Commit
•
1e34e40
1
Parent(s):
8619de8
Update trainer_state.json
Browse files- trainer_state.json +976 -0
trainer_state.json
CHANGED
|
@@ -0,0 +1,976 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 24.43494196701283,
|
| 5 |
+
"global_step": 80000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.15,
|
| 12 |
+
"learning_rate": 1.993891264508247e-05,
|
| 13 |
+
"loss": 0.4281,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.31,
|
| 18 |
+
"learning_rate": 1.987782529016494e-05,
|
| 19 |
+
"loss": 0.3603,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.46,
|
| 24 |
+
"learning_rate": 1.9816737935247404e-05,
|
| 25 |
+
"loss": 0.335,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.61,
|
| 30 |
+
"learning_rate": 1.9755650580329873e-05,
|
| 31 |
+
"loss": 0.324,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.76,
|
| 36 |
+
"learning_rate": 1.9694563225412342e-05,
|
| 37 |
+
"loss": 0.3191,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.92,
|
| 42 |
+
"learning_rate": 1.963347587049481e-05,
|
| 43 |
+
"loss": 0.3027,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.07,
|
| 48 |
+
"learning_rate": 1.9572388515577276e-05,
|
| 49 |
+
"loss": 0.2698,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.22,
|
| 54 |
+
"learning_rate": 1.9511301160659744e-05,
|
| 55 |
+
"loss": 0.2289,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.37,
|
| 60 |
+
"learning_rate": 1.9450213805742213e-05,
|
| 61 |
+
"loss": 0.2299,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.53,
|
| 66 |
+
"learning_rate": 1.938912645082468e-05,
|
| 67 |
+
"loss": 0.2369,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.68,
|
| 72 |
+
"learning_rate": 1.932803909590715e-05,
|
| 73 |
+
"loss": 0.241,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 1.83,
|
| 78 |
+
"learning_rate": 1.9266951740989616e-05,
|
| 79 |
+
"loss": 0.2375,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.99,
|
| 84 |
+
"learning_rate": 1.9205864386072085e-05,
|
| 85 |
+
"loss": 0.2334,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 2.14,
|
| 90 |
+
"learning_rate": 1.9144777031154553e-05,
|
| 91 |
+
"loss": 0.1654,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.29,
|
| 96 |
+
"learning_rate": 1.9083689676237022e-05,
|
| 97 |
+
"loss": 0.1569,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 2.44,
|
| 102 |
+
"learning_rate": 1.9022602321319487e-05,
|
| 103 |
+
"loss": 0.1689,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 2.6,
|
| 108 |
+
"learning_rate": 1.8961514966401956e-05,
|
| 109 |
+
"loss": 0.1641,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 2.75,
|
| 114 |
+
"learning_rate": 1.8900427611484425e-05,
|
| 115 |
+
"loss": 0.1654,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 2.9,
|
| 120 |
+
"learning_rate": 1.883934025656689e-05,
|
| 121 |
+
"loss": 0.1684,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 3.05,
|
| 126 |
+
"learning_rate": 1.8778252901649362e-05,
|
| 127 |
+
"loss": 0.1484,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 3.21,
|
| 132 |
+
"learning_rate": 1.8717165546731827e-05,
|
| 133 |
+
"loss": 0.1118,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 3.36,
|
| 138 |
+
"learning_rate": 1.8656078191814296e-05,
|
| 139 |
+
"loss": 0.1181,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 3.51,
|
| 144 |
+
"learning_rate": 1.8594990836896765e-05,
|
| 145 |
+
"loss": 0.118,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 3.67,
|
| 150 |
+
"learning_rate": 1.853390348197923e-05,
|
| 151 |
+
"loss": 0.1204,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 3.82,
|
| 156 |
+
"learning_rate": 1.84728161270617e-05,
|
| 157 |
+
"loss": 0.1255,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 3.97,
