Commit ·
17176b4
1
Parent(s): b5ecb1e
MRI Classification
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.9827025893699549,
|
| 4 |
+
"eval_loss": 0.045277200639247894,
|
| 5 |
+
"eval_runtime": 109.1502,
|
| 6 |
+
"eval_samples_per_second": 87.393,
|
| 7 |
+
"eval_steps_per_second": 10.93,
|
| 8 |
+
"total_flos": 1.6754388090591707e+19,
|
| 9 |
+
"train_loss": 0.017405257951027365,
|
| 10 |
+
"train_runtime": 6361.011,
|
| 11 |
+
"train_samples_per_second": 33.99,
|
| 12 |
+
"train_steps_per_second": 1.063
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.9827025893699549,
|
| 4 |
+
"eval_loss": 0.045277200639247894,
|
| 5 |
+
"eval_runtime": 109.1502,
|
| 6 |
+
"eval_samples_per_second": 87.393,
|
| 7 |
+
"eval_steps_per_second": 10.93
|
| 8 |
+
}
|
runs/Jun18_01-39-27_96a527910044/events.out.tfevents.1655523207.96a527910044.81.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c3e665c42be256dacd68270f6ff498d4bc659114074bf8f8480c13eff807efd
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"total_flos": 1.6754388090591707e+19,
|
| 4 |
+
"train_loss": 0.017405257951027365,
|
| 5 |
+
"train_runtime": 6361.011,
|
| 6 |
+
"train_samples_per_second": 33.99,
|
| 7 |
+
"train_steps_per_second": 1.063
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,952 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.045277200639247894,
|
| 3 |
+
"best_model_checkpoint": "./vit-base-mri/checkpoint-1500",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"global_step": 6760,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 2.977810650887574e-05,
|
| 13 |
+
"loss": 0.099,
|
| 14 |
+
"step": 50
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.06,
|
| 18 |
+
"learning_rate": 2.9556213017751482e-05,
|
| 19 |
+
"loss": 0.0913,
|
| 20 |
+
"step": 100
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.09,
|
| 24 |
+
"learning_rate": 2.933431952662722e-05,
|
| 25 |
+
"loss": 0.0751,
|
| 26 |
+
"step": 150
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.12,
|
| 30 |
+
"learning_rate": 2.911242603550296e-05,
|
| 31 |
+
"loss": 0.0663,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.15,
|
| 36 |
+
"learning_rate": 2.88905325443787e-05,
|
| 37 |
+
"loss": 0.0773,
|
| 38 |
+
"step": 250
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.18,
|
| 42 |
+
"learning_rate": 2.8668639053254437e-05,
|
| 43 |
+
"loss": 0.0628,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.21,
|
| 48 |
+
"learning_rate": 2.844674556213018e-05,
|
| 49 |
+
"loss": 0.0516,
|
| 50 |
+
"step": 350
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.24,
|
| 54 |
+
"learning_rate": 2.8224852071005918e-05,
|
| 55 |
+
"loss": 0.0537,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.27,
|
| 60 |
+
"learning_rate": 2.800295857988166e-05,
|
| 61 |
+
"loss": 0.0498,
|
| 62 |
+
"step": 450
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.3,
|
| 66 |
+
"learning_rate": 2.7781065088757396e-05,
|
| 67 |
+
"loss": 0.04,
|
| 68 |
+
"step": 500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.3,
|
| 72 |
+
"eval_accuracy": 0.9689694936576161,
|
| 73 |
+
"eval_loss": 0.08280089497566223,
|
| 74 |
+
"eval_runtime": 109.6629,
|
| 75 |
+
"eval_samples_per_second": 86.985,
|
| 76 |
+
"eval_steps_per_second": 10.879,
|
| 77 |
+
"step": 500
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.33,
|
| 81 |
+
"learning_rate": 2.7559171597633135e-05,
|
| 82 |
+
"loss": 0.0462,
|
| 83 |
+
"step": 550
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.36,
