| 228 |
+
"eval_loss": 1.2262649536132812,
|
| 229 |
+
"eval_runtime": 79.2235,
|
| 230 |
+
"eval_samples_per_second": 197.037,
|
| 231 |
+
"eval_steps_per_second": 2.057,
|
| 232 |
+
"step": 3500
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 1.21,
|
| 236 |
+
"learning_rate": 3.6298482293423274e-05,
|
| 237 |
+
"loss": 1.1643,
|
| 238 |
+
"step": 3750
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 1.21,
|
| 242 |
+
"eval_accuracy": 0.6368994234465086,
|
| 243 |
+
"eval_f1_score": 0.6215204096974251,
|
| 244 |
+
"eval_loss": 1.2120453119277954,
|
| 245 |
+
"eval_runtime": 79.1999,
|
| 246 |
+
"eval_samples_per_second": 197.096,
|
| 247 |
+
"eval_steps_per_second": 2.058,
|
| 248 |
+
"step": 3750
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.29,
|
| 252 |
+
"learning_rate": 3.524451939291737e-05,
|
| 253 |
+
"loss": 1.1226,
|
| 254 |
+
"step": 4000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.29,
|
| 258 |
+
"eval_accuracy": 0.6377322229340167,
|
| 259 |
+
"eval_f1_score": 0.6225832875038563,
|
| 260 |
+
"eval_loss": 1.2013684511184692,
|
| 261 |
+
"eval_runtime": 79.3501,
|
| 262 |
+
"eval_samples_per_second": 196.723,
|
| 263 |
+
"eval_steps_per_second": 2.054,
|
| 264 |
+
"step": 4000
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.38,
|
| 268 |
+
"learning_rate": 3.419055649241147e-05,
|
| 269 |
+
"loss": 1.1193,
|
| 270 |
+
"step": 4250
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.38,
|
| 274 |
+
"eval_accuracy": 0.6332479180012812,
|
| 275 |
+
"eval_f1_score": 0.6195484695149066,
|
| 276 |
+
"eval_loss": 1.2123792171478271,
|
| 277 |
+
"eval_runtime": 79.3987,
|
| 278 |
+
"eval_samples_per_second": 196.603,
|
| 279 |
+
"eval_steps_per_second": 2.053,
|
| 280 |
+
"step": 4250
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.46,
|
| 284 |
+
"learning_rate": 3.313659359190557e-05,
|
| 285 |
+
"loss": 1.1093,
|
| 286 |
+
"step": 4500
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 1.46,
|
| 290 |
+
"eval_accuracy": 0.6402306213965406,
|
| 291 |
+
"eval_f1_score": 0.6254312551097587,
|
| 292 |
+
"eval_loss": 1.1897130012512207,
|
| 293 |
+
"eval_runtime": 79.5616,
|
| 294 |
+
"eval_samples_per_second": 196.2,
|
| 295 |
+
"eval_steps_per_second": 2.049,
|
| 296 |
+
"step": 4500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.54,
|
| 300 |
+
"learning_rate": 3.208263069139966e-05,
|
| 301 |
+
"loss": 1.0848,
|
| 302 |
+
"step": 4750
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.54,
|
| 306 |
+
"eval_accuracy": 0.6440102498398462,
|
| 307 |
+
"eval_f1_score": 0.6291617598687145,
|
| 308 |
+
"eval_loss": 1.178646445274353,
|
| 309 |
+
"eval_runtime": 79.2558,
|
| 310 |
+
"eval_samples_per_second": 196.957,
|
| 311 |
+
"eval_steps_per_second": 2.057,
|
| 312 |
+
"step": 4750
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.62,
|
| 316 |
+
"learning_rate": 3.1028667790893765e-05,
|
| 317 |
+
"loss": 1.1105,
|
| 318 |
+
"step": 5000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.62,
|
| 322 |
+
"eval_accuracy": 0.649583600256246,
|
| 323 |
+
"eval_f1_score": 0.6335161931693469,
|
| 324 |
+
"eval_loss": 1.1682264804840088,
|
| 325 |
+
"eval_runtime": 79.3313,
|
| 326 |
+
"eval_samples_per_second": 196.77,
|
| 327 |
+
"eval_steps_per_second": 2.055,
|
| 328 |
+
"step": 5000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 1.7,
|
| 332 |
+
"learning_rate": 2.997470489038786e-05,