|
| 162 |
+
"learning_rate": 1.8411728772144168e-05,
|
| 163 |
+
"loss": 0.1284,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 4.12,
|
| 168 |
+
"learning_rate": 1.8350641417226636e-05,
|
| 169 |
+
"loss": 0.0925,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 4.28,
|
| 174 |
+
"learning_rate": 1.82895540623091e-05,
|
| 175 |
+
"loss": 0.0896,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 4.43,
|
| 180 |
+
"learning_rate": 1.8228466707391574e-05,
|
| 181 |
+
"loss": 0.0897,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 4.58,
|
| 186 |
+
"learning_rate": 1.816737935247404e-05,
|
| 187 |
+
"loss": 0.0969,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 4.73,
|
| 192 |
+
"learning_rate": 1.8106291997556508e-05,
|
| 193 |
+
"loss": 0.0971,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 4.89,
|
| 198 |
+
"learning_rate": 1.8045204642638976e-05,
|
| 199 |
+
"loss": 0.0963,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 5.04,
|
| 204 |
+
"learning_rate": 1.7984117287721442e-05,
|
| 205 |
+
"loss": 0.0906,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 5.19,
|
| 210 |
+
"learning_rate": 1.792302993280391e-05,
|
| 211 |
+
"loss": 0.0717,
|
| 212 |
+
"step": 17000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 5.35,
|
| 216 |
+
"learning_rate": 1.786194257788638e-05,
|
| 217 |
+
"loss": 0.0754,
|
| 218 |
+
"step": 17500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 5.5,
|
| 222 |
+
"learning_rate": 1.7800855222968848e-05,
|
| 223 |
+
"loss": 0.0709,
|
| 224 |
+
"step": 18000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 5.65,
|
| 228 |
+
"learning_rate": 1.7739767868051313e-05,
|
| 229 |
+
"loss": 0.0814,
|
| 230 |
+
"step": 18500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 5.8,
|
| 234 |
+
"learning_rate": 1.7678680513133785e-05,
|
| 235 |
+
"loss": 0.0797,
|
| 236 |
+
"step": 19000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 5.96,
|
| 240 |
+
"learning_rate": 1.761759315821625e-05,
|
| 241 |
+
"loss": 0.0914,
|
| 242 |
+
"step": 19500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 6.11,
|
| 246 |
+
"learning_rate": 1.755650580329872e-05,
|
| 247 |
+
"loss": 0.0648,
|
| 248 |
+
"step": 20000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 6.26,
|
| 252 |
+
"learning_rate": 1.7495418448381188e-05,
|
| 253 |
+
"loss": 0.0627,
|
| 254 |
+
"step": 20500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 6.41,
|
| 258 |
+
"learning_rate": 1.7434331093463653e-05,
|
| 259 |
+
"loss": 0.061,
|
| 260 |
+
"step": 21000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 6.57,
|
| 264 |
+
"learning_rate": 1.7373243738546122e-05,
|
| 265 |
+
"loss": 0.0678,
|
| 266 |
+
"step": 21500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 6.72,
|
| 270 |
+
"learning_rate": 1.731215638362859e-05,
|
| 271 |
+
"loss": 0.0656,
|
| 272 |
+
"step": 22000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 6.87,
|
| 276 |
+
"learning_rate": 1.725106902871106e-05,
|
| 277 |
+
"loss": 0.0702,
|
| 278 |
+
"step": 22500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 7.03,
|
| 282 |
+
"learning_rate": 1.7189981673793525e-05,
|
| 283 |
+
"loss": 0.0716,
|
| 284 |
+
"step": 23000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 7.18,
|
| 288 |
+
"learning_rate": 1.7128894318875993e-05,
|
| 289 |
+
"loss": 0.0553,
|
| 290 |
+
"step": 23500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 7.33,
|
| 294 |
+
"learning_rate": 1.7067806963958462e-05,
|
| 295 |
+
"loss": 0.0546,
|
| 296 |
+
"step": 24000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 7.48,
|
| 300 |
+
"learning_rate": 1.700671960904093e-05,
|
| 301 |
+
"loss": 0.0575,
|
| 302 |
+
"step": 24500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 7.64,