|
| 87 |
+
"learning_rate": 2.7337278106508877e-05,
|
| 88 |
+
"loss": 0.0382,
|
| 89 |
+
"step": 600
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.38,
|
| 93 |
+
"learning_rate": 2.7115384615384616e-05,
|
| 94 |
+
"loss": 0.0351,
|
| 95 |
+
"step": 650
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.41,
|
| 99 |
+
"learning_rate": 2.6893491124260358e-05,
|
| 100 |
+
"loss": 0.0274,
|
| 101 |
+
"step": 700
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.44,
|
| 105 |
+
"learning_rate": 2.6671597633136097e-05,
|
| 106 |
+
"loss": 0.029,
|
| 107 |
+
"step": 750
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.47,
|
| 111 |
+
"learning_rate": 2.6449704142011832e-05,
|
| 112 |
+
"loss": 0.0599,
|
| 113 |
+
"step": 800
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.5,
|
| 117 |
+
"learning_rate": 2.6227810650887575e-05,
|
| 118 |
+
"loss": 0.0621,
|
| 119 |
+
"step": 850
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.53,
|
| 123 |
+
"learning_rate": 2.6005917159763313e-05,
|
| 124 |
+
"loss": 0.0807,
|
| 125 |
+
"step": 900
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.56,
|
| 129 |
+
"learning_rate": 2.5784023668639056e-05,
|
| 130 |
+
"loss": 0.0535,
|
| 131 |
+
"step": 950
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.59,
|
| 135 |
+
"learning_rate": 2.5562130177514795e-05,
|
| 136 |
+
"loss": 0.0765,
|
| 137 |
+
"step": 1000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.59,
|
| 141 |
+
"eval_accuracy": 0.9750497955760562,
|
| 142 |
+
"eval_loss": 0.0623435452580452,
|
| 143 |
+
"eval_runtime": 109.6354,
|
| 144 |
+
"eval_samples_per_second": 87.007,
|
| 145 |
+
"eval_steps_per_second": 10.882,
|
| 146 |
+
"step": 1000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.62,
|
| 150 |
+
"learning_rate": 2.5340236686390533e-05,
|
| 151 |
+
"loss": 0.0574,
|
| 152 |
+
"step": 1050
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.65,
|
| 156 |
+
"learning_rate": 2.5118343195266272e-05,
|
| 157 |
+
"loss": 0.0608,
|
| 158 |
+
"step": 1100
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.68,
|
| 162 |
+
"learning_rate": 2.489644970414201e-05,
|
| 163 |
+
"loss": 0.0546,
|
| 164 |
+
"step": 1150
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.71,
|
| 168 |
+
"learning_rate": 2.467455621301775e-05,
|
| 169 |
+
"loss": 0.0509,
|
| 170 |
+
"step": 1200
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.74,
|
| 174 |
+
"learning_rate": 2.4452662721893492e-05,
|
| 175 |
+
"loss": 0.0509,
|
| 176 |
+
"step": 1250
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.77,
|
| 180 |
+
"learning_rate": 2.423076923076923e-05,
|
| 181 |
+
"loss": 0.0471,
|
| 182 |
+
"step": 1300
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.8,
|
| 186 |
+
"learning_rate": 2.4008875739644973e-05,
|
| 187 |
+
"loss": 0.0326,
|
| 188 |
+
"step": 1350
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.83,
|
| 192 |
+
"learning_rate": 2.3786982248520712e-05,
|
| 193 |
+
"loss": 0.0568,
|
| 194 |
+
"step": 1400
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.86,
|
| 198 |
+
"learning_rate": 2.3565088757396448e-05,
|
| 199 |
+
"loss": 0.0289,
|
| 200 |
+
"step": 1450
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.89,
|
| 204 |
+
"learning_rate": 2.334319526627219e-05,
|
| 205 |
+
"loss": 0.0479,
|
| 206 |
+
"step": 1500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.89,
|
| 210 |
+
"eval_accuracy": 0.9827025893699549,
|
| 211 |
+
"eval_loss": 0.045277200639247894,
|
| 212 |