|
| 333 |
+
"loss": 1.1014,
|
| 334 |
+
"step": 5250
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 1.7,
|
| 338 |
+
"eval_accuracy": 0.6516335682254965,
|
| 339 |
+
"eval_f1_score": 0.6377925627710681,
|
| 340 |
+
"eval_loss": 1.156283974647522,
|
| 341 |
+
"eval_runtime": 79.534,
|
| 342 |
+
"eval_samples_per_second": 196.268,
|
| 343 |
+
"eval_steps_per_second": 2.049,
|
| 344 |
+
"step": 5250
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.78,
|
| 348 |
+
"learning_rate": 2.8920741989881955e-05,
|
| 349 |
+
"loss": 1.0828,
|
| 350 |
+
"step": 5500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.78,
|
| 354 |
+
"eval_accuracy": 0.6459961563100577,
|
| 355 |
+
"eval_f1_score": 0.6313286940706383,
|
| 356 |
+
"eval_loss": 1.1570398807525635,
|
| 357 |
+
"eval_runtime": 79.5476,
|
| 358 |
+
"eval_samples_per_second": 196.235,
|
| 359 |
+
"eval_steps_per_second": 2.049,
|
| 360 |
+
"step": 5500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.86,
|
| 364 |
+
"learning_rate": 2.7866779089376055e-05,
|
| 365 |
+
"loss": 1.0803,
|
| 366 |
+
"step": 5750
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.86,
|
| 370 |
+
"eval_accuracy": 0.6530429212043561,
|
| 371 |
+
"eval_f1_score": 0.6404845558613215,
|
| 372 |
+
"eval_loss": 1.146980881690979,
|
| 373 |
+
"eval_runtime": 79.2229,
|
| 374 |
+
"eval_samples_per_second": 197.039,
|
| 375 |
+
"eval_steps_per_second": 2.057,
|
| 376 |
+
"step": 5750
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.94,
|
| 380 |
+
"learning_rate": 2.6812816188870155e-05,
|
| 381 |
+
"loss": 1.0799,
|
| 382 |
+
"step": 6000
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 1.94,
|
| 386 |
+
"eval_accuracy": 0.6516976297245356,
|
| 387 |
+
"eval_f1_score": 0.638943219006341,
|
| 388 |
+
"eval_loss": 1.1435405015945435,
|
| 389 |
+
"eval_runtime": 79.322,
|
| 390 |
+
"eval_samples_per_second": 196.793,
|
| 391 |
+
"eval_steps_per_second": 2.055,
|
| 392 |
+
"step": 6000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.02,
|
| 396 |
+
"learning_rate": 2.575885328836425e-05,
|
| 397 |
+
"loss": 1.0172,
|
| 398 |
+
"step": 6250
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.02,
|
| 402 |
+
"eval_accuracy": 0.647982062780269,
|
| 403 |
+
"eval_f1_score": 0.6342691459320453,
|
| 404 |
+
"eval_loss": 1.157272219657898,
|
| 405 |
+
"eval_runtime": 79.3295,
|
| 406 |
+
"eval_samples_per_second": 196.774,
|
| 407 |
+
"eval_steps_per_second": 2.055,
|
| 408 |
+
"step": 6250
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 2.1,
|
| 412 |
+
"learning_rate": 2.470489038785835e-05,
|
| 413 |
+
"loss": 0.8986,
|
| 414 |
+
"step": 6500
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 2.1,
|
| 418 |
+
"eval_accuracy": 0.6488789237668161,
|
| 419 |
+
"eval_f1_score": 0.6382130933006974,
|
| 420 |
+
"eval_loss": 1.1634544134140015,
|
| 421 |
+
"eval_runtime": 79.298,
|
| 422 |
+
"eval_samples_per_second": 196.852,
|
| 423 |
+
"eval_steps_per_second": 2.056,
|
| 424 |
+
"step": 6500
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 2.18,
|
| 428 |
+
"learning_rate": 2.3650927487352446e-05,
|
| 429 |
+
"loss": 0.8871,
|
| 430 |
+
"step": 6750
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 2.18,
|
| 434 |
+
"eval_accuracy": 0.6540679051889814,
|
| 435 |
+
"eval_f1_score": 0.6404856756767366,
|
| 436 |
+
"eval_loss": 1.1588941812515259,
|
| 437 |
+
"eval_runtime": 79.2578,
|