|
| 306 |
+
"learning_rate": 1.69456322541234e-05,
|
| 307 |
+
"loss": 0.0558,
|
| 308 |
+
"step": 25000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 7.79,
|
| 312 |
+
"learning_rate": 1.6884544899205865e-05,
|
| 313 |
+
"loss": 0.0628,
|
| 314 |
+
"step": 25500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 7.94,
|
| 318 |
+
"learning_rate": 1.6823457544288334e-05,
|
| 319 |
+
"loss": 0.06,
|
| 320 |
+
"step": 26000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 8.09,
|
| 324 |
+
"learning_rate": 1.6762370189370802e-05,
|
| 325 |
+
"loss": 0.0501,
|
| 326 |
+
"step": 26500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 8.25,
|
| 330 |
+
"learning_rate": 1.670128283445327e-05,
|
| 331 |
+
"loss": 0.0465,
|
| 332 |
+
"step": 27000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 8.4,
|
| 336 |
+
"learning_rate": 1.6640195479535736e-05,
|
| 337 |
+
"loss": 0.0497,
|
| 338 |
+
"step": 27500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 8.55,
|
| 342 |
+
"learning_rate": 1.6579108124618205e-05,
|
| 343 |
+
"loss": 0.0513,
|
| 344 |
+
"step": 28000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 8.7,
|
| 348 |
+
"learning_rate": 1.6518020769700674e-05,
|
| 349 |
+
"loss": 0.0526,
|
| 350 |
+
"step": 28500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 8.86,
|
| 354 |
+
"learning_rate": 1.645693341478314e-05,
|
| 355 |
+
"loss": 0.0503,
|
| 356 |
+
"step": 29000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 9.01,
|
| 360 |
+
"learning_rate": 1.639584605986561e-05,
|
| 361 |
+
"loss": 0.05,
|
| 362 |
+
"step": 29500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 9.16,
|
| 366 |
+
"learning_rate": 1.6334758704948076e-05,
|
| 367 |
+
"loss": 0.0371,
|
| 368 |
+
"step": 30000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 9.32,
|
| 372 |
+
"learning_rate": 1.6273671350030545e-05,
|
| 373 |
+
"loss": 0.0396,
|
| 374 |
+
"step": 30500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 9.47,
|
| 378 |
+
"learning_rate": 1.6212583995113014e-05,
|
| 379 |
+
"loss": 0.0417,
|
| 380 |
+
"step": 31000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 9.62,
|
| 384 |
+
"learning_rate": 1.6151496640195482e-05,
|
| 385 |
+
"loss": 0.0453,
|
| 386 |
+
"step": 31500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 9.77,
|
| 390 |
+
"learning_rate": 1.6090409285277948e-05,
|
| 391 |
+
"loss": 0.0402,
|
| 392 |
+
"step": 32000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 9.93,
|
| 396 |
+
"learning_rate": 1.6029321930360416e-05,
|
| 397 |
+
"loss": 0.0421,
|
| 398 |
+
"step": 32500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 10.08,
|
| 402 |
+
"learning_rate": 1.5968234575442885e-05,
|
| 403 |
+
"loss": 0.0389,
|
| 404 |
+
"step": 33000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 10.23,
|
| 408 |
+
"learning_rate": 1.590714722052535e-05,
|
| 409 |
+
"loss": 0.0372,
|
| 410 |
+
"step": 33500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 10.38,
|
| 414 |
+
"learning_rate": 1.5846059865607823e-05,
|
| 415 |
+
"loss": 0.0346,
|
| 416 |
+
"step": 34000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 10.54,
|
| 420 |
+
"learning_rate": 1.5784972510690288e-05,
|
| 421 |
+
"loss": 0.0397,
|
| 422 |
+
"step": 34500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 10.69,
|
| 426 |
+
"learning_rate": 1.5723885155772757e-05,
|
| 427 |
+
"loss": 0.0357,
|
| 428 |
+
"step": 35000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 10.84,
|
| 432 |
+
"learning_rate": 1.5662797800855225e-05,
|
| 433 |
+
"loss": 0.032,
|
| 434 |
+
"step": 35500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 11.0,
|
| 438 |
+
"learning_rate": 1.560171044593769e-05,
|
| 439 |
+
