End of training
Browse files- README.md +3 -3
- all_results.json +12 -0
- eval_results.json +8 -0
- runs/May24_08-48-29_DESKTOP-SKBE9FB/events.out.tfevents.1716563348.DESKTOP-SKBE9FB.12648.1 +3 -0
- train_results.json +7 -0
- trainer_state.json +708 -0
README.md
CHANGED
|
@@ -22,7 +22,7 @@ model-index:
|
|
| 22 |
metrics:
|
| 23 |
- name: Accuracy
|
| 24 |
type: accuracy
|
| 25 |
-
value: 0.
|
| 26 |
---
|
| 27 |
|
| 28 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -32,8 +32,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 32 |
|
| 33 |
This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the imagefolder dataset.
|
| 34 |
It achieves the following results on the evaluation set:
|
| 35 |
-
- Loss: 0.
|
| 36 |
-
- Accuracy: 0.
|
| 37 |
|
| 38 |
## Model description
|
| 39 |
|
|
|
|
| 22 |
metrics:
|
| 23 |
- name: Accuracy
|
| 24 |
type: accuracy
|
| 25 |
+
value: 0.85
|
| 26 |
---
|
| 27 |
|
| 28 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 32 |
|
| 33 |
This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the imagefolder dataset.
|
| 34 |
It achieves the following results on the evaluation set:
|
| 35 |
+
- Loss: 0.6150
|
| 36 |
+
- Accuracy: 0.85
|
| 37 |
|
| 38 |
## Model description
|
| 39 |
|
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 37.97,
|
| 3 |
+
"eval_accuracy": 0.85,
|
| 4 |
+
"eval_loss": 0.6149908304214478,
|
| 5 |
+
"eval_runtime": 1.3442,
|
| 6 |
+
"eval_samples_per_second": 44.635,
|
| 7 |
+
"eval_steps_per_second": 1.488,
|
| 8 |
+
"train_loss": 0.23852362962705748,
|
| 9 |
+
"train_runtime": 1111.7542,
|
| 10 |
+
"train_samples_per_second": 67.569,
|
| 11 |
+
"train_steps_per_second": 0.504
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 37.97,
|
| 3 |
+
"eval_accuracy": 0.85,
|
| 4 |
+
"eval_loss": 0.6149908304214478,
|
| 5 |
+
"eval_runtime": 1.3442,
|
| 6 |
+
"eval_samples_per_second": 44.635,
|
| 7 |
+
"eval_steps_per_second": 1.488
|
| 8 |
+
}
|
runs/May24_08-48-29_DESKTOP-SKBE9FB/events.out.tfevents.1716563348.DESKTOP-SKBE9FB.12648.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84ebbf48b6cd717c9d0432b404fc5f389985a00aafab95b1f8a7f5d49cc9e5a3
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 37.97,
|
| 3 |
+
"train_loss": 0.23852362962705748,
|
| 4 |
+
"train_runtime": 1111.7542,
|
| 5 |
+
"train_samples_per_second": 67.569,
|
| 6 |
+
"train_steps_per_second": 0.504
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,708 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.85,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-RU5-40\\checkpoint-177",
|
| 4 |
+
"epoch": 37.96610169491525,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 560,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.68,
|
| 13 |
+
"learning_rate": 1.785714285714286e-05,
|
| 14 |
+
"loss": 1.3806,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.95,
|
| 19 |
+
"eval_accuracy": 0.48333333333333334,
|
| 20 |
+
"eval_loss": 1.3384602069854736,
|
| 21 |
+
"eval_runtime": 1.0175,
|
| 22 |
+
"eval_samples_per_second": 58.969,
|
| 23 |
+
"eval_steps_per_second": 1.966,
|
| 24 |
+
"step": 14
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 1.36,
|
| 28 |
+
"learning_rate": 3.571428571428572e-05,
|
| 29 |
+
"loss": 1.3323,
|
| 30 |
+
"step": 20
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 1.97,
|
| 34 |
+
"eval_accuracy": 0.6,
|
| 35 |
+
"eval_loss": 1.180294156074524,
|
| 36 |
+
"eval_runtime": 1.0059,
|
| 37 |
+
"eval_samples_per_second": 59.648,
|
| 38 |
+
"eval_steps_per_second": 1.988,
|
| 39 |
+
"step": 29
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 2.03,
|
| 43 |
+
"learning_rate": 4.981203007518797e-05,
|
| 44 |
+
"loss": 1.2486,
|
| 45 |
+
"step": 30
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 2.71,
|
| 49 |
+
