End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 4.812819303609225e+18,
|
| 4 |
+
"train_loss": 1.6366877293570932,
|
| 5 |
+
"train_runtime": 3640.8494,
|
| 6 |
+
"train_samples_per_second": 53.054,
|
| 7 |
+
"train_steps_per_second": 0.414
|
| 8 |
+
}
|
runs/Jan11_09-42-41_e308505a2991/events.out.tfevents.1704972642.e308505a2991.24483.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:270421cd5d34ea94cbeace003168510107298aab65849e14656855c2eeca67f0
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 4.812819303609225e+18,
|
| 4 |
+
"train_loss": 1.6366877293570932,
|
| 5 |
+
"train_runtime": 3640.8494,
|
| 6 |
+
"train_samples_per_second": 53.054,
|
| 7 |
+
"train_steps_per_second": 0.414
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,964 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7760274575376221,
|
| 3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-food101/checkpoint-1509",
|
| 4 |
+
"epoch": 2.9985096870342773,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1509,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.02,
|
| 13 |
+
"learning_rate": 3.3112582781456956e-06,
|
| 14 |
+
"loss": 4.6663,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.04,
|
| 19 |
+
"learning_rate": 6.622516556291391e-06,
|
| 20 |
+
"loss": 4.6531,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.06,
|
| 25 |
+
"learning_rate": 9.933774834437086e-06,
|
| 26 |
+
"loss": 4.628,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.08,
|
| 31 |
+
"learning_rate": 1.3245033112582782e-05,
|
| 32 |
+
"loss": 4.613,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.1,
|
| 37 |
+
"learning_rate": 1.655629139072848e-05,
|
| 38 |
+
"loss": 4.5875,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.12,
|
| 43 |
+
"learning_rate": 1.9867549668874173e-05,
|
| 44 |
+
"loss": 4.5399,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14,
|
| 49 |
+
"learning_rate": 2.317880794701987e-05,
|
| 50 |
+
"loss": 4.5013,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.16,
|
| 55 |
+
"learning_rate": 2.6490066225165565e-05,
|
| 56 |
+
"loss": 4.42,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.18,
|
| 61 |
+
"learning_rate": 2.980132450331126e-05,
|
| 62 |
+
"loss": 4.298,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.2,
|
| 67 |
+
"learning_rate": 3.311258278145696e-05,
|
| 68 |
+
"loss": 4.1074,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.22,
|
| 73 |
+
"learning_rate": 3.6423841059602646e-05,
|
| 74 |
+
"loss": 3.9249,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.24,
|
| 79 |
+
"learning_rate": 3.9735099337748346e-05,
|
| 80 |
+
"loss": 3.6311,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.26,
|
| 85 |
+
"learning_rate": 4.304635761589404e-05,
|
| 86 |
+
"loss": 3.4282,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.28,
|
| 91 |
+
"learning_rate": 4.635761589403974e-05,
|
| 92 |
+
"loss": 3.1668,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.3,
|
| 97 |
+
"learning_rate": 4.966887417218543e-05,
|
| 98 |
+
"loss": 3.0268,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.32,
|
| 103 |
+
"learning_rate": 4.966863033873343e-05,
|
| 104 |
+
"loss": 2.7985,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.34,
|
| 109 |
+
"learning_rate": 4.9300441826215024e-05,
|
| 110 |
+
"loss": 2.6903,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.36,
|
| 115 |
+
"learning_rate": 4.8932253313696616e-05,
|
| 116 |
+
"loss": 2.4989,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.38,
|
| 121 |
+
"learning_rate": 4.856406480117821e-05,
|
| 122 |
+
"loss": 2.5033,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.4,
|
| 127 |
+
"learning_rate": 4.8195876288659794e-05,
|
| 128 |
+
"loss": 2.3169,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.42,
|
| 133 |
+
"learning_rate": 4.7827687776141386e-05,