"loss": 0.0474,
|
| 440 |
+
"step": 36000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 11.15,
|
| 444 |
+
"learning_rate": 1.554062309102016e-05,
|
| 445 |
+
"loss": 0.0298,
|
| 446 |
+
"step": 36500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 11.3,
|
| 450 |
+
"learning_rate": 1.5479535736102628e-05,
|
| 451 |
+
"loss": 0.0355,
|
| 452 |
+
"step": 37000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 11.45,
|
| 456 |
+
"learning_rate": 1.5418448381185097e-05,
|
| 457 |
+
"loss": 0.0368,
|
| 458 |
+
"step": 37500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 11.61,
|
| 462 |
+
"learning_rate": 1.5357361026267562e-05,
|
| 463 |
+
"loss": 0.0335,
|
| 464 |
+
"step": 38000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 11.76,
|
| 468 |
+
"learning_rate": 1.5296273671350034e-05,
|
| 469 |
+
"loss": 0.0328,
|
| 470 |
+
"step": 38500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 11.91,
|
| 474 |
+
"learning_rate": 1.52351863164325e-05,
|
| 475 |
+
"loss": 0.0317,
|
| 476 |
+
"step": 39000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 12.06,
|
| 480 |
+
"learning_rate": 1.5174098961514966e-05,
|
| 481 |
+
"loss": 0.0346,
|
| 482 |
+
"step": 39500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 12.22,
|
| 486 |
+
"learning_rate": 1.5113011606597437e-05,
|
| 487 |
+
"loss": 0.0279,
|
| 488 |
+
"step": 40000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 12.37,
|
| 492 |
+
"learning_rate": 1.5051924251679904e-05,
|
| 493 |
+
"loss": 0.0268,
|
| 494 |
+
"step": 40500
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 12.52,
|
| 498 |
+
"learning_rate": 1.4990836896762371e-05,
|
| 499 |
+
"loss": 0.0324,
|
| 500 |
+
"step": 41000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 12.68,
|
| 504 |
+
"learning_rate": 1.492974954184484e-05,
|
| 505 |
+
"loss": 0.0309,
|
| 506 |
+
"step": 41500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 12.83,
|
| 510 |
+
"learning_rate": 1.4868662186927307e-05,
|
| 511 |
+
"loss": 0.0316,
|
| 512 |
+
"step": 42000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 12.98,
|
| 516 |
+
"learning_rate": 1.4807574832009775e-05,
|
| 517 |
+
"loss": 0.0365,
|
| 518 |
+
"step": 42500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 13.13,
|
| 522 |
+
"learning_rate": 1.4746487477092244e-05,
|
| 523 |
+
"loss": 0.0241,
|
| 524 |
+
"step": 43000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 13.29,
|
| 528 |
+
"learning_rate": 1.4685400122174711e-05,
|
| 529 |
+
"loss": 0.0276,
|
| 530 |
+
"step": 43500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 13.44,
|
| 534 |
+
"learning_rate": 1.4624312767257178e-05,
|
| 535 |
+
"loss": 0.0249,
|
| 536 |
+
"step": 44000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 13.59,
|
| 540 |
+
"learning_rate": 1.4563225412339648e-05,
|
| 541 |
+
"loss": 0.0284,
|
| 542 |
+
"step": 44500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 13.74,
|
| 546 |
+
"learning_rate": 1.4502138057422115e-05,
|
| 547 |
+
"loss": 0.0277,
|
| 548 |
+
"step": 45000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 13.9,
|
| 552 |
+
"learning_rate": 1.4441050702504582e-05,
|
| 553 |
+
"loss": 0.0296,
|
| 554 |
+
"step": 45500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 14.05,
|
| 558 |
+
"learning_rate": 1.4379963347587051e-05,
|
| 559 |
+
"loss": 0.0245,
|
| 560 |
+
"step": 46000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 14.2,
|
| 564 |
+
"learning_rate": 1.4318875992669518e-05,
|
| 565 |
+
"loss": 0.0227,
|
| 566 |
+
"step": 46500
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 14.36,
|
| 570 |
+
"learning_rate": 1.4257788637751985e-05,
|
| 571 |
+
"loss": 0.0273,
|
| 572 |
+
"step": 47000