"learning_rate": 4.887218045112782e-05,
|
| 50 |
+
"loss": 1.1086,
|
| 51 |
+
"step": 40
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 2.98,
|
| 55 |
+
"eval_accuracy": 0.6333333333333333,
|
| 56 |
+
"eval_loss": 0.9834739565849304,
|
| 57 |
+
"eval_runtime": 1.0161,
|
| 58 |
+
"eval_samples_per_second": 59.048,
|
| 59 |
+
"eval_steps_per_second": 1.968,
|
| 60 |
+
"step": 44
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 3.39,
|
| 64 |
+
"learning_rate": 4.793233082706767e-05,
|
| 65 |
+
"loss": 0.927,
|
| 66 |
+
"step": 50
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 4.0,
|
| 70 |
+
"eval_accuracy": 0.7166666666666667,
|
| 71 |
+
"eval_loss": 0.8339998722076416,
|
| 72 |
+
"eval_runtime": 1.0317,
|
| 73 |
+
"eval_samples_per_second": 58.155,
|
| 74 |
+
"eval_steps_per_second": 1.939,
|
| 75 |
+
"step": 59
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 4.07,
|
| 79 |
+
"learning_rate": 4.699248120300752e-05,
|
| 80 |
+
"loss": 0.7855,
|
| 81 |
+
"step": 60
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 4.75,
|
| 85 |
+
"learning_rate": 4.605263157894737e-05,
|
| 86 |
+
"loss": 0.6591,
|
| 87 |
+
"step": 70
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 4.95,
|
| 91 |
+
"eval_accuracy": 0.7166666666666667,
|
| 92 |
+
"eval_loss": 0.7842686176300049,
|
| 93 |
+
"eval_runtime": 0.9874,
|
| 94 |
+
"eval_samples_per_second": 60.766,
|
| 95 |
+
"eval_steps_per_second": 2.026,
|
| 96 |
+
"step": 73
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 5.42,
|
| 100 |
+
"learning_rate": 4.511278195488722e-05,
|
| 101 |
+
"loss": 0.5201,
|
| 102 |
+
"step": 80
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 5.97,
|
| 106 |
+
"eval_accuracy": 0.7166666666666667,
|
| 107 |
+
"eval_loss": 0.7682653665542603,
|
| 108 |
+
"eval_runtime": 1.07,
|
| 109 |
+
"eval_samples_per_second": 56.076,
|
| 110 |
+
"eval_steps_per_second": 1.869,
|
| 111 |
+
"step": 88
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 6.1,
|
| 115 |
+
"learning_rate": 4.4172932330827074e-05,
|
| 116 |
+
"loss": 0.435,
|
| 117 |
+
"step": 90
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 6.78,
|
| 121 |
+
"learning_rate": 4.323308270676692e-05,
|
| 122 |
+
"loss": 0.3763,
|
| 123 |
+
"step": 100
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 6.98,
|
| 127 |
+
"eval_accuracy": 0.6833333333333333,
|
| 128 |
+
"eval_loss": 0.7880204319953918,
|
| 129 |
+
"eval_runtime": 1.0266,
|
| 130 |
+
"eval_samples_per_second": 58.446,
|
| 131 |
+
"eval_steps_per_second": 1.948,
|
| 132 |
+
"step": 103
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 7.46,
|
| 136 |
+
"learning_rate": 4.229323308270677e-05,
|
| 137 |
+
"loss": 0.26,
|
| 138 |
+
"step": 110
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 8.0,
|
| 142 |
+
"eval_accuracy": 0.7666666666666667,
|
| 143 |
+
"eval_loss": 0.6876354217529297,
|
| 144 |
+
"eval_runtime": 1.0561,
|
| 145 |
+
"eval_samples_per_second": 56.811,
|
| 146 |
+
"eval_steps_per_second": 1.894,
|
| 147 |
+
"step": 118
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 8.14,
|
| 151 |
+
"learning_rate": 4.135338345864662e-05,
|
| 152 |
+
"loss": 0.2651,
|
| 153 |
+
"step": 120
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 8.81,
|
| 157 |
+
"learning_rate": 4.041353383458647e-05,
|
| 158 |
+
"loss": 0.2219,
|
| 159 |
+
"step": 130
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 8.95,
|
| 163 |
+
"eval_accuracy": 0.7833333333333333,
|
| 164 |
+
"eval_loss": 0.7187811136245728,
|
| 165 |
+
"eval_runtime": 1.0257,
|
| 166 |
+
"eval_samples_per_second": 58.496,
|
| 167 |
+
"eval_steps_per_second": 1.95,
|
| 168 |
+
"step": 132
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 9.49,
|
| 172 |
+
"learning_rate": 3.9473684210526316e-05,
|
| 173 |
+
