Commit
•
f7f6f79
1
Parent(s):
4eb4026
End of training
Browse files- all_results.json +9 -9
- eval_results.json +5 -5
- runs/Aug31_13-48-52_psxiy4p4s/events.out.tfevents.1661957726.psxiy4p4s.1875.2 +3 -0
- train_results.json +4 -4
- trainer_state.json +1919 -95
all_results.json
CHANGED
|
@@ -1,12 +1,12 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 5.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
-
"train_loss":
|
| 9 |
-
"train_runtime":
|
| 10 |
-
"train_samples_per_second":
|
| 11 |
-
"train_steps_per_second":
|
| 12 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 0.9948888888888889,
|
| 4 |
+
"eval_loss": 0.02364550158381462,
|
| 5 |
+
"eval_runtime": 47.5029,
|
| 6 |
+
"eval_samples_per_second": 189.462,
|
| 7 |
+
"eval_steps_per_second": 23.683,
|
| 8 |
+
"train_loss": 0.3736599187514361,
|
| 9 |
+
"train_runtime": 3816.438,
|
| 10 |
+
"train_samples_per_second": 66.816,
|
| 11 |
+
"train_steps_per_second": 8.352
|
| 12 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 5.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 0.9948888888888889,
|
| 4 |
+
"eval_loss": 0.02364550158381462,
|
| 5 |
+
"eval_runtime": 47.5029,
|
| 6 |
+
"eval_samples_per_second": 189.462,
|
| 7 |
+
"eval_steps_per_second": 23.683
|
| 8 |
}
|
runs/Aug31_13-48-52_psxiy4p4s/events.out.tfevents.1661957726.psxiy4p4s.1875.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:866b399d39674566cfce3346fb0c5824f04bcbb039eb3b4f54490326dd9650ee
|
| 3 |
+
size 369
|
train_results.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 5.0,
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples_per_second":
|
| 6 |
-
"train_steps_per_second":
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 5.0,
|
| 3 |
+
"train_loss": 0.3736599187514361,
|
| 4 |
+
"train_runtime": 3816.438,
|
| 5 |
+
"train_samples_per_second": 66.816,
|
| 6 |
+
"train_steps_per_second": 8.352
|
| 7 |
}
|
trainer_state.json
CHANGED
|
@@ -1,154 +1,1978 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "/storage/img-cls-data/
|
| 4 |
"epoch": 5.0,
|
| 5 |
-
"global_step":
|
| 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.
|
| 12 |
-
"learning_rate": 1.
|
| 13 |
-
"loss": 2.
|
| 14 |
"step": 100
|
| 15 |
},
|
| 16 |
{
|
| 17 |
-
"epoch": 0.
|
| 18 |
-
"learning_rate": 1.
|
| 19 |
-
"loss": 1.
|
| 20 |
"step": 200
|
| 21 |
},
|
| 22 |
{
|
| 23 |
-
"epoch":
|
| 24 |
-
"
|
| 25 |
-
"
|
| 26 |
-
"eval_runtime": 2.1357,
|
| 27 |
-
"eval_samples_per_second": 191.039,
|
| 28 |
-
"eval_steps_per_second": 23.88,
|
| 29 |
-
"step": 289
|
| 30 |
-
},
|
| 31 |
-
{
|
| 32 |
-
"epoch": 1.04,
|
| 33 |
-
"learning_rate": 1.5847750865051904e-05,
|
| 34 |
-
"loss": 1.6347,
|
| 35 |
"step": 300
|
| 36 |
},
|
| 37 |
{
|
| 38 |
-
"epoch":
|
| 39 |
-
"learning_rate": 1.
|
| 40 |
-
"loss": 1.
|
| 41 |
"step": 400
|
| 42 |
},
|
| 43 |
{
|
| 44 |
-
"epoch":
|
| 45 |
-
"learning_rate": 1.
|
| 46 |
-
"loss":
|
| 47 |
"step": 500
|
| 48 |
},
|
| 49 |
{
|
| 50 |
-
"epoch":
|
| 51 |
-
"
|
| 52 |
-
"
|
| 53 |
-
"eval_runtime": 1.9774,
|
| 54 |
-
"eval_samples_per_second": 206.327,
|
| 55 |
-
"eval_steps_per_second": 25.791,
|
| 56 |
-
"step": 578
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 2.08,
|
| 60 |
-
"learning_rate": 1.1695501730103806e-05,
|
| 61 |
-
"loss": 1.2134,
|
| 62 |
"step": 600
|
| 63 |
},
|
| 64 |
{
|
| 65 |
-
"epoch":
|
| 66 |
-
"learning_rate": 1.
|
| 67 |
-
"loss":
|
| 68 |
"step": 700
|
| 69 |
},
|
| 70 |
{
|
| 71 |
-
"epoch":
|
| 72 |
-
"learning_rate":
|
| 73 |
-
"loss":
|
| 74 |
"step": 800
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch":
|
| 78 |
-
"
|
| 79 |
-
"
|
| 80 |
-
"eval_runtime": 2.1384,
|
| 81 |
-
"eval_samples_per_second": 190.795,
|
| 82 |
-
"eval_steps_per_second": 23.849,
|
| 83 |
-
"step": 867
|
| 84 |
-
},
|
| 85 |
-
{
|
| 86 |
-
"epoch": 3.11,
|
| 87 |
-
"learning_rate": 7.5432525951557104e-06,
|
| 88 |
-
"loss": 1.0903,
|
| 89 |
"step": 900
|
| 90 |
},
|
| 91 |
{
|
| 92 |
-
"epoch":
|
| 93 |
-
"learning_rate":
|
| 94 |
-
"loss":
|
| 95 |
"step": 1000
|
| 96 |
},
|
| 97 |
{
|
| 98 |
-
"epoch":
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 1100
|
| 102 |
},
|
| 103 |
{
|
| 104 |
-
"epoch":
|
| 105 |
-
"
|
| 106 |
-
"
|
| 107 |
-
"eval_runtime": 2.0398,
|
| 108 |
-
"eval_samples_per_second": 200.024,
|
| 109 |
-
"eval_steps_per_second": 25.003,
|
| 110 |
-
"step": 1156
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 4.15,
|
| 114 |
-
"learning_rate": 3.3910034602076125e-06,
|
| 115 |
-
"loss": 1.0053,
|
| 116 |
"step": 1200
|
| 117 |
},
|
| 118 |
{
|
| 119 |
-
"epoch":
|
| 120 |
-
"learning_rate":
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 1300
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch":
|
| 126 |
-
"learning_rate":
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 1400
|
| 129 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
{
|
| 131 |
"epoch": 5.0,
|
| 132 |
-
"eval_accuracy": 0.
|
| 133 |
-
"eval_loss": 0.