|
| 134 |
+
"loss": 2.3053,
|
| 135 |
+
"step": 210
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.44,
|
| 139 |
+
"learning_rate": 4.745949926362298e-05,
|
| 140 |
+
"loss": 2.1974,
|
| 141 |
+
"step": 220
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.46,
|
| 145 |
+
"learning_rate": 4.709131075110457e-05,
|
| 146 |
+
"loss": 2.0775,
|
| 147 |
+
"step": 230
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.48,
|
| 151 |
+
"learning_rate": 4.6723122238586156e-05,
|
| 152 |
+
"loss": 2.0065,
|
| 153 |
+
"step": 240
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.5,
|
| 157 |
+
"learning_rate": 4.635493372606775e-05,
|
| 158 |
+
"loss": 2.0521,
|
| 159 |
+
"step": 250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.52,
|
| 163 |
+
"learning_rate": 4.598674521354934e-05,
|
| 164 |
+
"loss": 1.9464,
|
| 165 |
+
"step": 260
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.54,
|
| 169 |
+
"learning_rate": 4.561855670103093e-05,
|
| 170 |
+
"loss": 2.0498,
|
| 171 |
+
"step": 270
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.56,
|
| 175 |
+
"learning_rate": 4.525036818851252e-05,
|
| 176 |
+
"loss": 1.9544,
|
| 177 |
+
"step": 280
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.58,
|
| 181 |
+
"learning_rate": 4.488217967599411e-05,
|
| 182 |
+
"loss": 1.902,
|
| 183 |
+
"step": 290
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.6,
|
| 187 |
+
"learning_rate": 4.45139911634757e-05,
|
| 188 |
+
"loss": 1.8034,
|
| 189 |
+
"step": 300
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.62,
|
| 193 |
+
"learning_rate": 4.4145802650957294e-05,
|
| 194 |
+
"loss": 1.7483,
|
| 195 |
+
"step": 310
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.64,
|
| 199 |
+
"learning_rate": 4.3777614138438886e-05,
|
| 200 |
+
"loss": 1.8315,
|
| 201 |
+
"step": 320
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.66,
|
| 205 |
+
"learning_rate": 4.340942562592047e-05,
|
| 206 |
+
"loss": 1.7785,
|
| 207 |
+
"step": 330
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.68,
|
| 211 |
+
"learning_rate": 4.3041237113402064e-05,
|
| 212 |
+
"loss": 1.85,
|
| 213 |
+
"step": 340
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.7,
|
| 217 |
+
"learning_rate": 4.2673048600883656e-05,
|
| 218 |
+
"loss": 1.7037,
|
| 219 |
+
"step": 350
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.72,
|
| 223 |
+
"learning_rate": 4.230486008836525e-05,
|
| 224 |
+
"loss": 1.6452,
|
| 225 |
+
"step": 360
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.74,
|
| 229 |
+
"learning_rate": 4.193667157584683e-05,
|
| 230 |
+
"loss": 1.6515,
|
| 231 |
+
"step": 370
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.76,
|
| 235 |
+
"learning_rate": 4.1568483063328425e-05,
|
| 236 |
+
"loss": 1.6375,
|
| 237 |
+
"step": 380
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.77,
|
| 241 |
+
"learning_rate": 4.120029455081002e-05,
|
| 242 |
+
"loss": 1.6522,
|
| 243 |
+
"step": 390
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.79,
|
| 247 |
+
"learning_rate": 4.083210603829161e-05,
|
| 248 |
+
"loss": 1.5982,
|
| 249 |
+
"step": 400
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.81,
|
| 253 |
+
"learning_rate": 4.0463917525773195e-05,
|
| 254 |
+
"loss": 1.5984,
|
| 255 |
+
"step": 410
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.83,
|
| 259 |
+
"learning_rate": 4.009572901325479e-05,
|
| 260 |
+
"loss": 1.5961,
|
| 261 |
+
"step": 420
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.85,
|
| 265 |
+
"learning_rate": 3.972754050073638e-05,
|
| 266 |
+
"loss": 1.6263,
|
| 267 |
+
"step": 430
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.87,
|
| 271 |
+
"learning_rate": 3.935935198821797e-05,
|
| 272 |
+
"loss": 1.5302,
|
| 273 |
+
"step": 440