+
"eval_runtime": 110.2271,
|
| 213 |
+
"eval_samples_per_second": 86.54,
|
| 214 |
+
"eval_steps_per_second": 10.823,
|
| 215 |
+
"step": 1500
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.92,
|
| 219 |
+
"learning_rate": 2.312130177514793e-05,
|
| 220 |
+
"loss": 0.055,
|
| 221 |
+
"step": 1550
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.95,
|
| 225 |
+
"learning_rate": 2.289940828402367e-05,
|
| 226 |
+
"loss": 0.0402,
|
| 227 |
+
"step": 1600
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.98,
|
| 231 |
+
"learning_rate": 2.267751479289941e-05,
|
| 232 |
+
"loss": 0.0546,
|
| 233 |
+
"step": 1650
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.01,
|
| 237 |
+
"learning_rate": 2.245562130177515e-05,
|
| 238 |
+
"loss": 0.0377,
|
| 239 |
+
"step": 1700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.04,
|
| 243 |
+
"learning_rate": 2.2238165680473372e-05,
|
| 244 |
+
"loss": 0.0164,
|
| 245 |
+
"step": 1750
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.07,
|
| 249 |
+
"learning_rate": 2.2016272189349114e-05,
|
| 250 |
+
"loss": 0.0081,
|
| 251 |
+
"step": 1800
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.09,
|
| 255 |
+
"learning_rate": 2.1794378698224853e-05,
|
| 256 |
+
"loss": 0.0097,
|
| 257 |
+
"step": 1850
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.12,
|
| 261 |
+
"learning_rate": 2.1572485207100592e-05,
|
| 262 |
+
"loss": 0.0139,
|
| 263 |
+
"step": 1900
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.15,
|
| 267 |
+
"learning_rate": 2.135059171597633e-05,
|
| 268 |
+
"loss": 0.0075,
|
| 269 |
+
"step": 1950
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.18,
|
| 273 |
+
"learning_rate": 2.112869822485207e-05,
|
| 274 |
+
"loss": 0.0199,
|
| 275 |
+
"step": 2000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.18,
|
| 279 |
+
"eval_accuracy": 0.985742740329175,
|
| 280 |
+
"eval_loss": 0.05243392661213875,
|
| 281 |
+
"eval_runtime": 109.2241,
|
| 282 |
+
"eval_samples_per_second": 87.334,
|
| 283 |
+
"eval_steps_per_second": 10.923,
|
| 284 |
+
"step": 2000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.21,
|
| 288 |
+
"learning_rate": 2.0906804733727812e-05,
|
| 289 |
+
"loss": 0.0079,
|
| 290 |
+
"step": 2050
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.24,
|
| 294 |
+
"learning_rate": 2.068491124260355e-05,
|
| 295 |
+
"loss": 0.0145,
|
| 296 |
+
"step": 2100
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.27,
|
| 300 |
+
"learning_rate": 2.0463017751479293e-05,
|
| 301 |
+
"loss": 0.0166,
|
| 302 |
+
"step": 2150
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.3,
|
| 306 |
+
"learning_rate": 2.0241124260355032e-05,
|
| 307 |
+
"loss": 0.0101,
|
| 308 |
+
"step": 2200
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.33,
|
| 312 |
+
"learning_rate": 2.0019230769230767e-05,
|
| 313 |
+
"loss": 0.0079,
|
| 314 |
+
"step": 2250
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.36,
|
| 318 |
+
"learning_rate": 1.979733727810651e-05,
|
| 319 |
+
"loss": 0.0075,
|
| 320 |
+
"step": 2300
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 1.39,
|
| 324 |
+
"learning_rate": 1.9575443786982248e-05,
|
| 325 |
+
"loss": 0.0176,
|
| 326 |
+
"step": 2350
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.42,
|
| 330 |
+
"learning_rate": 1.935355029585799e-05,
|
| 331 |
+
"loss": 0.0107,
|
| 332 |
+
"step": 2400
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.45,
|
| 336 |
+
"learning_rate": 1.913165680473373e-05,