|
| 134 |
-
"eval_runtime":
|
| 135 |
-
"eval_samples_per_second":
|
| 136 |
-
"eval_steps_per_second":
|
| 137 |
-
"step":
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 5.0,
|
| 141 |
-
"step":
|
| 142 |
-
"total_flos":
|
| 143 |
-
"train_loss":
|
| 144 |
-
"train_runtime":
|
| 145 |
-
"train_samples_per_second":
|
| 146 |
-
"train_steps_per_second":
|
| 147 |
}
|
| 148 |
],
|
| 149 |
-
"max_steps":
|
| 150 |
"num_train_epochs": 5,
|
| 151 |
-
"total_flos":
|
| 152 |
"trial_name": null,
|
| 153 |
"trial_params": null
|
| 154 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.02364550158381462,
|
| 3 |
+
"best_model_checkpoint": "/storage/img-cls-data/mnist_outputs/checkpoint-31875",
|
| 4 |
"epoch": 5.0,
|
| 5 |
+
"global_step": 31875,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 1.9937254901960785e-05,
|
| 13 |
+
"loss": 2.0433,
|
| 14 |
"step": 100
|
| 15 |
},
|
| 16 |
{
|
| 17 |
+
"epoch": 0.03,
|
| 18 |
+
"learning_rate": 1.987450980392157e-05,
|
| 19 |
+
"loss": 1.5446,
|
| 20 |
"step": 200
|
| 21 |
},
|
| 22 |
{
|
| 23 |
+
"epoch": 0.05,
|
| 24 |
+
"learning_rate": 1.9811764705882354e-05,
|
| 25 |
+
"loss": 1.2233,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
"step": 300
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.06,
|
| 30 |
+
"learning_rate": 1.9749019607843137e-05,
|
| 31 |
+
"loss": 1.0046,
|
| 32 |
"step": 400
|
| 33 |
},
|
| 34 |
{
|
| 35 |
+
"epoch": 0.08,
|
| 36 |
+
"learning_rate": 1.9686274509803924e-05,
|
| 37 |
+
"loss": 0.9327,
|
| 38 |
"step": 500
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.09,
|
| 42 |
+
"learning_rate": 1.9623529411764707e-05,
|
| 43 |
+
"loss": 0.8659,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
"step": 600
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.11,
|
| 48 |
+
"learning_rate": 1.9560784313725493e-05,
|
| 49 |
+
"loss": 0.7593,
|
| 50 |
"step": 700
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.13,
|
| 54 |
+
"learning_rate": 1.9498039215686276e-05,
|
| 55 |
+
"loss": 0.7057,
|
| 56 |
"step": 800
|
| 57 |
},
|
| 58 |
{
|
| 59 |
+
"epoch": 0.14,
|
| 60 |
+
"learning_rate": 1.9435294117647063e-05,
|
| 61 |
+
"loss": 0.6666,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
"step": 900
|
| 63 |
},
|
| 64 |
{
|
| 65 |
+
"epoch": 0.16,
|
| 66 |
+
"learning_rate": 1.9372549019607846e-05,
|
| 67 |
+
"loss": 0.6416,
|
| 68 |
"step": 1000
|
| 69 |
},
|
| 70 |
{
|
| 71 |
+
"epoch": 0.17,
|
| 72 |
+
"learning_rate": 1.930980392156863e-05,
|
| 73 |
+
"loss": 0.665,
|
| 74 |
"step": 1100
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 0.19,
|
| 78 |
+
"learning_rate": 1.9247058823529415e-05,
|
| 79 |
+
"loss": 0.6084,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 80 |
"step": 1200
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 0.2,
|
| 84 |
+
"learning_rate": 1.9184313725490198e-05,
|
| 85 |
+
"loss": 0.6203,
|
| 86 |
"step": 1300
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.22,
|
| 90 |
+
"learning_rate": 1.912156862745098e-05,
|
| 91 |
+
"loss": 0.6202,
|
| 92 |
"step": 1400
|
| 93 |
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.24,
|
| 96 |
+
"learning_rate": 1.9058823529411764e-05,
|
| 97 |
+
"loss": 0.5994,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.25,
|
| 102 |
+
"learning_rate": 1.899607843137255e-05,
|
| 103 |
+
"loss": 0.5837,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.27,
|
| 108 |
+
"learning_rate": 1.8933333333333334e-05,
|
| 109 |
+
"loss": 0.5787,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.28,
|
| 114 |
+
"learning_rate": 1.8870588235294117e-05,
|
| 115 |
+
"loss": 0.573,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.3,
|
| 120 |
+
"learning_rate": 1.8807843137254903e-05,
|
| 121 |
+
"loss": 0.5547,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.31,
|
| 126 |
+
"learning_rate": 1.8745098039215686e-05,
|
| 127 |
+
"loss": 0.5362,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.33,
|
| 132 |
+
"learning_rate": 1.8682352941176473e-05,
|
| 133 |
+
"loss": 0.5462,
|
| 134 |
+
"step": 2100
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.35,
|
| 138 |
+
"learning_rate": 1.8619607843137256e-05,
|
| 139 |
+
"loss": 0.4988,
|
| 140 |
+
"step": 2200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.36,
|
| 144 |
+
"learning_rate": 1.8556862745098042e-05,
|
| 145 |
+
"loss": 0.4175,
|
| 146 |
+
"step": 2300
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.38,
|
| 150 |
+
"learning_rate": 1.8494117647058825e-05,
|
| 151 |
+
"loss": 0.5065,
|
| 152 |
+
"step": 2400
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.39,
|
| 156 |
+
"learning_rate": 1.843137254901961e-05,
|
| 157 |
+
"loss": 0.5204,
|
| 158 |
+
"step": 2500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.41,
|
| 162 |
+
"learning_rate": 1.8368627450980395e-05,
|
| 163 |
+
"loss": 0.4434,
|
| 164 |
+
"step": 2600
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.42,
|
| 168 |
+
"learning_rate": 1.8305882352941178e-05,
|
| 169 |
+
"loss": 0.5012,
|
| 170 |
+
"step": 2700
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.44,
|
| 174 |
+
"learning_rate": 1.824313725490196e-05,
|
| 175 |
+
"loss": 0.5101,
|
| 176 |
+
"step": 2800
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.45,
|
| 180 |
+
"learning_rate": 1.8180392156862747e-05,
|
| 181 |
+
"loss": 0.4713,
|
| 182 |
+
"step": 2900
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.47,
|
| 186 |
+
"learning_rate": 1.811764705882353e-05,
|
| 187 |
+
"loss": 0.4327,
|
| 188 |
+
"step": 3000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.49,
|
| 192 |
+
"learning_rate": 1.8054901960784314e-05,
|
| 193 |
+
"loss": 0.5051,
|
| 194 |
+
"step": 3100
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.5,
|
| 198 |
+
"learning_rate": 1.79921568627451e-05,
|
| 199 |
+
"loss": 0.4257,
|
| 200 |
+
"step": 3200
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.52,
|
| 204 |
+
"learning_rate": 1.7929411764705883e-05,
|
| 205 |
+
"loss": 0.427,
|
| 206 |
+
"step": 3300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.53,
|
| 210 |
+
"learning_rate": 1.7866666666666666e-05,
|
| 211 |
+
"loss": 0.4886,
|
| 212 |
+
"step": 3400
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.55,
|
| 216 |
+
"learning_rate": 1.7803921568627453e-05,
|
| 217 |
+
"loss": 0.4276,
|
| 218 |
+
"step": 3500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.56,
|
| 222 |
+
"learning_rate": 1.7741176470588236e-05,
|
| 223 |
+
"loss": 0.4275,
|
| 224 |
+
"step": 3600
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.58,
|
| 228 |
+
"learning_rate": 1.7678431372549022e-05,
|
| 229 |
+
"loss": 0.4558,
|
| 230 |
+
"step": 3700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.6,
|
| 234 |
+
"learning_rate": 1.7615686274509805e-05,
|
| 235 |
+
"loss": 0.4459,
|
| 236 |
+
"step": 3800
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.61,
|
| 240 |
+
"learning_rate": 1.755294117647059e-05,
|
| 241 |
+
"loss": 0.5111,
|
| 242 |
+
"step": 3900
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.63,
|
| 246 |
+
"learning_rate": 1.7490196078431375e-05,
|
| 247 |
+
"loss": 0.3821,
|
| 248 |
+
"step": 4000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.64,
|
| 252 |
+
"learning_rate": 1.7427450980392158e-05,
|
| 253 |
+
"loss": 0.4291,
|
| 254 |
+
"step": 4100
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.66,
|
| 258 |
+
"learning_rate": 1.7364705882352944e-05,
|
| 259 |
+
"loss": 0.4001,
|
| 260 |
+
"step": 4200
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.67,
|
| 264 |
+
"learning_rate": 1.7301960784313727e-05,
|
| 265 |
+
"loss": 0.4144,
|
| 266 |
+
"step": 4300
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.69,
|
| 270 |
+
"learning_rate": 1.723921568627451e-05,
|
| 271 |
+
"loss": 0.3786,
|
| 272 |
+
"step": 4400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.71,
|
| 276 |
+
"learning_rate": 1.7176470588235293e-05,
|
| 277 |
+
"loss": 0.4464,