| 438 |
+
"eval_samples_per_second": 196.952,
|
| 439 |
+
"eval_steps_per_second": 2.057,
|
| 440 |
+
"step": 6750
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.27,
|
| 444 |
+
"learning_rate": 2.2596964586846546e-05,
|
| 445 |
+
"loss": 0.9066,
|
| 446 |
+
"step": 7000
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.27,
|
| 450 |
+
"eval_accuracy": 0.6515695067264574,
|
| 451 |
+
"eval_f1_score": 0.6402759354718203,
|
| 452 |
+
"eval_loss": 1.146422266960144,
|
| 453 |
+
"eval_runtime": 79.5354,
|
| 454 |
+
"eval_samples_per_second": 196.265,
|
| 455 |
+
"eval_steps_per_second": 2.049,
|
| 456 |
+
"step": 7000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 2.35,
|
| 460 |
+
"learning_rate": 2.1543001686340643e-05,
|
| 461 |
+
"loss": 0.897,
|
| 462 |
+
"step": 7250
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 2.35,
|
| 466 |
+
"eval_accuracy": 0.650352338244715,
|
| 467 |
+
"eval_f1_score": 0.6389635569093923,
|
| 468 |
+
"eval_loss": 1.1561025381088257,
|
| 469 |
+
"eval_runtime": 79.2894,
|
| 470 |
+
"eval_samples_per_second": 196.874,
|
| 471 |
+
"eval_steps_per_second": 2.056,
|
| 472 |
+
"step": 7250
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 2.43,
|
| 476 |
+
"learning_rate": 2.048903878583474e-05,
|
| 477 |
+
"loss": 0.8988,
|
| 478 |
+
"step": 7500
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 2.43,
|
| 482 |
+
"eval_accuracy": 0.6547725816784112,
|
| 483 |
+
"eval_f1_score": 0.6425914825215071,
|
| 484 |
+
"eval_loss": 1.1453956365585327,
|
| 485 |
+
"eval_runtime": 79.2343,
|
| 486 |
+
"eval_samples_per_second": 197.011,
|
| 487 |
+
"eval_steps_per_second": 2.057,
|
| 488 |
+
"step": 7500
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.51,
|
| 492 |
+
"learning_rate": 1.943507588532884e-05,
|
| 493 |
+
"loss": 0.9023,
|
| 494 |
+
"step": 7750
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.51,
|
| 498 |
+
"eval_accuracy": 0.6542600896860986,
|
| 499 |
+
"eval_f1_score": 0.6438243504000527,
|
| 500 |
+
"eval_loss": 1.1429415941238403,
|
| 501 |
+
"eval_runtime": 79.2597,
|
| 502 |
+
"eval_samples_per_second": 196.947,
|
| 503 |
+
"eval_steps_per_second": 2.057,
|
| 504 |
+
"step": 7750
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 2.59,
|
| 508 |
+
"learning_rate": 1.8381112984822936e-05,
|
| 509 |
+
"loss": 0.8791,
|
| 510 |
+
"step": 8000
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 2.59,
|
| 514 |
+
"eval_accuracy": 0.6572069186418962,
|
| 515 |
+
"eval_f1_score": 0.647070357907258,
|
| 516 |
+
"eval_loss": 1.1421723365783691,
|
| 517 |
+
"eval_runtime": 79.5074,
|
| 518 |
+
"eval_samples_per_second": 196.334,
|
| 519 |
+
"eval_steps_per_second": 2.05,
|
| 520 |
+
"step": 8000
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 2.67,
|
| 524 |
+
"learning_rate": 1.7327150084317033e-05,
|
| 525 |
+
"loss": 0.8971,
|
| 526 |
+
"step": 8250
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 2.67,
|
| 530 |
+
"eval_accuracy": 0.6581678411274824,
|
| 531 |
+
"eval_f1_score": 0.6474305978141667,
|
| 532 |
+
"eval_loss": 1.1367805004119873,
|
| 533 |
+
"eval_runtime": 79.1975,
|
| 534 |
+
"eval_samples_per_second": 197.102,
|
| 535 |
+
"eval_steps_per_second": 2.058,
|
| 536 |
+
"step": 8250
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.75,
|
| 540 |
+
"learning_rate": 1.627318718381113e-05,
|
| 541 |
+
"loss": 0.8885,
|
| 542 |
+