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.89,
|
| 277 |
+
"learning_rate": 3.899116347569956e-05,
|
| 278 |
+
"loss": 1.5545,
|
| 279 |
+
"step": 450
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.91,
|
| 283 |
+
"learning_rate": 3.862297496318115e-05,
|
| 284 |
+
"loss": 1.5918,
|
| 285 |
+
"step": 460
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.93,
|
| 289 |
+
"learning_rate": 3.825478645066274e-05,
|
| 290 |
+
"loss": 1.539,
|
| 291 |
+
"step": 470
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.95,
|
| 295 |
+
"learning_rate": 3.7886597938144333e-05,
|
| 296 |
+
"loss": 1.5363,
|
| 297 |
+
"step": 480
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.97,
|
| 301 |
+
"learning_rate": 3.7518409425625926e-05,
|
| 302 |
+
"loss": 1.5359,
|
| 303 |
+
"step": 490
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.99,
|
| 307 |
+
"learning_rate": 3.715022091310751e-05,
|
| 308 |
+
"loss": 1.4777,
|
| 309 |
+
"step": 500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.0,
|
| 313 |
+
"eval_accuracy": 0.7062395494147672,
|
| 314 |
+
"eval_loss": 1.1435766220092773,
|
| 315 |
+
"eval_runtime": 107.4159,
|
| 316 |
+
"eval_samples_per_second": 105.785,
|
| 317 |
+
"eval_steps_per_second": 3.314,
|
| 318 |
+
"step": 503
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.01,
|
| 322 |
+
"learning_rate": 3.67820324005891e-05,
|
| 323 |
+
"loss": 1.492,
|
| 324 |
+
"step": 510
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.03,
|
| 328 |
+
"learning_rate": 3.6413843888070695e-05,
|
| 329 |
+
"loss": 1.4226,
|
| 330 |
+
"step": 520
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.05,
|
| 334 |
+
"learning_rate": 3.604565537555229e-05,
|
| 335 |
+
"loss": 1.3538,
|
| 336 |
+
"step": 530
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.07,
|
| 340 |
+
"learning_rate": 3.567746686303387e-05,
|
| 341 |
+
"loss": 1.3888,
|
| 342 |
+
"step": 540
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.09,
|
| 346 |
+
"learning_rate": 3.5309278350515465e-05,
|
| 347 |
+
"loss": 1.3147,
|
| 348 |
+
"step": 550
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 1.11,
|
| 352 |
+
"learning_rate": 3.494108983799706e-05,
|
| 353 |
+
"loss": 1.414,
|
| 354 |
+
"step": 560
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.13,
|
| 358 |
+
"learning_rate": 3.457290132547865e-05,
|
| 359 |
+
"loss": 1.2426,
|
| 360 |
+
"step": 570
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.15,
|
| 364 |
+
"learning_rate": 3.4204712812960235e-05,
|
| 365 |
+
"loss": 1.3277,
|
| 366 |
+
"step": 580
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.17,
|
| 370 |
+
"learning_rate": 3.383652430044183e-05,
|
| 371 |
+
"loss": 1.3334,
|
| 372 |
+
"step": 590
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.19,
|
| 376 |
+
"learning_rate": 3.346833578792342e-05,
|
| 377 |
+
"loss": 1.3156,
|
| 378 |
+
"step": 600
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.21,
|
| 382 |
+
"learning_rate": 3.310014727540501e-05,
|
| 383 |
+
"loss": 1.2929,
|
| 384 |
+
"step": 610
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.23,
|
| 388 |
+
"learning_rate": 3.2731958762886596e-05,
|
| 389 |
+
"loss": 1.274,
|
| 390 |
+
"step": 620
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.25,
|
| 394 |
+
"learning_rate": 3.236377025036819e-05,
|
| 395 |
+
"loss": 1.3538,
|
| 396 |
+
"step": 630
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 1.27,
|
| 400 |
+
"learning_rate": 3.199558173784978e-05,
|
| 401 |
+
"loss": 1.3097,
|
| 402 |
+
"step": 640
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.29,
|
| 406 |
+
"learning_rate": 3.162739322533137e-05,
|
| 407 |
+
"loss": 1.3836,
|
| 408 |
+
"step": 650
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.31,
|
| 412 |
+
"learning_rate": 3.1259204712812965e-05,
|
| 413 |
+
"loss": 1.2941,
|