|
| 337 |
+
"loss": 0.0213,
|
| 338 |
+
"step": 2450
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.48,
|
| 342 |
+
"learning_rate": 1.8909763313609468e-05,
|
| 343 |
+
"loss": 0.0114,
|
| 344 |
+
"step": 2500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.48,
|
| 348 |
+
"eval_accuracy": 0.9860572387042666,
|
| 349 |
+
"eval_loss": 0.048441529273986816,
|
| 350 |
+
"eval_runtime": 108.7938,
|
| 351 |
+
"eval_samples_per_second": 87.68,
|
| 352 |
+
"eval_steps_per_second": 10.966,
|
| 353 |
+
"step": 2500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.51,
|
| 357 |
+
"learning_rate": 1.8687869822485207e-05,
|
| 358 |
+
"loss": 0.0158,
|
| 359 |
+
"step": 2550
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.54,
|
| 363 |
+
"learning_rate": 1.8465976331360946e-05,
|
| 364 |
+
"loss": 0.0105,
|
| 365 |
+
"step": 2600
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.57,
|
| 369 |
+
"learning_rate": 1.8244082840236688e-05,
|
| 370 |
+
"loss": 0.0078,
|
| 371 |
+
"step": 2650
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.6,
|
| 375 |
+
"learning_rate": 1.8022189349112427e-05,
|
| 376 |
+
"loss": 0.0131,
|
| 377 |
+
"step": 2700
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.63,
|
| 381 |
+
"learning_rate": 1.7800295857988166e-05,
|
| 382 |
+
"loss": 0.0063,
|
| 383 |
+
"step": 2750
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.66,
|
| 387 |
+
"learning_rate": 1.7578402366863908e-05,
|
| 388 |
+
"loss": 0.0064,
|
| 389 |
+
"step": 2800
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.69,
|
| 393 |
+
"learning_rate": 1.7356508875739643e-05,
|
| 394 |
+
"loss": 0.0077,
|
| 395 |
+
"step": 2850
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.72,
|
| 399 |
+
"learning_rate": 1.7134615384615386e-05,
|
| 400 |
+
"loss": 0.0063,
|
| 401 |
+
"step": 2900
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.75,
|
| 405 |
+
"learning_rate": 1.6912721893491125e-05,
|
| 406 |
+
"loss": 0.0098,
|
| 407 |
+
"step": 2950
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.78,
|
| 411 |
+
"learning_rate": 1.6690828402366863e-05,
|
| 412 |
+
"loss": 0.008,
|
| 413 |
+
"step": 3000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.78,
|
| 417 |
+
"eval_accuracy": 0.9852185763706888,
|
| 418 |
+
"eval_loss": 0.056569721549749374,
|
| 419 |
+
"eval_runtime": 108.3934,
|
| 420 |
+
"eval_samples_per_second": 88.004,
|
| 421 |
+
"eval_steps_per_second": 11.006,
|
| 422 |
+
"step": 3000
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.8,
|
| 426 |
+
"learning_rate": 1.6468934911242606e-05,
|
| 427 |
+
"loss": 0.0133,
|
| 428 |
+
"step": 3050
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.83,
|
| 432 |
+
"learning_rate": 1.6247041420118344e-05,
|
| 433 |
+
"loss": 0.0134,
|
| 434 |
+
"step": 3100
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.86,
|
| 438 |
+
"learning_rate": 1.6025147928994083e-05,
|
| 439 |
+
"loss": 0.0067,
|
| 440 |
+
"step": 3150
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.89,
|
| 444 |
+
"learning_rate": 1.5803254437869822e-05,
|
| 445 |
+
"loss": 0.009,
|
| 446 |
+
"step": 3200
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.92,
|
| 450 |
+
"learning_rate": 1.558136094674556e-05,
|
| 451 |
+
"loss": 0.0132,
|
| 452 |
+
"step": 3250
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.95,
|
| 456 |
+
"learning_rate": 1.5359467455621303e-05,
|
| 457 |
+
"loss": 0.0131,
|
| 458 |
+
"step": 3300
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.98,