|
| 278 |
+
"step": 4500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.72,
|
| 282 |
+
"learning_rate": 1.711372549019608e-05,
|
| 283 |
+
"loss": 0.449,
|
| 284 |
+
"step": 4600
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.74,
|
| 288 |
+
"learning_rate": 1.7050980392156863e-05,
|
| 289 |
+
"loss": 0.4206,
|
| 290 |
+
"step": 4700
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.75,
|
| 294 |
+
"learning_rate": 1.698823529411765e-05,
|
| 295 |
+
"loss": 0.4625,
|
| 296 |
+
"step": 4800
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.77,
|
| 300 |
+
"learning_rate": 1.6925490196078432e-05,
|
| 301 |
+
"loss": 0.4383,
|
| 302 |
+
"step": 4900
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.78,
|
| 306 |
+
"learning_rate": 1.686274509803922e-05,
|
| 307 |
+
"loss": 0.3844,
|
| 308 |
+
"step": 5000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.8,
|
| 312 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 313 |
+
"loss": 0.4103,
|
| 314 |
+
"step": 5100
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.82,
|
| 318 |
+
"learning_rate": 1.6737254901960788e-05,
|
| 319 |
+
"loss": 0.4965,
|
| 320 |
+
"step": 5200
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.83,
|
| 324 |
+
"learning_rate": 1.667450980392157e-05,
|
| 325 |
+
"loss": 0.368,
|
| 326 |
+
"step": 5300
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.85,
|
| 330 |
+
"learning_rate": 1.6611764705882354e-05,
|
| 331 |
+
"loss": 0.4474,
|
| 332 |
+
"step": 5400
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.86,
|
| 336 |
+
"learning_rate": 1.6549019607843137e-05,
|
| 337 |
+
"loss": 0.4173,
|
| 338 |
+
"step": 5500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.88,
|
| 342 |
+
"learning_rate": 1.6486274509803924e-05,
|
| 343 |
+
"loss": 0.4226,
|
| 344 |
+
"step": 5600
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.89,
|
| 348 |
+
"learning_rate": 1.6423529411764707e-05,
|
| 349 |
+
"loss": 0.4163,
|
| 350 |
+
"step": 5700
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.91,
|
| 354 |
+
"learning_rate": 1.636078431372549e-05,
|
| 355 |
+
"loss": 0.4096,
|
| 356 |
+
"step": 5800
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.93,
|
| 360 |
+
"learning_rate": 1.6298039215686276e-05,
|
| 361 |
+
"loss": 0.3749,
|
| 362 |
+
"step": 5900
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.94,
|
| 366 |
+
"learning_rate": 1.623529411764706e-05,
|
| 367 |
+
"loss": 0.4418,
|
| 368 |
+
"step": 6000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.96,
|
| 372 |
+
"learning_rate": 1.6172549019607842e-05,
|
| 373 |
+
"loss": 0.4591,
|
| 374 |
+
"step": 6100
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.97,
|
| 378 |
+
"learning_rate": 1.610980392156863e-05,
|
| 379 |
+
"loss": 0.4181,
|
| 380 |
+
"step": 6200
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.99,
|
| 384 |
+
"learning_rate": 1.6047058823529412e-05,
|
| 385 |
+
"loss": 0.3717,
|
| 386 |
+
"step": 6300
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.0,
|
| 390 |
+
"eval_accuracy": 0.9893333333333333,
|
| 391 |
+
"eval_loss": 0.05222630500793457,
|
| 392 |
+
"eval_runtime": 45.5882,
|
| 393 |
+
"eval_samples_per_second": 197.42,
|
| 394 |
+
"eval_steps_per_second": 24.677,
|
| 395 |
+
"step": 6375
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.0,
|
| 399 |
+
"learning_rate": 1.59843137254902e-05,
|
| 400 |
+
"loss": 0.3852,
|
| 401 |
+
"step": 6400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.02,
|
| 405 |
+
"learning_rate": 1.592156862745098e-05,
|
| 406 |
+
"loss": 0.4254,
|
| 407 |
+
"step": 6500
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.04,
|
| 411 |
+
"learning_rate": 1.5858823529411768e-05,
|
| 412 |
+
"loss": 0.4035,
|
| 413 |
+
"step": 6600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.05,
|
| 417 |
+
"learning_rate": 1.579607843137255e-05,
|
| 418 |
+
"loss": 0.3704,
|
| 419 |
+
"step": 6700
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.07,
|
| 423 |
+
"learning_rate": 1.5733333333333334e-05,
|
| 424 |
+
"loss": 0.4037,
|
| 425 |
+
"step": 6800
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.08,
|
| 429 |
+
"learning_rate": 1.567058823529412e-05,
|
| 430 |
+
"loss": 0.3574,
|
| 431 |
+
"step": 6900
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.1,
|
| 435 |
+
"learning_rate": 1.5607843137254904e-05,
|
| 436 |
+
"loss": 0.3766,
|
| 437 |
+
"step": 7000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.11,
|
| 441 |
+
"learning_rate": 1.5545098039215687e-05,
|
| 442 |
+
"loss": 0.3677,
|
| 443 |
+
"step": 7100
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.13,
|
| 447 |
+
"learning_rate": 1.5482352941176473e-05,
|
| 448 |
+
"loss": 0.5049,
|
| 449 |
+
"step": 7200
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.15,
|
| 453 |
+
"learning_rate": 1.5419607843137256e-05,
|
| 454 |
+
"loss": 0.3507,
|
| 455 |
+
"step": 7300
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.16,
|
| 459 |
+
"learning_rate": 1.535686274509804e-05,
|
| 460 |
+
"loss": 0.4128,
|
| 461 |
+
"step": 7400
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.18,
|
| 465 |
+
"learning_rate": 1.5294117647058822e-05,
|
| 466 |
+
"loss": 0.3476,
|
| 467 |
+
"step": 7500
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 1.19,
|
| 471 |
+
"learning_rate": 1.523137254901961e-05,
|
| 472 |
+
"loss": 0.3652,
|
| 473 |
+
"step": 7600
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.21,
|
| 477 |
+
"learning_rate": 1.5168627450980393e-05,
|
| 478 |
+
"loss": 0.3738,
|
| 479 |
+
"step": 7700
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.22,
|
| 483 |
+
"learning_rate": 1.5105882352941176e-05,
|
| 484 |
+
"loss": 0.3927,
|
| 485 |
+
"step": 7800
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.24,
|
| 489 |
+
"learning_rate": 1.5043137254901963e-05,
|
| 490 |
+
"loss": 0.3556,
|
| 491 |
+
"step": 7900
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 1.25,
|
| 495 |
+
"learning_rate": 1.4980392156862746e-05,
|
| 496 |
+
"loss": 0.3917,
|
| 497 |
+
"step": 8000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1.27,
|
| 501 |
+
"learning_rate": 1.491764705882353e-05,
|
| 502 |
+
"loss": 0.3965,
|
| 503 |
+
"step": 8100
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.29,
|
| 507 |
+
"learning_rate": 1.4854901960784315e-05,
|
| 508 |
+
"loss": 0.3356,
|
| 509 |
+
"step": 8200
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 1.3,
|
| 513 |
+
"learning_rate": 1.47921568627451e-05,
|
| 514 |
+
"loss": 0.3583,
|
| 515 |
+
"step": 8300
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.32,
|
| 519 |
+
"learning_rate": 1.4729411764705883e-05,
|
| 520 |
+
"loss": 0.4065,
|
| 521 |
+
"step": 8400
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.33,
|
| 525 |
+
"learning_rate": 1.4666666666666666e-05,
|
| 526 |
+
"loss": 0.4424,
|
| 527 |
+
"step": 8500
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 1.35,
|
| 531 |
+
"learning_rate": 1.4603921568627453e-05,
|
| 532 |
+
"loss": 0.3177,
|
| 533 |
+
"step": 8600
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 1.36,
|
| 537 |
+
"learning_rate": 1.4541176470588236e-05,
|
| 538 |
+
"loss": 0.3676,
|
| 539 |
+
"step": 8700
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 1.38,
|
| 543 |
+
"learning_rate": 1.447843137254902e-05,
|
| 544 |
+
"loss": 0.3968,
|
| 545 |
+
"step": 8800
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.4,
|
| 549 |
+
"learning_rate": 1.4415686274509805e-05,
|
| 550 |
+
"loss": 0.3891,
|
| 551 |
+
"step": 8900
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 1.41,
|
| 555 |
+
"learning_rate": 1.435294117647059e-05,
|
| 556 |
+
"loss": 0.4299,
|
| 557 |
+
"step": 9000
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.43,
|
| 561 |
+
"learning_rate": 1.4290196078431373e-05,
|
| 562 |
+
"loss": 0.3649,
|
| 563 |
+
"step": 9100
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.44,
|
| 567 |
+