| 414 |
+
"step": 660
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.33,
|
| 418 |
+
"learning_rate": 3.089101620029455e-05,
|
| 419 |
+
"loss": 1.3041,
|
| 420 |
+
"step": 670
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.35,
|
| 424 |
+
"learning_rate": 3.052282768777614e-05,
|
| 425 |
+
"loss": 1.35,
|
| 426 |
+
"step": 680
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 1.37,
|
| 430 |
+
"learning_rate": 3.015463917525773e-05,
|
| 431 |
+
"loss": 1.2847,
|
| 432 |
+
"step": 690
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.39,
|
| 436 |
+
"learning_rate": 2.9786450662739324e-05,
|
| 437 |
+
"loss": 1.3606,
|
| 438 |
+
"step": 700
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.41,
|
| 442 |
+
"learning_rate": 2.9418262150220916e-05,
|
| 443 |
+
"loss": 1.2239,
|
| 444 |
+
"step": 710
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.43,
|
| 448 |
+
"learning_rate": 2.9050073637702508e-05,
|
| 449 |
+
"loss": 1.2794,
|
| 450 |
+
"step": 720
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.45,
|
| 454 |
+
"learning_rate": 2.8681885125184093e-05,
|
| 455 |
+
"loss": 1.2988,
|
| 456 |
+
"step": 730
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.47,
|
| 460 |
+
"learning_rate": 2.8313696612665685e-05,
|
| 461 |
+
"loss": 1.2714,
|
| 462 |
+
"step": 740
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 1.49,
|
| 466 |
+
"learning_rate": 2.7945508100147278e-05,
|
| 467 |
+
"loss": 1.2118,
|
| 468 |
+
"step": 750
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.51,
|
| 472 |
+
"learning_rate": 2.757731958762887e-05,
|
| 473 |
+
"loss": 1.2792,
|
| 474 |
+
"step": 760
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 1.53,
|
| 478 |
+
"learning_rate": 2.7209131075110455e-05,
|
| 479 |
+
"loss": 1.3307,
|
| 480 |
+
"step": 770
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.55,
|
| 484 |
+
"learning_rate": 2.6840942562592047e-05,
|
| 485 |
+
"loss": 1.2683,
|
| 486 |
+
"step": 780
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.57,
|
| 490 |
+
"learning_rate": 2.647275405007364e-05,
|
| 491 |
+
"loss": 1.2789,
|
| 492 |
+
"step": 790
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.59,
|
| 496 |
+
"learning_rate": 2.610456553755523e-05,
|
| 497 |
+
"loss": 1.261,
|
| 498 |
+
"step": 800
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 1.61,
|
| 502 |
+
"learning_rate": 2.5736377025036817e-05,
|
| 503 |
+
"loss": 1.3408,
|
| 504 |
+
"step": 810
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 1.63,
|
| 508 |
+
"learning_rate": 2.536818851251841e-05,
|
| 509 |
+
"loss": 1.27,
|
| 510 |
+
"step": 820
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 1.65,
|
| 514 |
+
"learning_rate": 2.5e-05,
|
| 515 |
+
"loss": 1.1915,
|
| 516 |
+
"step": 830
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 1.67,
|
| 520 |
+
"learning_rate": 2.4631811487481593e-05,
|
| 521 |
+
"loss": 1.1396,
|
| 522 |
+
"step": 840
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.69,
|
| 526 |
+
"learning_rate": 2.4263622974963182e-05,
|
| 527 |
+
"loss": 1.1942,
|
| 528 |
+
"step": 850
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.71,
|
| 532 |
+
"learning_rate": 2.3895434462444774e-05,
|
| 533 |
+
"loss": 1.1793,
|
| 534 |
+
"step": 860
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 1.73,
|
| 538 |
+
"learning_rate": 2.3527245949926363e-05,
|
| 539 |
+
"loss": 1.192,
|
| 540 |
+
"step": 870
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 1.75,
|
| 544 |
+
"learning_rate": 2.3159057437407955e-05,
|
| 545 |
+
"loss": 1.2951,
|
| 546 |
+
"step": 880
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 1.77,
|
| 550 |
+
"learning_rate": 2.2790868924889544e-05,
|
| 551 |
+
"loss": 1.2129,
|
| 552 |
+
"step": 890
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.79,
|