|
| 462 |
+
"learning_rate": 1.5137573964497042e-05,
|
| 463 |
+
"loss": 0.0071,
|
| 464 |
+
"step": 3350
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.01,
|
| 468 |
+
"learning_rate": 1.4915680473372781e-05,
|
| 469 |
+
"loss": 0.0115,
|
| 470 |
+
"step": 3400
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.04,
|
| 474 |
+
"learning_rate": 1.4693786982248521e-05,
|
| 475 |
+
"loss": 0.0017,
|
| 476 |
+
"step": 3450
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.07,
|
| 480 |
+
"learning_rate": 1.4471893491124262e-05,
|
| 481 |
+
"loss": 0.0051,
|
| 482 |
+
"step": 3500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.07,
|
| 486 |
+
"eval_accuracy": 0.9874200649963308,
|
| 487 |
+
"eval_loss": 0.051312096416950226,
|
| 488 |
+
"eval_runtime": 108.8914,
|
| 489 |
+
"eval_samples_per_second": 87.601,
|
| 490 |
+
"eval_steps_per_second": 10.956,
|
| 491 |
+
"step": 3500
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 2.1,
|
| 495 |
+
"learning_rate": 1.4249999999999999e-05,
|
| 496 |
+
"loss": 0.0023,
|
| 497 |
+
"step": 3550
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 2.13,
|
| 501 |
+
"learning_rate": 1.402810650887574e-05,
|
| 502 |
+
"loss": 0.0037,
|
| 503 |
+
"step": 3600
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 2.16,
|
| 507 |
+
"learning_rate": 1.380621301775148e-05,
|
| 508 |
+
"loss": 0.001,
|
| 509 |
+
"step": 3650
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 2.19,
|
| 513 |
+
"learning_rate": 1.3584319526627219e-05,
|
| 514 |
+
"loss": 0.0037,
|
| 515 |
+
"step": 3700
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 2.22,
|
| 519 |
+
"learning_rate": 1.336242603550296e-05,
|
| 520 |
+
"loss": 0.0026,
|
| 521 |
+
"step": 3750
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 2.25,
|
| 525 |
+
"learning_rate": 1.3140532544378698e-05,
|
| 526 |
+
"loss": 0.0012,
|
| 527 |
+
"step": 3800
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 2.28,
|
| 531 |
+
"learning_rate": 1.2918639053254437e-05,
|
| 532 |
+
"loss": 0.0022,
|
| 533 |
+
"step": 3850
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 2.31,
|
| 537 |
+
"learning_rate": 1.2696745562130178e-05,
|
| 538 |
+
"loss": 0.0006,
|
| 539 |
+
"step": 3900
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 2.34,
|
| 543 |
+
"learning_rate": 1.2474852071005918e-05,
|
| 544 |
+
"loss": 0.0005,
|
| 545 |
+
"step": 3950
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 2.37,
|
| 549 |
+
"learning_rate": 1.2252958579881657e-05,
|
| 550 |
+
"loss": 0.0008,
|
| 551 |
+
"step": 4000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 2.37,
|
| 555 |
+
"eval_accuracy": 0.9874200649963308,
|
| 556 |
+
"eval_loss": 0.061718400567770004,
|
| 557 |
+
"eval_runtime": 108.6858,
|
| 558 |
+
"eval_samples_per_second": 87.767,
|
| 559 |
+
"eval_steps_per_second": 10.977,
|
| 560 |
+
"step": 4000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.4,
|
| 564 |
+
"learning_rate": 1.2031065088757396e-05,
|
| 565 |
+
"loss": 0.005,
|
| 566 |
+
"step": 4050
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.43,
|
| 570 |
+
"learning_rate": 1.1809171597633137e-05,
|
| 571 |
+
"loss": 0.0005,
|
| 572 |
+
"step": 4100
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.46,
|
| 576 |
+
"learning_rate": 1.1587278106508877e-05,
|
| 577 |
+
"loss": 0.0019,
|
| 578 |
+
"step": 4150
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.49,
|
| 582 |
+
"learning_rate": 1.1365384615384616e-05,
|
| 583 |
+
"loss": 0.0015,