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 14.51,
|
| 576 |
+
"learning_rate": 1.4196701282834456e-05,
|
| 577 |
+
"loss": 0.0263,
|
| 578 |
+
"step": 47500
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 14.66,
|
| 582 |
+
"learning_rate": 1.4135613927916923e-05,
|
| 583 |
+
"loss": 0.0243,
|
| 584 |
+
"step": 48000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 14.81,
|
| 588 |
+
"learning_rate": 1.407452657299939e-05,
|
| 589 |
+
"loss": 0.0256,
|
| 590 |
+
"step": 48500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 14.97,
|
| 594 |
+
"learning_rate": 1.4013439218081858e-05,
|
| 595 |
+
"loss": 0.0285,
|
| 596 |
+
"step": 49000
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 15.12,
|
| 600 |
+
"learning_rate": 1.3952351863164327e-05,
|
| 601 |
+
"loss": 0.0189,
|
| 602 |
+
"step": 49500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 15.27,
|
| 606 |
+
"learning_rate": 1.3891264508246794e-05,
|
| 607 |
+
"loss": 0.0186,
|
| 608 |
+
"step": 50000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 15.42,
|
| 612 |
+
"learning_rate": 1.3830177153329263e-05,
|
| 613 |
+
"loss": 0.0272,
|
| 614 |
+
"step": 50500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 15.58,
|
| 618 |
+
"learning_rate": 1.376908979841173e-05,
|
| 619 |
+
"loss": 0.0264,
|
| 620 |
+
"step": 51000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 15.73,
|
| 624 |
+
"learning_rate": 1.3708002443494197e-05,
|
| 625 |
+
"loss": 0.0239,
|
| 626 |
+
"step": 51500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 15.88,
|
| 630 |
+
"learning_rate": 1.3646915088576667e-05,
|
| 631 |
+
"loss": 0.0255,
|
| 632 |
+
"step": 52000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 16.04,
|
| 636 |
+
"learning_rate": 1.3585827733659134e-05,
|
| 637 |
+
"loss": 0.0261,
|
| 638 |
+
"step": 52500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 16.19,
|
| 642 |
+
"learning_rate": 1.3524740378741601e-05,
|
| 643 |
+
"loss": 0.0239,
|
| 644 |
+
"step": 53000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 16.34,
|
| 648 |
+
"learning_rate": 1.346365302382407e-05,
|
| 649 |
+
"loss": 0.0187,
|
| 650 |
+
"step": 53500
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 16.49,
|
| 654 |
+
"learning_rate": 1.3402565668906537e-05,
|
| 655 |
+
"loss": 0.023,
|
| 656 |
+
"step": 54000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 16.65,
|
| 660 |
+
"learning_rate": 1.3341478313989005e-05,
|
| 661 |
+
"loss": 0.0215,
|
| 662 |
+
"step": 54500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 16.8,
|
| 666 |
+
"learning_rate": 1.3280390959071474e-05,
|
| 667 |
+
"loss": 0.0264,
|
| 668 |
+
"step": 55000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 16.95,
|
| 672 |
+
"learning_rate": 1.3219303604153941e-05,
|
| 673 |
+
"loss": 0.0198,
|
| 674 |
+
"step": 55500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 17.1,
|
| 678 |
+
"learning_rate": 1.3158216249236408e-05,
|
| 679 |
+
"loss": 0.0191,
|
| 680 |
+
"step": 56000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 17.26,
|
| 684 |
+
"learning_rate": 1.3097128894318879e-05,
|
| 685 |
+
"loss": 0.0174,
|
| 686 |
+
"step": 56500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 17.41,
|
| 690 |
+
"learning_rate": 1.3036041539401346e-05,
|
| 691 |
+
"loss": 0.0149,
|
| 692 |
+
"step": 57000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 17.56,
|
| 696 |
+
"learning_rate": 1.2974954184483813e-05,
|
| 697 |
+
"loss": 0.019,
|
| 698 |
+
"step": 57500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 17.72,
|
| 702 |
+
"learning_rate": 1.2913866829566281e-05,
|
| 703 |
+
"loss": 0.0188,
|
| 704 |
+
"step": 58000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 17.87,
|
| 708 |
+