"learning_rate": 1.422745098039216e-05,
|
| 568 |
+
"loss": 0.3936,
|
| 569 |
+
"step": 9200
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 1.46,
|
| 573 |
+
"learning_rate": 1.4164705882352943e-05,
|
| 574 |
+
"loss": 0.3488,
|
| 575 |
+
"step": 9300
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 1.47,
|
| 579 |
+
"learning_rate": 1.4101960784313726e-05,
|
| 580 |
+
"loss": 0.3544,
|
| 581 |
+
"step": 9400
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 1.49,
|
| 585 |
+
"learning_rate": 1.403921568627451e-05,
|
| 586 |
+
"loss": 0.3886,
|
| 587 |
+
"step": 9500
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.51,
|
| 591 |
+
"learning_rate": 1.3976470588235295e-05,
|
| 592 |
+
"loss": 0.3687,
|
| 593 |
+
"step": 9600
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.52,
|
| 597 |
+
"learning_rate": 1.391372549019608e-05,
|
| 598 |
+
"loss": 0.3756,
|
| 599 |
+
"step": 9700
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.54,
|
| 603 |
+
"learning_rate": 1.3850980392156863e-05,
|
| 604 |
+
"loss": 0.3966,
|
| 605 |
+
"step": 9800
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.55,
|
| 609 |
+
"learning_rate": 1.378823529411765e-05,
|
| 610 |
+
"loss": 0.3965,
|
| 611 |
+
"step": 9900
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.57,
|
| 615 |
+
"learning_rate": 1.3725490196078432e-05,
|
| 616 |
+
"loss": 0.4041,
|
| 617 |
+
"step": 10000
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 1.58,
|
| 621 |
+
"learning_rate": 1.3662745098039215e-05,
|
| 622 |
+
"loss": 0.321,
|
| 623 |
+
"step": 10100
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.6,
|
| 627 |
+
"learning_rate": 1.3600000000000002e-05,
|
| 628 |
+
"loss": 0.34,
|
| 629 |
+
"step": 10200
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.62,
|
| 633 |
+
"learning_rate": 1.3537254901960785e-05,
|
| 634 |
+
"loss": 0.3859,
|
| 635 |
+
"step": 10300
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 1.63,
|
| 639 |
+
"learning_rate": 1.347450980392157e-05,
|
| 640 |
+
"loss": 0.3574,
|
| 641 |
+
"step": 10400
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.65,
|
| 645 |
+
"learning_rate": 1.3411764705882353e-05,
|
| 646 |
+
"loss": 0.3645,
|
| 647 |
+
"step": 10500
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.66,
|
| 651 |
+
"learning_rate": 1.334901960784314e-05,
|
| 652 |
+
"loss": 0.3378,
|
| 653 |
+
"step": 10600
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 1.68,
|
| 657 |
+
"learning_rate": 1.3286274509803922e-05,
|
| 658 |
+
"loss": 0.4109,
|
| 659 |
+
"step": 10700
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 1.69,
|
| 663 |
+
"learning_rate": 1.3223529411764705e-05,
|
| 664 |
+
"loss": 0.3732,
|
| 665 |
+
"step": 10800
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.71,
|
| 669 |
+
"learning_rate": 1.3160784313725492e-05,
|
| 670 |
+
"loss": 0.3548,
|
| 671 |
+
"step": 10900
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.73,
|
| 675 |
+
"learning_rate": 1.3098039215686275e-05,
|
| 676 |
+
"loss": 0.3076,
|
| 677 |
+
"step": 11000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 1.74,
|
| 681 |
+
"learning_rate": 1.303529411764706e-05,
|
| 682 |
+
"loss": 0.3707,
|
| 683 |
+
"step": 11100
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.76,
|
| 687 |
+
"learning_rate": 1.2972549019607846e-05,
|
| 688 |
+
"loss": 0.367,
|
| 689 |
+
"step": 11200
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.77,
|
| 693 |
+
"learning_rate": 1.2909803921568629e-05,
|
| 694 |
+
"loss": 0.3242,
|
| 695 |
+
"step": 11300
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 1.79,
|
| 699 |
+
"learning_rate": 1.2847058823529412e-05,
|
| 700 |
+
"loss": 0.3913,
|
| 701 |
+
"step": 11400
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.8,
|
| 705 |
+
"learning_rate": 1.2784313725490197e-05,
|
| 706 |
+
"loss": 0.3605,
|
| 707 |
+
"step": 11500
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.82,
|
| 711 |
+
"learning_rate": 1.2721568627450982e-05,
|
| 712 |
+
"loss": 0.352,
|
| 713 |
+
"step": 11600
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.84,
|
| 717 |
+
"learning_rate": 1.2658823529411766e-05,
|
| 718 |
+
"loss": 0.3326,
|
| 719 |
+
"step": 11700
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 1.85,
|
| 723 |
+
"learning_rate": 1.259607843137255e-05,
|
| 724 |
+
"loss": 0.3601,
|
| 725 |
+
"step": 11800
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.87,
|
| 729 |
+
"learning_rate": 1.2533333333333336e-05,
|
| 730 |
+
"loss": 0.389,
|
| 731 |
+
"step": 11900
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.88,
|
| 735 |
+
"learning_rate": 1.2470588235294119e-05,
|
| 736 |
+
"loss": 0.3516,
|
| 737 |
+
"step": 12000
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.9,
|
| 741 |
+
"learning_rate": 1.2407843137254902e-05,
|
| 742 |
+
"loss": 0.3759,
|
| 743 |
+
"step": 12100
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.91,
|
| 747 |
+
"learning_rate": 1.2345098039215688e-05,
|
| 748 |
+
"loss": 0.3405,
|
| 749 |
+
"step": 12200
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.93,
|
| 753 |
+
"learning_rate": 1.2282352941176471e-05,
|
| 754 |
+
"loss": 0.3621,
|
| 755 |
+
"step": 12300
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.95,
|
| 759 |
+
"learning_rate": 1.2219607843137256e-05,
|
| 760 |
+
"loss": 0.3429,
|
| 761 |
+
"step": 12400
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 1.96,
|
| 765 |
+
"learning_rate": 1.215686274509804e-05,
|
| 766 |
+
"loss": 0.3789,
|
| 767 |
+
"step": 12500
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.98,
|
| 771 |
+
"learning_rate": 1.2094117647058826e-05,
|
| 772 |
+
"loss": 0.3011,
|
| 773 |
+
"step": 12600
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.99,
|
| 777 |
+
"learning_rate": 1.2031372549019609e-05,
|
| 778 |
+
"loss": 0.3453,
|
| 779 |
+
"step": 12700
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 2.0,
|
| 783 |
+
"eval_accuracy": 0.9905555555555555,
|
| 784 |
+
"eval_loss": 0.0370011106133461,
|
| 785 |
+
"eval_runtime": 47.336,
|
| 786 |
+
"eval_samples_per_second": 190.13,
|
| 787 |
+
"eval_steps_per_second": 23.766,
|
| 788 |
+
"step": 12750
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.01,
|
| 792 |
+
"learning_rate": 1.1968627450980392e-05,
|
| 793 |
+
"loss": 0.3666,
|
| 794 |
+
"step": 12800
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 2.02,
|
| 798 |
+
"learning_rate": 1.1905882352941178e-05,
|
| 799 |
+
"loss": 0.3447,
|
| 800 |
+
"step": 12900
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 2.04,
|
| 804 |
+
"learning_rate": 1.1843137254901961e-05,
|
| 805 |
+
"loss": 0.3669,
|
| 806 |
+
"step": 13000
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.05,
|
| 810 |
+
"learning_rate": 1.1780392156862746e-05,
|
| 811 |
+
"loss": 0.3423,
|
| 812 |
+
"step": 13100
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 2.07,
|
| 816 |
+
"learning_rate": 1.171764705882353e-05,
|
| 817 |
+
"loss": 0.3372,
|
| 818 |
+
"step": 13200
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.09,
|
| 822 |
+
"learning_rate": 1.1654901960784316e-05,
|
| 823 |
+
"loss": 0.356,
|
| 824 |
+
"step": 13300
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.1,
|
| 828 |
+
"learning_rate": 1.1592156862745099e-05,
|
| 829 |
+
"loss": 0.306,
|
| 830 |
+
"step": 13400
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 2.12,
|
| 834 |
+
"learning_rate": 1.1529411764705882e-05,
|
| 835 |
+
"loss": 0.354,
|
| 836 |
+
"step": 13500
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 2.13,
|
| 840 |
+
"learning_rate": 1.1466666666666668e-05,
|
| 841 |
+
"loss": 0.3664,
|
| 842 |
+
"step": 13600
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.15,
|
| 846 |
+
"learning_rate": 1.1403921568627451e-05,
|
| 847 |
+
"loss": 0.3568,
|
| 848 |
+
"step": 13700
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 2.16,
|
| 852 |
+
"learning_rate": 1.1341176470588236e-05,
|
| 853 |
+
"loss": 0.3334,