|
| 584 |
+
"step": 4200
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.51,
|
| 588 |
+
"learning_rate": 1.1143491124260355e-05,
|
| 589 |
+
"loss": 0.0044,
|
| 590 |
+
"step": 4250
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.54,
|
| 594 |
+
"learning_rate": 1.0921597633136095e-05,
|
| 595 |
+
"loss": 0.0005,
|
| 596 |
+
"step": 4300
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.57,
|
| 600 |
+
"learning_rate": 1.0699704142011834e-05,
|
| 601 |
+
"loss": 0.0036,
|
| 602 |
+
"step": 4350
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.6,
|
| 606 |
+
"learning_rate": 1.0477810650887575e-05,
|
| 607 |
+
"loss": 0.0009,
|
| 608 |
+
"step": 4400
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.63,
|
| 612 |
+
"learning_rate": 1.0255917159763315e-05,
|
| 613 |
+
"loss": 0.0004,
|
| 614 |
+
"step": 4450
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.66,
|
| 618 |
+
"learning_rate": 1.0034023668639052e-05,
|
| 619 |
+
"loss": 0.0021,
|
| 620 |
+
"step": 4500
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.66,
|
| 624 |
+
"eval_accuracy": 0.9870007338295419,
|
| 625 |
+
"eval_loss": 0.06636156141757965,
|
| 626 |
+
"eval_runtime": 108.716,
|
| 627 |
+
"eval_samples_per_second": 87.742,
|
| 628 |
+
"eval_steps_per_second": 10.974,
|
| 629 |
+
"step": 4500
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 2.69,
|
| 633 |
+
"learning_rate": 9.812130177514793e-06,
|
| 634 |
+
"loss": 0.0022,
|
| 635 |
+
"step": 4550
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 2.72,
|
| 639 |
+
"learning_rate": 9.590236686390533e-06,
|
| 640 |
+
"loss": 0.0043,
|
| 641 |
+
"step": 4600
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 2.75,
|
| 645 |
+
"learning_rate": 9.368343195266272e-06,
|
| 646 |
+
"loss": 0.0047,
|
| 647 |
+
"step": 4650
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 2.78,
|
| 651 |
+
"learning_rate": 9.146449704142013e-06,
|
| 652 |
+
"loss": 0.0008,
|
| 653 |
+
"step": 4700
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 2.81,
|
| 657 |
+
"learning_rate": 8.924556213017752e-06,
|
| 658 |
+
"loss": 0.0008,
|
| 659 |
+
"step": 4750
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 2.84,
|
| 663 |
+
"learning_rate": 8.70266272189349e-06,
|
| 664 |
+
"loss": 0.0022,
|
| 665 |
+
"step": 4800
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 2.87,
|
| 669 |
+
"learning_rate": 8.480769230769231e-06,
|
| 670 |
+
"loss": 0.0015,
|
| 671 |
+
"step": 4850
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 2.9,
|
| 675 |
+
"learning_rate": 8.258875739644972e-06,
|
| 676 |
+
"loss": 0.0004,
|
| 677 |
+
"step": 4900
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 2.93,
|
| 681 |
+
"learning_rate": 8.036982248520709e-06,
|
| 682 |
+
"loss": 0.0071,
|
| 683 |
+
"step": 4950
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.96,
|
| 687 |
+
"learning_rate": 7.81508875739645e-06,
|
| 688 |
+
"loss": 0.0005,
|
| 689 |
+
"step": 5000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 2.96,
|
| 693 |
+
"eval_accuracy": 0.9872103994129363,
|
| 694 |
+
"eval_loss": 0.06394314020872116,
|
| 695 |
+
"eval_runtime": 108.7628,
|
| 696 |
+
"eval_samples_per_second": 87.705,
|
| 697 |
+
"eval_steps_per_second": 10.969,
|
| 698 |
+
"step": 5000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.99,
|
| 702 |
+
"learning_rate": 7.59319526627219e-06,
|
| 703 |
+
"loss": 0.0006,
|
| 704 |
+
"step": 5050
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 3.02,
|
| 708 |