"learning_rate": 1.2852779474648748e-05,
|
| 709 |
+
"loss": 0.022,
|
| 710 |
+
"step": 58500
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 18.02,
|
| 714 |
+
"learning_rate": 1.2791692119731215e-05,
|
| 715 |
+
"loss": 0.0248,
|
| 716 |
+
"step": 59000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 18.17,
|
| 720 |
+
"learning_rate": 1.2730604764813686e-05,
|
| 721 |
+
"loss": 0.0162,
|
| 722 |
+
"step": 59500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 18.33,
|
| 726 |
+
"learning_rate": 1.2669517409896153e-05,
|
| 727 |
+
"loss": 0.017,
|
| 728 |
+
"step": 60000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 18.48,
|
| 732 |
+
"learning_rate": 1.260843005497862e-05,
|
| 733 |
+
"loss": 0.0149,
|
| 734 |
+
"step": 60500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 18.63,
|
| 738 |
+
"learning_rate": 1.2547342700061088e-05,
|
| 739 |
+
"loss": 0.0194,
|
| 740 |
+
"step": 61000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 18.78,
|
| 744 |
+
"learning_rate": 1.2486255345143557e-05,
|
| 745 |
+
"loss": 0.0185,
|
| 746 |
+
"step": 61500
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 18.94,
|
| 750 |
+
"learning_rate": 1.2425167990226024e-05,
|
| 751 |
+
"loss": 0.0161,
|
| 752 |
+
"step": 62000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 19.09,
|
| 756 |
+
"learning_rate": 1.2364080635308493e-05,
|
| 757 |
+
"loss": 0.0144,
|
| 758 |
+
"step": 62500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 19.24,
|
| 762 |
+
"learning_rate": 1.230299328039096e-05,
|
| 763 |
+
"loss": 0.0141,
|
| 764 |
+
"step": 63000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 19.4,
|
| 768 |
+
"learning_rate": 1.2241905925473427e-05,
|
| 769 |
+
"loss": 0.0149,
|
| 770 |
+
"step": 63500
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 19.55,
|
| 774 |
+
"learning_rate": 1.2180818570555897e-05,
|
| 775 |
+
"loss": 0.0178,
|
| 776 |
+
"step": 64000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 19.7,
|
| 780 |
+
"learning_rate": 1.2119731215638364e-05,
|
| 781 |
+
"loss": 0.0159,
|
| 782 |
+
"step": 64500
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 19.85,
|
| 786 |
+
"learning_rate": 1.2058643860720831e-05,
|
| 787 |
+
"loss": 0.0194,
|
| 788 |
+
"step": 65000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 20.01,
|
| 792 |
+
"learning_rate": 1.19975565058033e-05,
|
| 793 |
+
"loss": 0.0196,
|
| 794 |
+
"step": 65500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 20.16,
|
| 798 |
+
"learning_rate": 1.1936469150885767e-05,
|
| 799 |
+
"loss": 0.0123,
|
| 800 |
+
"step": 66000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 20.31,
|
| 804 |
+
"learning_rate": 1.1875381795968236e-05,
|
| 805 |
+
"loss": 0.0133,
|
| 806 |
+
"step": 66500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 20.46,
|
| 810 |
+
"learning_rate": 1.1814294441050704e-05,
|
| 811 |
+
"loss": 0.0188,
|
| 812 |
+
"step": 67000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 20.62,
|
| 816 |
+
"learning_rate": 1.1753207086133171e-05,
|
| 817 |
+
"loss": 0.0141,
|
| 818 |
+
"step": 67500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 20.77,
|
| 822 |
+
"learning_rate": 1.1692119731215638e-05,
|
| 823 |
+
"loss": 0.0176,
|
| 824 |
+
"step": 68000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 20.92,
|
| 828 |
+
"learning_rate": 1.1631032376298109e-05,
|
| 829 |
+
"loss": 0.0151,
|
| 830 |
+
"step": 68500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 21.08,
|
| 834 |
+
"learning_rate": 1.1569945021380576e-05,
|
| 835 |
+
"loss": 0.0151,
|
| 836 |
+
"step": 69000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 21.23,
|
| 840 |
+
"learning_rate": 1.1508857666463043e-05,
|
| 841 |
+
"loss": 0.0085,
|
| 842 |
+