|
| 854 |
+
"step": 13800
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 2.18,
|
| 858 |
+
"learning_rate": 1.127843137254902e-05,
|
| 859 |
+
"loss": 0.3405,
|
| 860 |
+
"step": 13900
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.2,
|
| 864 |
+
"learning_rate": 1.1215686274509805e-05,
|
| 865 |
+
"loss": 0.3107,
|
| 866 |
+
"step": 14000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 2.21,
|
| 870 |
+
"learning_rate": 1.1152941176470588e-05,
|
| 871 |
+
"loss": 0.2895,
|
| 872 |
+
"step": 14100
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 2.23,
|
| 876 |
+
"learning_rate": 1.1090196078431375e-05,
|
| 877 |
+
"loss": 0.2832,
|
| 878 |
+
"step": 14200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.24,
|
| 882 |
+
"learning_rate": 1.1027450980392158e-05,
|
| 883 |
+
"loss": 0.3823,
|
| 884 |
+
"step": 14300
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.26,
|
| 888 |
+
"learning_rate": 1.0964705882352941e-05,
|
| 889 |
+
"loss": 0.2835,
|
| 890 |
+
"step": 14400
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 2.27,
|
| 894 |
+
"learning_rate": 1.0901960784313726e-05,
|
| 895 |
+
"loss": 0.3859,
|
| 896 |
+
"step": 14500
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 2.29,
|
| 900 |
+
"learning_rate": 1.083921568627451e-05,
|
| 901 |
+
"loss": 0.3486,
|
| 902 |
+
"step": 14600
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 2.31,
|
| 906 |
+
"learning_rate": 1.0776470588235295e-05,
|
| 907 |
+
"loss": 0.3202,
|
| 908 |
+
"step": 14700
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 2.32,
|
| 912 |
+
"learning_rate": 1.0713725490196078e-05,
|
| 913 |
+
"loss": 0.3276,
|
| 914 |
+
"step": 14800
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 2.34,
|
| 918 |
+
"learning_rate": 1.0650980392156865e-05,
|
| 919 |
+
"loss": 0.3562,
|
| 920 |
+
"step": 14900
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.35,
|
| 924 |
+
"learning_rate": 1.0588235294117648e-05,
|
| 925 |
+
"loss": 0.3344,
|
| 926 |
+
"step": 15000
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 2.37,
|
| 930 |
+
"learning_rate": 1.0525490196078431e-05,
|
| 931 |
+
"loss": 0.3356,
|
| 932 |
+
"step": 15100
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.38,
|
| 936 |
+
"learning_rate": 1.0462745098039217e-05,
|
| 937 |
+
"loss": 0.322,
|
| 938 |
+
"step": 15200
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 2.4,
|
| 942 |
+
"learning_rate": 1.04e-05,
|
| 943 |
+
"loss": 0.3114,
|
| 944 |
+
"step": 15300
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.42,
|
| 948 |
+
"learning_rate": 1.0337254901960785e-05,
|
| 949 |
+
"loss": 0.3401,
|
| 950 |
+
"step": 15400
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 2.43,
|
| 954 |
+
"learning_rate": 1.0274509803921568e-05,
|
| 955 |
+
"loss": 0.3624,
|
| 956 |
+
"step": 15500
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 2.45,
|
| 960 |
+
"learning_rate": 1.0211764705882355e-05,
|
| 961 |
+
"loss": 0.3468,
|
| 962 |
+
"step": 15600
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.46,
|
| 966 |
+
"learning_rate": 1.0149019607843138e-05,
|
| 967 |
+
"loss": 0.3263,
|
| 968 |
+
"step": 15700
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 2.48,
|
| 972 |
+
"learning_rate": 1.0086274509803922e-05,
|
| 973 |
+
"loss": 0.317,
|
| 974 |
+
"step": 15800
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.49,
|
| 978 |
+
"learning_rate": 1.0023529411764707e-05,
|
| 979 |
+
"loss": 0.4155,
|
| 980 |
+
"step": 15900
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 2.51,
|
| 984 |
+
"learning_rate": 9.960784313725492e-06,
|
| 985 |
+
"loss": 0.2799,
|
| 986 |
+
"step": 16000
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 2.53,
|
| 990 |
+
"learning_rate": 9.898039215686275e-06,
|
| 991 |
+
"loss": 0.3184,
|
| 992 |
+
"step": 16100
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 2.54,
|
| 996 |
+
"learning_rate": 9.83529411764706e-06,
|
| 997 |
+
"loss": 0.3319,
|
| 998 |
+
"step": 16200
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.56,
|
| 1002 |
+
"learning_rate": 9.772549019607844e-06,
|
| 1003 |
+
"loss": 0.339,
|
| 1004 |
+
"step": 16300
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.57,
|
| 1008 |
+
"learning_rate": 9.709803921568628e-06,
|
| 1009 |
+
"loss": 0.3767,
|
| 1010 |
+
"step": 16400
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 2.59,
|
| 1014 |
+
"learning_rate": 9.647058823529412e-06,
|
| 1015 |
+
"loss": 0.2787,
|
| 1016 |
+
"step": 16500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.6,
|
| 1020 |
+
"learning_rate": 9.584313725490197e-06,
|
| 1021 |
+
"loss": 0.2856,
|
| 1022 |
+
"step": 16600
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 2.62,
|
| 1026 |
+
"learning_rate": 9.521568627450982e-06,
|
| 1027 |
+
"loss": 0.3085,
|
| 1028 |
+
"step": 16700
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 2.64,
|
| 1032 |
+
"learning_rate": 9.458823529411767e-06,
|
| 1033 |
+
"loss": 0.3445,
|
| 1034 |
+
"step": 16800
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 2.65,
|
| 1038 |
+
"learning_rate": 9.39607843137255e-06,
|
| 1039 |
+
"loss": 0.3586,
|
| 1040 |
+
"step": 16900
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 2.67,
|
| 1044 |
+
"learning_rate": 9.333333333333334e-06,
|
| 1045 |
+
"loss": 0.2763,
|
| 1046 |
+
"step": 17000
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.68,
|
| 1050 |
+
"learning_rate": 9.270588235294117e-06,
|
| 1051 |
+
"loss": 0.3879,
|
| 1052 |
+
"step": 17100
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 2.7,
|
| 1056 |
+
"learning_rate": 9.207843137254902e-06,
|
| 1057 |
+
"loss": 0.3333,
|
| 1058 |
+
"step": 17200
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 2.71,
|
| 1062 |
+
"learning_rate": 9.145098039215687e-06,
|
| 1063 |
+
"loss": 0.3134,
|
| 1064 |
+
"step": 17300
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 2.73,
|
| 1068 |
+
"learning_rate": 9.082352941176472e-06,
|
| 1069 |
+
"loss": 0.3143,
|
| 1070 |
+
"step": 17400
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 2.75,
|
| 1074 |
+
"learning_rate": 9.019607843137256e-06,
|
| 1075 |
+
"loss": 0.3189,
|
| 1076 |
+
"step": 17500
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 2.76,
|
| 1080 |
+
"learning_rate": 8.95686274509804e-06,
|
| 1081 |
+
"loss": 0.3751,
|
| 1082 |
+
"step": 17600
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 2.78,
|
| 1086 |
+
"learning_rate": 8.894117647058824e-06,
|
| 1087 |
+
"loss": 0.3392,
|
| 1088 |
+
"step": 17700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.79,
|
| 1092 |
+
"learning_rate": 8.831372549019609e-06,
|
| 1093 |
+
"loss": 0.4085,
|
| 1094 |
+
"step": 17800
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 2.81,
|
| 1098 |
+
"learning_rate": 8.768627450980392e-06,
|
| 1099 |
+
"loss": 0.351,
|
| 1100 |
+
"step": 17900
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 2.82,
|
| 1104 |
+
"learning_rate": 8.705882352941177e-06,
|
| 1105 |
+
"loss": 0.4239,
|
| 1106 |
+
"step": 18000
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 2.84,
|
| 1110 |
+
"learning_rate": 8.643137254901961e-06,
|
| 1111 |
+
"loss": 0.3577,
|
| 1112 |
+
"step": 18100
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.85,
|
| 1116 |
+
"learning_rate": 8.580392156862746e-06,
|
| 1117 |
+
"loss": 0.3651,
|
| 1118 |
+
"step": 18200
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 2.87,
|
| 1122 |
+
"learning_rate": 8.517647058823531e-06,
|
| 1123 |
+
"loss": 0.2912,
|
| 1124 |
+
"step": 18300
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 2.89,
|
| 1128 |
+
"learning_rate": 8.454901960784314e-06,
|
| 1129 |
+
"loss": 0.3367,
|
| 1130 |
+
"step": 18400
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.9,
|
| 1134 |
+
"learning_rate": 8.392156862745099e-06,
|
| 1135 |
+
"loss": 0.3616,
|
| 1136 |
+
"step": 18500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 2.92,
|
| 1140 |
+
"learning_rate": 8.329411764705882e-06,