+
"learning_rate": 7.3713017751479295e-06,
|
| 709 |
+
"loss": 0.0004,
|
| 710 |
+
"step": 5100
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 3.05,
|
| 714 |
+
"learning_rate": 7.149408284023668e-06,
|
| 715 |
+
"loss": 0.0006,
|
| 716 |
+
"step": 5150
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 3.08,
|
| 720 |
+
"learning_rate": 6.9319526627218935e-06,
|
| 721 |
+
"loss": 0.0106,
|
| 722 |
+
"step": 5200
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 3.11,
|
| 726 |
+
"learning_rate": 6.710059171597633e-06,
|
| 727 |
+
"loss": 0.0029,
|
| 728 |
+
"step": 5250
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 3.14,
|
| 732 |
+
"learning_rate": 6.488165680473374e-06,
|
| 733 |
+
"loss": 0.001,
|
| 734 |
+
"step": 5300
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 3.17,
|
| 738 |
+
"learning_rate": 6.2662721893491125e-06,
|
| 739 |
+
"loss": 0.0097,
|
| 740 |
+
"step": 5350
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 3.2,
|
| 744 |
+
"learning_rate": 6.044378698224852e-06,
|
| 745 |
+
"loss": 0.0037,
|
| 746 |
+
"step": 5400
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 3.22,
|
| 750 |
+
"learning_rate": 5.822485207100592e-06,
|
| 751 |
+
"loss": 0.0016,
|
| 752 |
+
"step": 5450
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 3.25,
|
| 756 |
+
"learning_rate": 5.600591715976332e-06,
|
| 757 |
+
"loss": 0.001,
|
| 758 |
+
"step": 5500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 3.25,
|
| 762 |
+
"eval_accuracy": 0.987944228954817,
|
| 763 |
+
"eval_loss": 0.06436183303594589,
|
| 764 |
+
"eval_runtime": 108.9405,
|
| 765 |
+
"eval_samples_per_second": 87.562,
|
| 766 |
+
"eval_steps_per_second": 10.951,
|
| 767 |
+
"step": 5500
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 3.28,
|
| 771 |
+
"learning_rate": 5.3786982248520704e-06,
|
| 772 |
+
"loss": 0.0006,
|
| 773 |
+
"step": 5550
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 3.31,
|
| 777 |
+
"learning_rate": 5.156804733727811e-06,
|
| 778 |
+
"loss": 0.0004,
|
| 779 |
+
"step": 5600
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 3.34,
|
| 783 |
+
"learning_rate": 4.934911242603551e-06,
|
| 784 |
+
"loss": 0.0002,
|
| 785 |
+
"step": 5650
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 3.37,
|
| 789 |
+
"learning_rate": 4.7130177514792895e-06,
|
| 790 |
+
"loss": 0.0003,
|
| 791 |
+
"step": 5700
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 3.4,
|
| 795 |
+
"learning_rate": 4.49112426035503e-06,
|
| 796 |
+
"loss": 0.0021,
|
| 797 |
+
"step": 5750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.43,
|
| 801 |
+
"learning_rate": 4.269230769230769e-06,
|
| 802 |
+
"loss": 0.0003,
|
| 803 |
+
"step": 5800
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 3.46,
|
| 807 |
+
"learning_rate": 4.047337278106509e-06,
|
| 808 |
+
"loss": 0.0005,
|
| 809 |
+
"step": 5850
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.49,
|
| 813 |
+
"learning_rate": 3.825443786982248e-06,
|
| 814 |
+
"loss": 0.0003,
|
| 815 |
+
"step": 5900
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 3.52,
|
| 819 |
+
"learning_rate": 3.6035502958579884e-06,
|
| 820 |
+
"loss": 0.0003,
|
| 821 |
+
"step": 5950
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 3.55,
|
| 825 |
+
"learning_rate": 3.3816568047337276e-06,
|
| 826 |
+
"loss": 0.0004,
|
| 827 |
+
"step": 6000
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 3.55,
|
| 831 |
+
"eval_accuracy": 0.9875248977880281,
|
| 832 |
+