"step": 8500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.75,
|
| 546 |
+
"eval_accuracy": 0.6608584240871236,
|
| 547 |
+
"eval_f1_score": 0.6487602589453412,
|
| 548 |
+
"eval_loss": 1.122239351272583,
|
| 549 |
+
"eval_runtime": 79.2042,
|
| 550 |
+
"eval_samples_per_second": 197.085,
|
| 551 |
+
"eval_steps_per_second": 2.058,
|
| 552 |
+
"step": 8500
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 2.83,
|
| 556 |
+
"learning_rate": 1.5219224283305228e-05,
|
| 557 |
+
"loss": 0.8745,
|
| 558 |
+
"step": 8750
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 2.83,
|
| 562 |
+
"eval_accuracy": 0.6579756566303652,
|
| 563 |
+
"eval_f1_score": 0.6483741964856444,
|
| 564 |
+
"eval_loss": 1.1291719675064087,
|
| 565 |
+
"eval_runtime": 79.4248,
|
| 566 |
+
"eval_samples_per_second": 196.538,
|
| 567 |
+
"eval_steps_per_second": 2.052,
|
| 568 |
+
"step": 8750
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 2.91,
|
| 572 |
+
"learning_rate": 1.4165261382799327e-05,
|
| 573 |
+
"loss": 0.8837,
|
| 574 |
+
"step": 9000
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 2.91,
|
| 578 |
+
"eval_accuracy": 0.6574631646380525,
|
| 579 |
+
"eval_f1_score": 0.64725568770933,
|
| 580 |
+
"eval_loss": 1.1298834085464478,
|
| 581 |
+
"eval_runtime": 79.278,
|
| 582 |
+
"eval_samples_per_second": 196.902,
|
| 583 |
+
"eval_steps_per_second": 2.056,
|
| 584 |
+
"step": 9000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.99,
|
| 588 |
+
"learning_rate": 1.3111298482293424e-05,
|
| 589 |
+
"loss": 0.8604,
|
| 590 |
+
"step": 9250
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.99,
|
| 594 |
+
"eval_accuracy": 0.6556694426649584,
|
| 595 |
+
"eval_f1_score": 0.6459143709441728,
|
| 596 |
+
"eval_loss": 1.1267211437225342,
|
| 597 |
+
"eval_runtime": 79.3015,
|
| 598 |
+
"eval_samples_per_second": 196.844,
|
| 599 |
+
"eval_steps_per_second": 2.055,
|
| 600 |
+
"step": 9250
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 3.07,
|
| 604 |
+
"learning_rate": 1.205733558178752e-05,
|
| 605 |
+
"loss": 0.754,
|
| 606 |
+
"step": 9500
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 3.07,
|
| 610 |
+
"eval_accuracy": 0.657847533632287,
|
| 611 |
+
"eval_f1_score": 0.6475765832075672,
|
| 612 |
+
"eval_loss": 1.1378750801086426,
|
| 613 |
+
"eval_runtime": 79.3879,
|
| 614 |
+
"eval_samples_per_second": 196.63,
|
| 615 |
+
"eval_steps_per_second": 2.053,
|
| 616 |
+
"step": 9500
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 3.16,
|
| 620 |
+
"learning_rate": 1.1003372681281619e-05,
|
| 621 |
+
"loss": 0.7371,
|
| 622 |
+
"step": 9750
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 3.16,
|
| 626 |
+
"eval_accuracy": 0.658744394618834,
|
| 627 |
+
"eval_f1_score": 0.648976579094328,
|
| 628 |
+
"eval_loss": 1.1473287343978882,
|
| 629 |
+
"eval_runtime": 79.2492,
|
| 630 |
+
"eval_samples_per_second": 196.974,
|
| 631 |
+
"eval_steps_per_second": 2.057,
|
| 632 |
+
"step": 9750
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.24,
|
| 636 |
+
"learning_rate": 9.949409780775717e-06,
|
| 637 |
+
"loss": 0.7418,
|
| 638 |
+
"step": 10000
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.24,
|
| 642 |
+
"eval_accuracy": 0.6577194106342088,
|
| 643 |
+
"eval_f1_score": 0.6481259620512635,
|
| 644 |
+
"eval_loss": 1.1429522037506104,
|
| 645 |
+
"eval_runtime": 79.2653,
|
| 646 |
+
"eval_samples_per_second": 196.934,
|
| 647 |
+