"loss": 0.2243,
|
| 174 |
+
"step": 140
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 9.97,
|
| 178 |
+
"eval_accuracy": 0.7,
|
| 179 |
+
"eval_loss": 0.8730494379997253,
|
| 180 |
+
"eval_runtime": 1.0464,
|
| 181 |
+
"eval_samples_per_second": 57.337,
|
| 182 |
+
"eval_steps_per_second": 1.911,
|
| 183 |
+
"step": 147
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 10.17,
|
| 187 |
+
"learning_rate": 3.8533834586466165e-05,
|
| 188 |
+
"loss": 0.1972,
|
| 189 |
+
"step": 150
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 10.85,
|
| 193 |
+
"learning_rate": 3.759398496240601e-05,
|
| 194 |
+
"loss": 0.178,
|
| 195 |
+
"step": 160
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 10.98,
|
| 199 |
+
"eval_accuracy": 0.7833333333333333,
|
| 200 |
+
"eval_loss": 0.6872482895851135,
|
| 201 |
+
"eval_runtime": 1.0432,
|
| 202 |
+
"eval_samples_per_second": 57.515,
|
| 203 |
+
"eval_steps_per_second": 1.917,
|
| 204 |
+
"step": 162
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 11.53,
|
| 208 |
+
"learning_rate": 3.665413533834587e-05,
|
| 209 |
+
"loss": 0.1944,
|
| 210 |
+
"step": 170
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 12.0,
|
| 214 |
+
"eval_accuracy": 0.85,
|
| 215 |
+
"eval_loss": 0.6149908304214478,
|
| 216 |
+
"eval_runtime": 1.0719,
|
| 217 |
+
"eval_samples_per_second": 55.973,
|
| 218 |
+
"eval_steps_per_second": 1.866,
|
| 219 |
+
"step": 177
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 12.2,
|
| 223 |
+
"learning_rate": 3.571428571428572e-05,
|
| 224 |
+
"loss": 0.1628,
|
| 225 |
+
"step": 180
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 12.88,
|
| 229 |
+
"learning_rate": 3.4774436090225565e-05,
|
| 230 |
+
"loss": 0.1422,
|
| 231 |
+
"step": 190
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 12.95,
|
| 235 |
+
"eval_accuracy": 0.7833333333333333,
|
| 236 |
+
"eval_loss": 0.6831592917442322,
|
| 237 |
+
"eval_runtime": 1.034,
|
| 238 |
+
"eval_samples_per_second": 58.029,
|
| 239 |
+
"eval_steps_per_second": 1.934,
|
| 240 |
+
"step": 191
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 13.56,
|
| 244 |
+
"learning_rate": 3.3834586466165414e-05,
|
| 245 |
+
"loss": 0.1117,
|
| 246 |
+
"step": 200
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 13.97,
|
| 250 |
+
"eval_accuracy": 0.7833333333333333,
|
| 251 |
+
"eval_loss": 0.7590230107307434,
|
| 252 |
+
"eval_runtime": 1.1059,
|
| 253 |
+
"eval_samples_per_second": 54.254,
|
| 254 |
+
"eval_steps_per_second": 1.808,
|
| 255 |
+
"step": 206
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 14.24,
|
| 259 |
+
"learning_rate": 3.289473684210527e-05,
|
| 260 |
+
"loss": 0.1081,
|
| 261 |
+
"step": 210
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 14.92,
|
| 265 |
+
"learning_rate": 3.195488721804512e-05,
|
| 266 |
+
"loss": 0.117,
|
| 267 |
+
"step": 220
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 14.98,
|
| 271 |
+
"eval_accuracy": 0.7666666666666667,
|
| 272 |
+
"eval_loss": 0.8428652882575989,
|
| 273 |
+
"eval_runtime": 1.0698,
|
| 274 |
+
"eval_samples_per_second": 56.085,
|
| 275 |
+
"eval_steps_per_second": 1.87,
|
| 276 |
+
"step": 221
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 15.59,
|
| 280 |
+
"learning_rate": 3.1015037593984966e-05,
|
| 281 |
+
"loss": 0.1176,
|
| 282 |
+
"step": 230
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 16.0,
|
| 286 |
+
"eval_accuracy": 0.7666666666666667,
|
| 287 |
+
"eval_loss": 0.9740858674049377,
|
| 288 |
+
"eval_runtime": 1.0637,
|
| 289 |
+
"eval_samples_per_second": 56.408,
|
| 290 |
+
"eval_steps_per_second": 1.88,
|
| 291 |
+
"step": 236
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 16.27,
|
| 295 |
+
"learning_rate": 3.007518796992481e-05,
|
| 296 |
+
"loss": 0.0974,
|