"eval_loss": 0.06724812090396881,
|
| 833 |
+
"eval_runtime": 108.7744,
|
| 834 |
+
"eval_samples_per_second": 87.695,
|
| 835 |
+
"eval_steps_per_second": 10.968,
|
| 836 |
+
"step": 6000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 3.58,
|
| 840 |
+
"learning_rate": 3.1597633136094673e-06,
|
| 841 |
+
"loss": 0.0018,
|
| 842 |
+
"step": 6050
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 3.61,
|
| 846 |
+
"learning_rate": 2.9378698224852074e-06,
|
| 847 |
+
"loss": 0.0003,
|
| 848 |
+
"step": 6100
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 3.64,
|
| 852 |
+
"learning_rate": 2.7159763313609467e-06,
|
| 853 |
+
"loss": 0.0016,
|
| 854 |
+
"step": 6150
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 3.67,
|
| 858 |
+
"learning_rate": 2.4940828402366864e-06,
|
| 859 |
+
"loss": 0.0009,
|
| 860 |
+
"step": 6200
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 3.7,
|
| 864 |
+
"learning_rate": 2.272189349112426e-06,
|
| 865 |
+
"loss": 0.0015,
|
| 866 |
+
"step": 6250
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 3.73,
|
| 870 |
+
"learning_rate": 2.0502958579881658e-06,
|
| 871 |
+
"loss": 0.0004,
|
| 872 |
+
"step": 6300
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 3.76,
|
| 876 |
+
"learning_rate": 1.8284023668639055e-06,
|
| 877 |
+
"loss": 0.0003,
|
| 878 |
+
"step": 6350
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 3.79,
|
| 882 |
+
"learning_rate": 1.606508875739645e-06,
|
| 883 |
+
"loss": 0.0017,
|
| 884 |
+
"step": 6400
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 3.82,
|
| 888 |
+
"learning_rate": 1.3846153846153846e-06,
|
| 889 |
+
"loss": 0.0004,
|
| 890 |
+
"step": 6450
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 3.85,
|
| 894 |
+
"learning_rate": 1.1627218934911243e-06,
|
| 895 |
+
"loss": 0.0003,
|
| 896 |
+
"step": 6500
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 3.85,
|
| 900 |
+
"eval_accuracy": 0.987944228954817,
|
| 901 |
+
"eval_loss": 0.06903128325939178,
|
| 902 |
+
"eval_runtime": 108.9231,
|
| 903 |
+
"eval_samples_per_second": 87.576,
|
| 904 |
+
"eval_steps_per_second": 10.953,
|
| 905 |
+
"step": 6500
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.88,
|
| 909 |
+
"learning_rate": 9.40828402366864e-07,
|
| 910 |
+
"loss": 0.0016,
|
| 911 |
+
"step": 6550
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.91,
|
| 915 |
+
"learning_rate": 7.189349112426036e-07,
|
| 916 |
+
"loss": 0.0002,
|
| 917 |
+
"step": 6600
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 3.93,
|
| 921 |
+
"learning_rate": 4.970414201183432e-07,
|
| 922 |
+
"loss": 0.0003,
|
| 923 |
+
"step": 6650
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.96,
|
| 927 |
+
"learning_rate": 2.751479289940828e-07,
|
| 928 |
+
"loss": 0.0016,
|
| 929 |
+
"step": 6700
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.99,
|
| 933 |
+
"learning_rate": 5.3254437869822486e-08,
|
| 934 |
+
"loss": 0.0013,
|
| 935 |
+
"step": 6750
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 4.0,
|
| 939 |
+
"step": 6760,
|
| 940 |
+
"total_flos": 1.6754388090591707e+19,
|
| 941 |
+
"train_loss": 0.017405257951027365,
|
| 942 |
+
"train_runtime": 6361.011,
|
| 943 |
+
"train_samples_per_second": 33.99,
|
| 944 |
+
"train_steps_per_second": 1.063
|
| 945 |
+
}
|
| 946 |
+
],
|
| 947 |
+
"max_steps": 6760,
|
| 948 |
+
"num_train_epochs": 4,
|
| 949 |
+
"total_flos": 1.6754388090591707e+19,
|
| 950 |
+
"trial_name": null,
|
| 951 |
+
"trial_params": null
|
| 952 |
+
}
|