|
| 1141 |
+
"loss": 0.3608,
|
| 1142 |
+
"step": 18600
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 2.93,
|
| 1146 |
+
"learning_rate": 8.266666666666667e-06,
|
| 1147 |
+
"loss": 0.3373,
|
| 1148 |
+
"step": 18700
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 2.95,
|
| 1152 |
+
"learning_rate": 8.203921568627451e-06,
|
| 1153 |
+
"loss": 0.3132,
|
| 1154 |
+
"step": 18800
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.96,
|
| 1158 |
+
"learning_rate": 8.141176470588236e-06,
|
| 1159 |
+
"loss": 0.3279,
|
| 1160 |
+
"step": 18900
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 2.98,
|
| 1164 |
+
"learning_rate": 8.07843137254902e-06,
|
| 1165 |
+
"loss": 0.3052,
|
| 1166 |
+
"step": 19000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 3.0,
|
| 1170 |
+
"learning_rate": 8.015686274509804e-06,
|
| 1171 |
+
"loss": 0.3736,
|
| 1172 |
+
"step": 19100
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 3.0,
|
| 1176 |
+
"eval_accuracy": 0.9915555555555555,
|
| 1177 |
+
"eval_loss": 0.030801517888903618,
|
| 1178 |
+
"eval_runtime": 47.8145,
|
| 1179 |
+
"eval_samples_per_second": 188.227,
|
| 1180 |
+
"eval_steps_per_second": 23.528,
|
| 1181 |
+
"step": 19125
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 3.01,
|
| 1185 |
+
"learning_rate": 7.952941176470589e-06,
|
| 1186 |
+
"loss": 0.3429,
|
| 1187 |
+
"step": 19200
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 3.03,
|
| 1191 |
+
"learning_rate": 7.890196078431373e-06,
|
| 1192 |
+
"loss": 0.3234,
|
| 1193 |
+
"step": 19300
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 3.04,
|
| 1197 |
+
"learning_rate": 7.827450980392156e-06,
|
| 1198 |
+
"loss": 0.3176,
|
| 1199 |
+
"step": 19400
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 3.06,
|
| 1203 |
+
"learning_rate": 7.764705882352941e-06,
|
| 1204 |
+
"loss": 0.3483,
|
| 1205 |
+
"step": 19500
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 3.07,
|
| 1209 |
+
"learning_rate": 7.701960784313726e-06,
|
| 1210 |
+
"loss": 0.2904,
|
| 1211 |
+
"step": 19600
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 3.09,
|
| 1215 |
+
"learning_rate": 7.63921568627451e-06,
|
| 1216 |
+
"loss": 0.3505,
|
| 1217 |
+
"step": 19700
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 3.11,
|
| 1221 |
+
"learning_rate": 7.576470588235295e-06,
|
| 1222 |
+
"loss": 0.2773,
|
| 1223 |
+
"step": 19800
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 3.12,
|
| 1227 |
+
"learning_rate": 7.5137254901960785e-06,
|
| 1228 |
+
"loss": 0.3035,
|
| 1229 |
+
"step": 19900
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 3.14,
|
| 1233 |
+
"learning_rate": 7.450980392156863e-06,
|
| 1234 |
+
"loss": 0.262,
|
| 1235 |
+
"step": 20000
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 3.15,
|
| 1239 |
+
"learning_rate": 7.388235294117647e-06,
|
| 1240 |
+
"loss": 0.288,
|
| 1241 |
+
"step": 20100
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 3.17,
|
| 1245 |
+
"learning_rate": 7.325490196078432e-06,
|
| 1246 |
+
"loss": 0.2816,
|
| 1247 |
+
"step": 20200
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 3.18,
|
| 1251 |
+
"learning_rate": 7.262745098039217e-06,
|
| 1252 |
+
"loss": 0.3427,
|
| 1253 |
+
"step": 20300
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 3.2,
|
| 1257 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 1258 |
+
"loss": 0.3235,
|
| 1259 |
+
"step": 20400
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 3.22,
|
| 1263 |
+
"learning_rate": 7.137254901960785e-06,
|
| 1264 |
+
"loss": 0.305,
|
| 1265 |
+
"step": 20500
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 3.23,
|
| 1269 |
+
"learning_rate": 7.074509803921568e-06,
|
| 1270 |
+
"loss": 0.3213,
|
| 1271 |
+
"step": 20600
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 3.25,
|
| 1275 |
+
"learning_rate": 7.011764705882353e-06,
|
| 1276 |
+
"loss": 0.2998,
|
| 1277 |
+
"step": 20700
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.26,
|
| 1281 |
+
"learning_rate": 6.949019607843138e-06,
|
| 1282 |
+
"loss": 0.2733,
|
| 1283 |
+
"step": 20800
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 3.28,
|
| 1287 |
+
"learning_rate": 6.886274509803922e-06,
|
| 1288 |
+
"loss": 0.2812,
|
| 1289 |
+
"step": 20900
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 3.29,
|
| 1293 |
+
"learning_rate": 6.8235294117647065e-06,
|
| 1294 |
+
"loss": 0.2979,
|
| 1295 |
+
"step": 21000
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 3.31,
|
| 1299 |
+
"learning_rate": 6.76078431372549e-06,
|
| 1300 |
+
"loss": 0.3359,
|
| 1301 |
+
"step": 21100
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 3.33,
|
| 1305 |
+
"learning_rate": 6.698039215686275e-06,
|
| 1306 |
+
"loss": 0.2952,
|
| 1307 |
+
"step": 21200
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 3.34,
|
| 1311 |
+
"learning_rate": 6.63529411764706e-06,
|
| 1312 |
+
"loss": 0.3092,
|
| 1313 |
+
"step": 21300
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 3.36,
|
| 1317 |
+
"learning_rate": 6.572549019607844e-06,
|
| 1318 |
+
"loss": 0.2614,
|
| 1319 |
+
"step": 21400
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 3.37,
|
| 1323 |
+
"learning_rate": 6.5098039215686285e-06,
|
| 1324 |
+
"loss": 0.3431,
|
| 1325 |
+
"step": 21500
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 3.39,
|
| 1329 |
+
"learning_rate": 6.4470588235294116e-06,
|
| 1330 |
+
"loss": 0.355,
|
| 1331 |
+
"step": 21600
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 3.4,
|
| 1335 |
+
"learning_rate": 6.384313725490196e-06,
|
| 1336 |
+
"loss": 0.2789,
|
| 1337 |
+
"step": 21700
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 3.42,
|
| 1341 |
+
"learning_rate": 6.321568627450981e-06,
|
| 1342 |
+
"loss": 0.3669,
|
| 1343 |
+
"step": 21800
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 3.44,
|
| 1347 |
+
"learning_rate": 6.258823529411765e-06,
|
| 1348 |
+
"loss": 0.2962,
|
| 1349 |
+
"step": 21900
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 3.45,
|
| 1353 |
+
"learning_rate": 6.19607843137255e-06,
|
| 1354 |
+
"loss": 0.3174,
|
| 1355 |
+
"step": 22000
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 3.47,
|
| 1359 |
+
"learning_rate": 6.133333333333334e-06,
|
| 1360 |
+
"loss": 0.3251,
|
| 1361 |
+
"step": 22100
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.48,
|
| 1365 |
+
"learning_rate": 6.070588235294118e-06,
|
| 1366 |
+
"loss": 0.3237,
|
| 1367 |
+
"step": 22200
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 3.5,
|
| 1371 |
+
"learning_rate": 6.007843137254903e-06,
|
| 1372 |
+
"loss": 0.2878,
|
| 1373 |
+
"step": 22300
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 3.51,
|
| 1377 |
+
"learning_rate": 5.945098039215686e-06,
|
| 1378 |
+
"loss": 0.2878,
|
| 1379 |
+
"step": 22400
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 3.53,
|
| 1383 |
+
"learning_rate": 5.882352941176471e-06,
|
| 1384 |
+
"loss": 0.3223,
|
| 1385 |
+
"step": 22500
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 3.55,
|
| 1389 |
+
"learning_rate": 5.819607843137255e-06,
|
| 1390 |
+
"loss": 0.3064,
|
| 1391 |
+
"step": 22600
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 3.56,
|
| 1395 |
+
"learning_rate": 5.7568627450980396e-06,
|
| 1396 |
+
"loss": 0.3291,
|
| 1397 |
+
"step": 22700
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 3.58,
|
| 1401 |
+
"learning_rate": 5.694117647058824e-06,
|
| 1402 |
+
"loss": 0.3049,
|
| 1403 |
+
"step": 22800
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.59,
|
| 1407 |
+
"learning_rate": 5.631372549019608e-06,
|
| 1408 |
+
"loss": 0.3109,
|
| 1409 |
+
"step": 22900
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 3.61,
|
| 1413 |
+
"learning_rate": 5.568627450980393e-06,
|
| 1414 |
+
"loss": 0.374,
|
| 1415 |
+
"step": 23000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 3.62,
|
| 1419 |
+
"learning_rate": 5.505882352941177e-06,
|
| 1420 |
+
"loss": 0.3325,