"step": 69500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 21.38,
|
| 846 |
+
"learning_rate": 1.1447770311545512e-05,
|
| 847 |
+
"loss": 0.0165,
|
| 848 |
+
"step": 70000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 21.53,
|
| 852 |
+
"learning_rate": 1.1386682956627979e-05,
|
| 853 |
+
"loss": 0.015,
|
| 854 |
+
"step": 70500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 21.69,
|
| 858 |
+
"learning_rate": 1.1325595601710446e-05,
|
| 859 |
+
"loss": 0.0165,
|
| 860 |
+
"step": 71000
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 21.84,
|
| 864 |
+
"learning_rate": 1.1264508246792916e-05,
|
| 865 |
+
"loss": 0.0139,
|
| 866 |
+
"step": 71500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 21.99,
|
| 870 |
+
"learning_rate": 1.1203420891875383e-05,
|
| 871 |
+
"loss": 0.0152,
|
| 872 |
+
"step": 72000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 22.14,
|
| 876 |
+
"learning_rate": 1.114233353695785e-05,
|
| 877 |
+
"loss": 0.0116,
|
| 878 |
+
"step": 72500
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 22.3,
|
| 882 |
+
"learning_rate": 1.1081246182040319e-05,
|
| 883 |
+
"loss": 0.0134,
|
| 884 |
+
"step": 73000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 22.45,
|
| 888 |
+
"learning_rate": 1.1020158827122787e-05,
|
| 889 |
+
"loss": 0.0146,
|
| 890 |
+
"step": 73500
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 22.6,
|
| 894 |
+
"learning_rate": 1.0959071472205254e-05,
|
| 895 |
+
"loss": 0.0148,
|
| 896 |
+
"step": 74000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 22.76,
|
| 900 |
+
"learning_rate": 1.0897984117287723e-05,
|
| 901 |
+
"loss": 0.0124,
|
| 902 |
+
"step": 74500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 22.91,
|
| 906 |
+
"learning_rate": 1.083689676237019e-05,
|
| 907 |
+
"loss": 0.013,
|
| 908 |
+
"step": 75000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 23.06,
|
| 912 |
+
"learning_rate": 1.0775809407452657e-05,
|
| 913 |
+
"loss": 0.0124,
|
| 914 |
+
"step": 75500
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 23.21,
|
| 918 |
+
"learning_rate": 1.0714722052535128e-05,
|
| 919 |
+
"loss": 0.0092,
|
| 920 |
+
"step": 76000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 23.37,
|
| 924 |
+
"learning_rate": 1.0653634697617595e-05,
|
| 925 |
+
"loss": 0.014,
|
| 926 |
+
"step": 76500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 23.52,
|
| 930 |
+
"learning_rate": 1.0592547342700062e-05,
|
| 931 |
+
"loss": 0.0124,
|
| 932 |
+
"step": 77000
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 23.67,
|
| 936 |
+
"learning_rate": 1.053145998778253e-05,
|
| 937 |
+
"loss": 0.0116,
|
| 938 |
+
"step": 77500
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 23.82,
|
| 942 |
+
"learning_rate": 1.0470372632864997e-05,
|
| 943 |
+
"loss": 0.01,
|
| 944 |
+
"step": 78000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 23.98,
|
| 948 |
+
"learning_rate": 1.0409285277947466e-05,
|
| 949 |
+
"loss": 0.0146,
|
| 950 |
+
"step": 78500
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 24.13,
|
| 954 |
+
"learning_rate": 1.0348197923029935e-05,
|
| 955 |
+
"loss": 0.011,
|
| 956 |
+
"step": 79000
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 24.28,
|
| 960 |
+
"learning_rate": 1.0287110568112402e-05,
|
| 961 |
+
"loss": 0.0101,
|
| 962 |
+
"step": 79500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 24.43,
|
| 966 |
+
"learning_rate": 1.0226023213194869e-05,
|
| 967 |
+
"loss": 0.009,
|
| 968 |
+
"step": 80000
|
| 969 |
+
}
|
| 970 |
+
],
|
| 971 |
+
"max_steps": 163700,
|
| 972 |
+
"num_train_epochs": 50,
|
| 973 |
+
"total_flos": 2.152034161253376e+17,
|
| 974 |
+
"trial_name": null,
|
| 975 |
+
"trial_params": null
|
| 976 |
+
}
|