"eval_steps_per_second": 2.056,
|
| 648 |
+
"step": 10000
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 3.32,
|
| 652 |
+
"learning_rate": 8.895446880269814e-06,
|
| 653 |
+
"loss": 0.7478,
|
| 654 |
+
"step": 10250
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 3.32,
|
| 658 |
+
"eval_accuracy": 0.6581678411274824,
|
| 659 |
+
"eval_f1_score": 0.6480298727343113,
|
| 660 |
+
"eval_loss": 1.147200107574463,
|
| 661 |
+
"eval_runtime": 79.3533,
|
| 662 |
+
"eval_samples_per_second": 196.715,
|
| 663 |
+
"eval_steps_per_second": 2.054,
|
| 664 |
+
"step": 10250
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 3.4,
|
| 668 |
+
"learning_rate": 7.841483979763913e-06,
|
| 669 |
+
"loss": 0.7214,
|
| 670 |
+
"step": 10500
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 3.4,
|
| 674 |
+
"eval_accuracy": 0.6593209481101858,
|
| 675 |
+
"eval_f1_score": 0.6487748216018238,
|
| 676 |
+
"eval_loss": 1.1483687162399292,
|
| 677 |
+
"eval_runtime": 79.2151,
|
| 678 |
+
"eval_samples_per_second": 197.058,
|
| 679 |
+
"eval_steps_per_second": 2.058,
|
| 680 |
+
"step": 10500
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 3.48,
|
| 684 |
+
"learning_rate": 6.787521079258011e-06,
|
| 685 |
+
"loss": 0.7378,
|
| 686 |
+
"step": 10750
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 3.48,
|
| 690 |
+
"eval_accuracy": 0.6613709160794362,
|
| 691 |
+
"eval_f1_score": 0.6515662944938931,
|
| 692 |
+
"eval_loss": 1.1435444355010986,
|
| 693 |
+
"eval_runtime": 79.1971,
|
| 694 |
+
"eval_samples_per_second": 197.103,
|
| 695 |
+
"eval_steps_per_second": 2.058,
|
| 696 |
+
"step": 10750
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 3.56,
|
| 700 |
+
"learning_rate": 5.733558178752108e-06,
|
| 701 |
+
"loss": 0.746,
|
| 702 |
+
"step": 11000
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 3.56,
|
| 706 |
+
"eval_accuracy": 0.6582959641255606,
|
| 707 |
+
"eval_f1_score": 0.649775469900882,
|
| 708 |
+
"eval_loss": 1.142891764640808,
|
| 709 |
+
"eval_runtime": 79.304,
|
| 710 |
+
"eval_samples_per_second": 196.838,
|
| 711 |
+
"eval_steps_per_second": 2.055,
|
| 712 |
+
"step": 11000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 3.64,
|
| 716 |
+
"learning_rate": 4.6795952782462055e-06,
|
| 717 |
+
"loss": 0.7355,
|
| 718 |
+
"step": 11250
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 3.64,
|
| 722 |
+
"eval_accuracy": 0.6614349775784754,
|
| 723 |
+
"eval_f1_score": 0.6523276061496733,
|
| 724 |
+
"eval_loss": 1.1420927047729492,
|
| 725 |
+
"eval_runtime": 79.4328,
|
| 726 |
+
"eval_samples_per_second": 196.518,
|
| 727 |
+
"eval_steps_per_second": 2.052,
|
| 728 |
+
"step": 11250
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 3.72,
|
| 732 |
+
"learning_rate": 3.625632377740304e-06,
|
| 733 |
+
"loss": 0.7207,
|
| 734 |
+
"step": 11500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 3.72,
|
| 738 |
+
"eval_accuracy": 0.6618193465727098,
|
| 739 |
+
"eval_f1_score": 0.6531447427234552,
|
| 740 |
+
"eval_loss": 1.1392197608947754,
|
| 741 |
+
"eval_runtime": 79.3578,
|
| 742 |
+
"eval_samples_per_second": 196.704,
|
| 743 |
+
"eval_steps_per_second": 2.054,
|
| 744 |
+
"step": 11500
|
| 745 |
+
}
|
| 746 |
+
],
|
| 747 |
+
"max_steps": 12360,
|
| 748 |
+
"num_train_epochs": 4,
|
| 749 |
+
"total_flos": 1.4666146788722688e+17,
|
| 750 |
+
"trial_name": null,
|
| 751 |
+
"trial_params": null
|
| 752 |
+
}
|