|
| 1421 |
+
"step": 23100
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 3.64,
|
| 1425 |
+
"learning_rate": 5.443137254901962e-06,
|
| 1426 |
+
"loss": 0.3247,
|
| 1427 |
+
"step": 23200
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 3.65,
|
| 1431 |
+
"learning_rate": 5.380392156862746e-06,
|
| 1432 |
+
"loss": 0.3483,
|
| 1433 |
+
"step": 23300
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 3.67,
|
| 1437 |
+
"learning_rate": 5.317647058823529e-06,
|
| 1438 |
+
"loss": 0.3321,
|
| 1439 |
+
"step": 23400
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 3.69,
|
| 1443 |
+
"learning_rate": 5.254901960784314e-06,
|
| 1444 |
+
"loss": 0.2998,
|
| 1445 |
+
"step": 23500
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.7,
|
| 1449 |
+
"learning_rate": 5.192156862745098e-06,
|
| 1450 |
+
"loss": 0.3337,
|
| 1451 |
+
"step": 23600
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 3.72,
|
| 1455 |
+
"learning_rate": 5.129411764705883e-06,
|
| 1456 |
+
"loss": 0.399,
|
| 1457 |
+
"step": 23700
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 3.73,
|
| 1461 |
+
"learning_rate": 5.0666666666666676e-06,
|
| 1462 |
+
"loss": 0.2966,
|
| 1463 |
+
"step": 23800
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 3.75,
|
| 1467 |
+
"learning_rate": 5.0039215686274515e-06,
|
| 1468 |
+
"loss": 0.307,
|
| 1469 |
+
"step": 23900
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 3.76,
|
| 1473 |
+
"learning_rate": 4.941176470588236e-06,
|
| 1474 |
+
"loss": 0.2964,
|
| 1475 |
+
"step": 24000
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 3.78,
|
| 1479 |
+
"learning_rate": 4.87843137254902e-06,
|
| 1480 |
+
"loss": 0.3022,
|
| 1481 |
+
"step": 24100
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 3.8,
|
| 1485 |
+
"learning_rate": 4.815686274509804e-06,
|
| 1486 |
+
"loss": 0.2502,
|
| 1487 |
+
"step": 24200
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 3.81,
|
| 1491 |
+
"learning_rate": 4.752941176470589e-06,
|
| 1492 |
+
"loss": 0.2361,
|
| 1493 |
+
"step": 24300
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 3.83,
|
| 1497 |
+
"learning_rate": 4.690196078431373e-06,
|
| 1498 |
+
"loss": 0.2654,
|
| 1499 |
+
"step": 24400
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 3.84,
|
| 1503 |
+
"learning_rate": 4.627450980392157e-06,
|
| 1504 |
+
"loss": 0.2892,
|
| 1505 |
+
"step": 24500
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 3.86,
|
| 1509 |
+
"learning_rate": 4.564705882352941e-06,
|
| 1510 |
+
"loss": 0.3175,
|
| 1511 |
+
"step": 24600
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 3.87,
|
| 1515 |
+
"learning_rate": 4.501960784313726e-06,
|
| 1516 |
+
"loss": 0.3644,
|
| 1517 |
+
"step": 24700
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 3.89,
|
| 1521 |
+
"learning_rate": 4.43921568627451e-06,
|
| 1522 |
+
"loss": 0.2597,
|
| 1523 |
+
"step": 24800
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 3.91,
|
| 1527 |
+
"learning_rate": 4.376470588235294e-06,
|
| 1528 |
+
"loss": 0.3063,
|
| 1529 |
+
"step": 24900
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 3.92,
|
| 1533 |
+
"learning_rate": 4.313725490196079e-06,
|
| 1534 |
+
"loss": 0.3096,
|
| 1535 |
+
"step": 25000
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 3.94,
|
| 1539 |
+
"learning_rate": 4.250980392156863e-06,
|
| 1540 |
+
"loss": 0.3088,
|
| 1541 |
+
"step": 25100
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 3.95,
|
| 1545 |
+
"learning_rate": 4.188235294117647e-06,
|
| 1546 |
+
"loss": 0.3462,
|
| 1547 |
+
"step": 25200
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 3.97,
|
| 1551 |
+
"learning_rate": 4.125490196078432e-06,
|
| 1552 |
+
"loss": 0.2733,
|
| 1553 |
+
"step": 25300
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 3.98,
|
| 1557 |
+
"learning_rate": 4.062745098039216e-06,
|
| 1558 |
+
"loss": 0.2905,
|
| 1559 |
+
"step": 25400
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 4.0,
|
| 1563 |
+
"learning_rate": 4.000000000000001e-06,
|
| 1564 |
+
"loss": 0.3224,
|
| 1565 |
+
"step": 25500
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 4.0,
|
| 1569 |
+
"eval_accuracy": 0.9938888888888889,
|
| 1570 |
+
"eval_loss": 0.026916783303022385,
|
| 1571 |
+
"eval_runtime": 48.3846,
|
| 1572 |
+
"eval_samples_per_second": 186.009,
|
| 1573 |
+
"eval_steps_per_second": 23.251,
|
| 1574 |
+
"step": 25500
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 4.02,
|
| 1578 |
+
"learning_rate": 3.9372549019607846e-06,
|
| 1579 |
+
"loss": 0.2968,
|
| 1580 |
+
"step": 25600
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 4.03,
|
| 1584 |
+
"learning_rate": 3.874509803921569e-06,
|
| 1585 |
+
"loss": 0.238,
|
| 1586 |
+
"step": 25700
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 4.05,
|
| 1590 |
+
"learning_rate": 3.8117647058823532e-06,
|
| 1591 |
+
"loss": 0.2669,
|
| 1592 |
+
"step": 25800
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 4.06,
|
| 1596 |
+
"learning_rate": 3.7490196078431375e-06,
|
| 1597 |
+
"loss": 0.2983,
|
| 1598 |
+
"step": 25900
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 4.08,
|
| 1602 |
+
"learning_rate": 3.6862745098039223e-06,
|
| 1603 |
+
"loss": 0.2686,
|
| 1604 |
+
"step": 26000
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 4.09,
|
| 1608 |
+
"learning_rate": 3.623529411764706e-06,
|
| 1609 |
+
"loss": 0.2907,
|
| 1610 |
+
"step": 26100
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 4.11,
|
| 1614 |
+
"learning_rate": 3.5607843137254905e-06,
|
| 1615 |
+
"loss": 0.3151,
|
| 1616 |
+
"step": 26200
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 4.13,
|
| 1620 |
+
"learning_rate": 3.498039215686275e-06,
|
| 1621 |
+
"loss": 0.2993,
|
| 1622 |
+
"step": 26300
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 4.14,
|
| 1626 |
+
"learning_rate": 3.4352941176470587e-06,
|
| 1627 |
+
"loss": 0.2732,
|
| 1628 |
+
"step": 26400
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 4.16,
|
| 1632 |
+
"learning_rate": 3.3725490196078435e-06,
|
| 1633 |
+
"loss": 0.2435,
|
| 1634 |
+
"step": 26500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 4.17,
|
| 1638 |
+
"learning_rate": 3.309803921568628e-06,
|
| 1639 |
+
"loss": 0.3121,
|
| 1640 |
+
"step": 26600
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 4.19,
|
| 1644 |
+
"learning_rate": 3.247058823529412e-06,
|
| 1645 |
+
"loss": 0.2927,
|
| 1646 |
+
"step": 26700
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 4.2,
|
| 1650 |
+
"learning_rate": 3.184313725490196e-06,
|
| 1651 |
+
"loss": 0.3005,
|
| 1652 |
+
"step": 26800
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 4.22,
|
| 1656 |
+
"learning_rate": 3.1215686274509804e-06,
|
| 1657 |
+
"loss": 0.286,
|
| 1658 |
+
"step": 26900
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 4.24,
|
| 1662 |
+
"learning_rate": 3.058823529411765e-06,
|
| 1663 |
+
"loss": 0.2461,
|
| 1664 |
+
"step": 27000
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 4.25,
|
| 1668 |
+
"learning_rate": 2.9960784313725494e-06,
|
| 1669 |
+
"loss": 0.2951,
|
| 1670 |
+
"step": 27100
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 4.27,
|
| 1674 |
+
"learning_rate": 2.9333333333333338e-06,
|
| 1675 |
+
"loss": 0.2699,
|
| 1676 |
+
"step": 27200
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 4.28,
|
| 1680 |
+
"learning_rate": 2.8705882352941177e-06,
|
| 1681 |
+
"loss": 0.3023,
|
| 1682 |
+
"step": 27300
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 4.3,
|
| 1686 |
+
"learning_rate": 2.807843137254902e-06,
|
| 1687 |
+
"loss": 0.2508,
|
| 1688 |
+
"step": 27400
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 4.31,
|
| 1692 |
+
"learning_rate": 2.7450980392156867e-06,
|
| 1693 |
+
"loss": 0.2879,
|
| 1694 |
+
"step": 27500
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 4.33,
|
| 1698 |
+
"learning_rate": 2.682352941176471e-06,
|
| 1699 |
+
"loss": 0.3172,
|
| 1700 |
+
"step": 27600