| 556 |
+
"learning_rate": 2.2422680412371136e-05,
|
| 557 |
+
"loss": 1.192,
|
| 558 |
+
"step": 900
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 1.81,
|
| 562 |
+
"learning_rate": 2.2054491899852725e-05,
|
| 563 |
+
"loss": 1.2012,
|
| 564 |
+
"step": 910
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.83,
|
| 568 |
+
"learning_rate": 2.1686303387334317e-05,
|
| 569 |
+
"loss": 1.2543,
|
| 570 |
+
"step": 920
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.85,
|
| 574 |
+
"learning_rate": 2.1318114874815906e-05,
|
| 575 |
+
"loss": 1.2318,
|
| 576 |
+
"step": 930
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.87,
|
| 580 |
+
"learning_rate": 2.0949926362297498e-05,
|
| 581 |
+
"loss": 1.223,
|
| 582 |
+
"step": 940
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.89,
|
| 586 |
+
"learning_rate": 2.0581737849779087e-05,
|
| 587 |
+
"loss": 1.2347,
|
| 588 |
+
"step": 950
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 1.91,
|
| 592 |
+
"learning_rate": 2.021354933726068e-05,
|
| 593 |
+
"loss": 1.2412,
|
| 594 |
+
"step": 960
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.93,
|
| 598 |
+
"learning_rate": 1.9845360824742268e-05,
|
| 599 |
+
"loss": 1.1686,
|
| 600 |
+
"step": 970
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.95,
|
| 604 |
+
"learning_rate": 1.947717231222386e-05,
|
| 605 |
+
"loss": 1.1424,
|
| 606 |
+
"step": 980
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.97,
|
| 610 |
+
"learning_rate": 1.910898379970545e-05,
|
| 611 |
+
"loss": 1.2094,
|
| 612 |
+
"step": 990
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.99,
|
| 616 |
+
"learning_rate": 1.874079528718704e-05,
|
| 617 |
+
"loss": 1.2418,
|
| 618 |
+
"step": 1000
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 2.0,
|
| 622 |
+
"eval_accuracy": 0.7571063979582857,
|
| 623 |
+
"eval_loss": 0.9226512908935547,
|
| 624 |
+
"eval_runtime": 106.3435,
|
| 625 |
+
"eval_samples_per_second": 106.852,
|
| 626 |
+
"eval_steps_per_second": 3.348,
|
| 627 |
+
"step": 1006
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.01,
|
| 631 |
+
"learning_rate": 1.8372606774668633e-05,
|
| 632 |
+
"loss": 1.2256,
|
| 633 |
+
"step": 1010
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.03,
|
| 637 |
+
"learning_rate": 1.800441826215022e-05,
|
| 638 |
+
"loss": 1.1317,
|
| 639 |
+
"step": 1020
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.05,
|
| 643 |
+
"learning_rate": 1.7636229749631814e-05,
|
| 644 |
+
"loss": 1.1062,
|
| 645 |
+
"step": 1030
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 2.07,
|
| 649 |
+
"learning_rate": 1.7268041237113403e-05,
|
| 650 |
+
"loss": 1.1442,
|
| 651 |
+
"step": 1040
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 2.09,
|
| 655 |
+
"learning_rate": 1.6899852724594995e-05,
|
| 656 |
+
"loss": 1.0836,
|
| 657 |
+
"step": 1050
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 2.11,
|
| 661 |
+
"learning_rate": 1.6531664212076583e-05,
|
| 662 |
+
"loss": 1.1139,
|
| 663 |
+
"step": 1060
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 2.13,
|
| 667 |
+
"learning_rate": 1.6163475699558176e-05,
|
| 668 |
+
"loss": 1.0567,
|
| 669 |
+
"step": 1070
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.15,
|
| 673 |
+
"learning_rate": 1.5795287187039764e-05,
|
| 674 |
+
"loss": 1.0716,
|
| 675 |
+
"step": 1080
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 2.17,
|
| 679 |
+
"learning_rate": 1.5427098674521357e-05,
|
| 680 |
+
"loss": 1.1631,
|
| 681 |
+
"step": 1090
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 2.19,
|
| 685 |
+
"learning_rate": 1.5058910162002945e-05,
|
| 686 |
+
"loss": 1.0727,
|
| 687 |
+
"step": 1100
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 2.21,
|
| 691 |
+
"learning_rate": 1.4690721649484537e-05,
|
| 692 |
+