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 4.35,
|
| 1704 |
+
"learning_rate": 2.619607843137255e-06,
|
| 1705 |
+
"loss": 0.3033,
|
| 1706 |
+
"step": 27700
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 4.36,
|
| 1710 |
+
"learning_rate": 2.5568627450980393e-06,
|
| 1711 |
+
"loss": 0.3222,
|
| 1712 |
+
"step": 27800
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 4.38,
|
| 1716 |
+
"learning_rate": 2.4941176470588236e-06,
|
| 1717 |
+
"loss": 0.373,
|
| 1718 |
+
"step": 27900
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 4.39,
|
| 1722 |
+
"learning_rate": 2.431372549019608e-06,
|
| 1723 |
+
"loss": 0.3396,
|
| 1724 |
+
"step": 28000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 4.41,
|
| 1728 |
+
"learning_rate": 2.3686274509803927e-06,
|
| 1729 |
+
"loss": 0.2794,
|
| 1730 |
+
"step": 28100
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 4.42,
|
| 1734 |
+
"learning_rate": 2.3058823529411766e-06,
|
| 1735 |
+
"loss": 0.3277,
|
| 1736 |
+
"step": 28200
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 4.44,
|
| 1740 |
+
"learning_rate": 2.243137254901961e-06,
|
| 1741 |
+
"loss": 0.3244,
|
| 1742 |
+
"step": 28300
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 4.45,
|
| 1746 |
+
"learning_rate": 2.1803921568627452e-06,
|
| 1747 |
+
"loss": 0.2688,
|
| 1748 |
+
"step": 28400
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 4.47,
|
| 1752 |
+
"learning_rate": 2.1176470588235296e-06,
|
| 1753 |
+
"loss": 0.2043,
|
| 1754 |
+
"step": 28500
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 4.49,
|
| 1758 |
+
"learning_rate": 2.054901960784314e-06,
|
| 1759 |
+
"loss": 0.3008,
|
| 1760 |
+
"step": 28600
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 4.5,
|
| 1764 |
+
"learning_rate": 1.9921568627450982e-06,
|
| 1765 |
+
"loss": 0.281,
|
| 1766 |
+
"step": 28700
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 4.52,
|
| 1770 |
+
"learning_rate": 1.9294117647058825e-06,
|
| 1771 |
+
"loss": 0.32,
|
| 1772 |
+
"step": 28800
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 4.53,
|
| 1776 |
+
"learning_rate": 1.8666666666666669e-06,
|
| 1777 |
+
"loss": 0.2855,
|
| 1778 |
+
"step": 28900
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 4.55,
|
| 1782 |
+
"learning_rate": 1.8039215686274512e-06,
|
| 1783 |
+
"loss": 0.2691,
|
| 1784 |
+
"step": 29000
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 4.56,
|
| 1788 |
+
"learning_rate": 1.7411764705882353e-06,
|
| 1789 |
+
"loss": 0.2638,
|
| 1790 |
+
"step": 29100
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 4.58,
|
| 1794 |
+
"learning_rate": 1.6784313725490198e-06,
|
| 1795 |
+
"loss": 0.2563,
|
| 1796 |
+
"step": 29200
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 4.6,
|
| 1800 |
+
"learning_rate": 1.615686274509804e-06,
|
| 1801 |
+
"loss": 0.3033,
|
| 1802 |
+
"step": 29300
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 4.61,
|
| 1806 |
+
"learning_rate": 1.5529411764705885e-06,
|
| 1807 |
+
"loss": 0.3405,
|
| 1808 |
+
"step": 29400
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 4.63,
|
| 1812 |
+
"learning_rate": 1.4901960784313726e-06,
|
| 1813 |
+
"loss": 0.2761,
|
| 1814 |
+
"step": 29500
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 4.64,
|
| 1818 |
+
"learning_rate": 1.427450980392157e-06,
|
| 1819 |
+
"loss": 0.2985,
|
| 1820 |
+
"step": 29600
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 4.66,
|
| 1824 |
+
"learning_rate": 1.3647058823529413e-06,
|
| 1825 |
+
"loss": 0.2562,
|
| 1826 |
+
"step": 29700
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 4.67,
|
| 1830 |
+
"learning_rate": 1.3019607843137256e-06,
|
| 1831 |
+
"loss": 0.3165,
|
| 1832 |
+
"step": 29800
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 4.69,
|
| 1836 |
+
"learning_rate": 1.23921568627451e-06,
|
| 1837 |
+
"loss": 0.3435,
|
| 1838 |
+
"step": 29900
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 4.71,
|
| 1842 |
+
"learning_rate": 1.1764705882352942e-06,
|
| 1843 |
+
"loss": 0.2907,
|
| 1844 |
+
"step": 30000
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 4.72,
|
| 1848 |
+
"learning_rate": 1.1137254901960786e-06,
|
| 1849 |
+
"loss": 0.2614,
|
| 1850 |
+
"step": 30100
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 4.74,
|
| 1854 |
+
"learning_rate": 1.0509803921568629e-06,
|
| 1855 |
+
"loss": 0.2767,
|
| 1856 |
+
"step": 30200
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 4.75,
|
| 1860 |
+
"learning_rate": 9.88235294117647e-07,
|
| 1861 |
+
"loss": 0.2349,
|
| 1862 |
+
"step": 30300
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 4.77,
|
| 1866 |
+
"learning_rate": 9.254901960784314e-07,
|
| 1867 |
+
"loss": 0.3014,
|
| 1868 |
+
"step": 30400
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 4.78,
|
| 1872 |
+
"learning_rate": 8.627450980392157e-07,
|
| 1873 |
+
"loss": 0.284,
|
| 1874 |
+
"step": 30500
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 4.8,
|
| 1878 |
+
"learning_rate": 8.000000000000001e-07,
|
| 1879 |
+
"loss": 0.2762,
|
| 1880 |
+
"step": 30600
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 4.82,
|
| 1884 |
+
"learning_rate": 7.372549019607844e-07,
|
| 1885 |
+
"loss": 0.2582,
|
| 1886 |
+
"step": 30700
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 4.83,
|
| 1890 |
+
"learning_rate": 6.745098039215686e-07,
|
| 1891 |
+
"loss": 0.3047,
|
| 1892 |
+
"step": 30800
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 4.85,
|
| 1896 |
+
"learning_rate": 6.11764705882353e-07,
|
| 1897 |
+
"loss": 0.3008,
|
| 1898 |
+
"step": 30900
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 4.86,
|
| 1902 |
+
"learning_rate": 5.490196078431373e-07,
|
| 1903 |
+
"loss": 0.2766,
|
| 1904 |
+
"step": 31000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 4.88,
|
| 1908 |
+
"learning_rate": 4.862745098039216e-07,
|
| 1909 |
+
"loss": 0.3692,
|
| 1910 |
+
"step": 31100
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 4.89,
|
| 1914 |
+
"learning_rate": 4.235294117647059e-07,
|
| 1915 |
+
"loss": 0.2916,
|
| 1916 |
+
"step": 31200
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 4.91,
|
| 1920 |
+
"learning_rate": 3.6078431372549024e-07,
|
| 1921 |
+
"loss": 0.317,
|
| 1922 |
+
"step": 31300
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 4.93,
|
| 1926 |
+
"learning_rate": 2.9803921568627456e-07,
|
| 1927 |
+
"loss": 0.267,
|
| 1928 |
+
"step": 31400
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 4.94,
|
| 1932 |
+
"learning_rate": 2.3529411764705883e-07,
|
| 1933 |
+
"loss": 0.2752,
|
| 1934 |
+
"step": 31500
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 4.96,
|
| 1938 |
+
"learning_rate": 1.7254901960784313e-07,
|
| 1939 |
+
"loss": 0.2656,
|
| 1940 |
+
"step": 31600
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 4.97,
|
| 1944 |
+
"learning_rate": 1.0980392156862746e-07,
|
| 1945 |
+
"loss": 0.2457,
|
| 1946 |
+
"step": 31700
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 4.99,
|
| 1950 |
+
"learning_rate": 4.705882352941176e-08,
|
| 1951 |
+
"loss": 0.2846,
|
| 1952 |
+
"step": 31800
|
| 1953 |
+
},
|
| 1954 |
{
|
| 1955 |
"epoch": 5.0,
|
| 1956 |
+
"eval_accuracy": 0.9948888888888889,
|
| 1957 |
+
"eval_loss": 0.02364550158381462,
|
| 1958 |
+
"eval_runtime": 47.8839,
|
| 1959 |
+
"eval_samples_per_second": 187.955,
|
| 1960 |
+
"eval_steps_per_second": 23.494,
|
| 1961 |
+
"step": 31875
|
| 1962 |
},
|
| 1963 |
{
|
| 1964 |
"epoch": 5.0,
|
| 1965 |
+
"step": 31875,
|
| 1966 |
+
"total_flos": 1.976187420555264e+19,
|
| 1967 |
+
"train_loss": 0.3736599187514361,
|
| 1968 |
+
"train_runtime": 3816.438,
|
| 1969 |
+
"train_samples_per_second": 66.816,
|
| 1970 |
+
"train_steps_per_second": 8.352
|
| 1971 |
}
|
| 1972 |
],
|
| 1973 |
+
"max_steps": 31875,
|
| 1974 |
"num_train_epochs": 5,
|
| 1975 |
+
"total_flos": 1.976187420555264e+19,
|
| 1976 |
"trial_name": null,
|
| 1977 |
"trial_params": null
|
| 1978 |
}
|