"loss": 1.1132,
|
| 693 |
+
"step": 1110
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 2.23,
|
| 697 |
+
"learning_rate": 1.4322533136966126e-05,
|
| 698 |
+
"loss": 1.0497,
|
| 699 |
+
"step": 1120
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 2.25,
|
| 703 |
+
"learning_rate": 1.3954344624447718e-05,
|
| 704 |
+
"loss": 1.1202,
|
| 705 |
+
"step": 1130
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.27,
|
| 709 |
+
"learning_rate": 1.3586156111929307e-05,
|
| 710 |
+
"loss": 1.0817,
|
| 711 |
+
"step": 1140
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.29,
|
| 715 |
+
"learning_rate": 1.32179675994109e-05,
|
| 716 |
+
"loss": 1.1048,
|
| 717 |
+
"step": 1150
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 2.31,
|
| 721 |
+
"learning_rate": 1.2849779086892488e-05,
|
| 722 |
+
"loss": 1.1784,
|
| 723 |
+
"step": 1160
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 2.32,
|
| 727 |
+
"learning_rate": 1.248159057437408e-05,
|
| 728 |
+
"loss": 1.08,
|
| 729 |
+
"step": 1170
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 2.34,
|
| 733 |
+
"learning_rate": 1.211340206185567e-05,
|
| 734 |
+
"loss": 1.1461,
|
| 735 |
+
"step": 1180
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.36,
|
| 739 |
+
"learning_rate": 1.1745213549337261e-05,
|
| 740 |
+
"loss": 1.0777,
|
| 741 |
+
"step": 1190
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 2.38,
|
| 745 |
+
"learning_rate": 1.1377025036818852e-05,
|
| 746 |
+
"loss": 1.0184,
|
| 747 |
+
"step": 1200
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 2.4,
|
| 751 |
+
"learning_rate": 1.1008836524300442e-05,
|
| 752 |
+
"loss": 1.0455,
|
| 753 |
+
"step": 1210
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 2.42,
|
| 757 |
+
"learning_rate": 1.0640648011782033e-05,
|
| 758 |
+
"loss": 1.095,
|
| 759 |
+
"step": 1220
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 2.44,
|
| 763 |
+
"learning_rate": 1.0272459499263623e-05,
|
| 764 |
+
"loss": 1.1394,
|
| 765 |
+
"step": 1230
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.46,
|
| 769 |
+
"learning_rate": 9.904270986745213e-06,
|
| 770 |
+
"loss": 1.1076,
|
| 771 |
+
"step": 1240
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 2.48,
|
| 775 |
+
"learning_rate": 9.536082474226804e-06,
|
| 776 |
+
"loss": 1.0063,
|
| 777 |
+
"step": 1250
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 2.5,
|
| 781 |
+
"learning_rate": 9.167893961708394e-06,
|
| 782 |
+
"loss": 1.1044,
|
| 783 |
+
"step": 1260
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.52,
|
| 787 |
+
"learning_rate": 8.799705449189986e-06,
|
| 788 |
+
"loss": 1.041,
|
| 789 |
+
"step": 1270
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 2.54,
|
| 793 |
+
"learning_rate": 8.431516936671577e-06,
|
| 794 |
+
"loss": 1.0551,
|
| 795 |
+
"step": 1280
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 2.56,
|
| 799 |
+
"learning_rate": 8.063328424153167e-06,
|
| 800 |
+
"loss": 1.0601,
|
| 801 |
+
"step": 1290
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.58,
|
| 805 |
+
"learning_rate": 7.695139911634758e-06,
|
| 806 |
+
"loss": 1.1787,
|
| 807 |
+
"step": 1300
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 2.6,
|
| 811 |
+
"learning_rate": 7.326951399116348e-06,
|
| 812 |
+
"loss": 1.0445,
|
| 813 |
+
"step": 1310
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 2.62,
|
| 817 |
+
"learning_rate": 6.958762886597939e-06,
|
| 818 |
+
"loss": 1.105,
|
| 819 |
+
"step": 1320
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.64,
|
| 823 |
+
"learning_rate": 6.590574374079529e-06,
|
| 824 |
+
"loss": 1.0418,
|
| 825 |
+
"step": 1330
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.66,
|
| 829 |
+
"learning_rate": 6.22238586156112e-06,
|
| 830 |
+
"loss": 0.9967,
|
| 831 |
+
"step": 1340
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.68,
|
| 835 |
+
"learning_rate": 5.85419734904271e-06,
|
| 836 |
+
"loss": 1.0506,
|
| 837 |
+
"step": 1350
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.7,
|
| 841 |
+
"learning_rate": 5.486008836524301e-06,
|
| 842 |
+
"loss": 1.1154,
|
| 843 |
+
"step": 1360
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 2.72,
|
| 847 |
+
"learning_rate": 5.117820324005891e-06,
|
| 848 |
+
"loss": 1.058,
|
| 849 |
+
"step": 1370
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.74,
|
| 853 |
+
"learning_rate": 4.7496318114874815e-06,
|
| 854 |
+
"loss": 1.0997,
|
| 855 |
+
"step": 1380
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.76,
|
| 859 |
+
"learning_rate": 4.381443298969072e-06,
|
| 860 |
+
"loss": 1.0751,
|
| 861 |
+
"step": 1390
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 2.78,
|
| 865 |
+
"learning_rate": 4.013254786450663e-06,
|
| 866 |
+
"loss": 1.0658,
|
| 867 |
+
"step": 1400
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.8,
|
| 871 |
+
"learning_rate": 3.6450662739322538e-06,
|
| 872 |
+
"loss": 1.0243,
|
| 873 |
+
"step": 1410
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.82,
|
| 877 |
+
"learning_rate": 3.2768777614138442e-06,
|
| 878 |
+
"loss": 1.003,
|
| 879 |
+
"step": 1420
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 2.84,
|
| 883 |
+
"learning_rate": 2.9086892488954347e-06,
|
| 884 |
+
"loss": 1.0251,
|
| 885 |
+
"step": 1430
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.86,
|
| 889 |
+
"learning_rate": 2.540500736377025e-06,
|
| 890 |
+
"loss": 1.0049,
|
| 891 |
+
"step": 1440
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.88,
|
| 895 |
+
"learning_rate": 2.1723122238586156e-06,
|
| 896 |
+
"loss": 1.1317,
|
| 897 |
+
"step": 1450
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 2.9,
|
| 901 |
+
"learning_rate": 1.804123711340206e-06,
|
| 902 |
+
"loss": 1.063,
|
| 903 |
+
"step": 1460
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 2.92,
|
| 907 |
+
"learning_rate": 1.4359351988217967e-06,
|
| 908 |
+
"loss": 1.0558,
|
| 909 |
+
"step": 1470
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.94,
|
| 913 |
+
"learning_rate": 1.0677466863033874e-06,
|
| 914 |
+
"loss": 1.0737,
|
| 915 |
+
"step": 1480
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 2.96,
|
| 919 |
+
"learning_rate": 6.99558173784978e-07,
|
| 920 |
+
"loss": 1.0705,
|
| 921 |
+
"step": 1490
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 2.98,
|
| 925 |
+
"learning_rate": 3.313696612665685e-07,
|
| 926 |
+
"loss": 1.0657,
|
| 927 |
+
"step": 1500
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 3.0,
|
| 931 |
+
"eval_accuracy": 0.7760274575376221,
|
| 932 |
+
"eval_loss": 0.8486079573631287,
|
| 933 |
+
"eval_runtime": 106.6601,
|
| 934 |
+
"eval_samples_per_second": 106.535,
|
| 935 |
+
"eval_steps_per_second": 3.338,
|
| 936 |
+
"step": 1509
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 3.0,
|
| 940 |
+
"step": 1509,
|
| 941 |
+
"total_flos": 4.812819303609225e+18,
|
| 942 |
+
"train_loss": 1.6366877293570932,
|
| 943 |
+
"train_runtime": 3640.8494,
|
| 944 |
+
"train_samples_per_second": 53.054,
|
| 945 |
+
"train_steps_per_second": 0.414
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 3.0,
|
| 949 |
+
"eval_accuracy": 0.7760274575376221,
|
| 950 |
+
"eval_loss": 0.8486079573631287,
|
| 951 |
+
"eval_runtime": 130.1364,
|
| 952 |
+
"eval_samples_per_second": 87.316,
|
| 953 |
+
"eval_steps_per_second": 2.736,
|
| 954 |
+
"step": 1509
|
| 955 |
+
}
|
| 956 |
+
],
|
| 957 |
+
"logging_steps": 10,
|
| 958 |
+
"max_steps": 1509,
|
| 959 |
+
"num_train_epochs": 3,
|
| 960 |
+
"save_steps": 500,
|
| 961 |
+
"total_flos": 4.812819303609225e+18,
|
| 962 |
+
"trial_name": null,
|
| 963 |
+
"trial_params": null
|
| 964 |
+
}
|