| 297 |
+
"step": 240
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 16.95,
|
| 301 |
+
"learning_rate": 2.9135338345864667e-05,
|
| 302 |
+
"loss": 0.1081,
|
| 303 |
+
"step": 250
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 16.95,
|
| 307 |
+
"eval_accuracy": 0.7833333333333333,
|
| 308 |
+
"eval_loss": 0.9106397032737732,
|
| 309 |
+
"eval_runtime": 1.0253,
|
| 310 |
+
"eval_samples_per_second": 58.521,
|
| 311 |
+
"eval_steps_per_second": 1.951,
|
| 312 |
+
"step": 250
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 17.63,
|
| 316 |
+
"learning_rate": 2.8195488721804515e-05,
|
| 317 |
+
"loss": 0.0928,
|
| 318 |
+
"step": 260
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 17.97,
|
| 322 |
+
"eval_accuracy": 0.7333333333333333,
|
| 323 |
+
"eval_loss": 0.9178593158721924,
|
| 324 |
+
"eval_runtime": 1.0767,
|
| 325 |
+
"eval_samples_per_second": 55.725,
|
| 326 |
+
"eval_steps_per_second": 1.857,
|
| 327 |
+
"step": 265
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 18.31,
|
| 331 |
+
"learning_rate": 2.7255639097744363e-05,
|
| 332 |
+
"loss": 0.0763,
|
| 333 |
+
"step": 270
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 18.98,
|
| 337 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 338 |
+
"loss": 0.0848,
|
| 339 |
+
"step": 280
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 18.98,
|
| 343 |
+
"eval_accuracy": 0.7666666666666667,
|
| 344 |
+
"eval_loss": 0.9694532155990601,
|
| 345 |
+
"eval_runtime": 1.0431,
|
| 346 |
+
"eval_samples_per_second": 57.519,
|
| 347 |
+
"eval_steps_per_second": 1.917,
|
| 348 |
+
"step": 280
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 19.66,
|
| 352 |
+
"learning_rate": 2.5375939849624064e-05,
|
| 353 |
+
"loss": 0.1045,
|
| 354 |
+
"step": 290
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 20.0,
|
| 358 |
+
"eval_accuracy": 0.8,
|
| 359 |
+
"eval_loss": 0.8805408477783203,
|
| 360 |
+
"eval_runtime": 1.0552,
|
| 361 |
+
"eval_samples_per_second": 56.862,
|
| 362 |
+
"eval_steps_per_second": 1.895,
|
| 363 |
+
"step": 295
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 20.34,
|
| 367 |
+
"learning_rate": 2.443609022556391e-05,
|
| 368 |
+
"loss": 0.1159,
|
| 369 |
+
"step": 300
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 20.95,
|
| 373 |
+
"eval_accuracy": 0.7666666666666667,
|
| 374 |
+
"eval_loss": 0.9457910060882568,
|
| 375 |
+
"eval_runtime": 1.0314,
|
| 376 |
+
"eval_samples_per_second": 58.171,
|
| 377 |
+
"eval_steps_per_second": 1.939,
|
| 378 |
+
"step": 309
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 21.02,
|
| 382 |
+
"learning_rate": 2.349624060150376e-05,
|
| 383 |
+
"loss": 0.084,
|
| 384 |
+
"step": 310
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 21.69,
|
| 388 |
+
"learning_rate": 2.255639097744361e-05,
|
| 389 |
+
"loss": 0.0748,
|
| 390 |
+
"step": 320
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 21.97,
|
| 394 |
+
"eval_accuracy": 0.7666666666666667,
|
| 395 |
+
"eval_loss": 0.8462807536125183,
|
| 396 |
+
"eval_runtime": 1.0447,
|
| 397 |
+
"eval_samples_per_second": 57.434,
|
| 398 |
+
"eval_steps_per_second": 1.914,
|
| 399 |
+
"step": 324
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 22.37,
|
| 403 |
+
"learning_rate": 2.161654135338346e-05,
|
| 404 |
+
"loss": 0.0641,
|
| 405 |
+
"step": 330
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 22.98,
|
| 409 |
+
"eval_accuracy": 0.8,
|
| 410 |
+
"eval_loss": 0.8815339207649231,
|
| 411 |
+
"eval_runtime": 1.0477,
|
| 412 |
+
"eval_samples_per_second": 57.27,
|
| 413 |
+
"eval_steps_per_second": 1.909,
|
| 414 |
+
"step": 339
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 23.05,
|
| 418 |
+
"learning_rate": 2.067669172932331e-05,
|
| 419 |
+
"loss": 0.0892,
|
| 420 |
+
"step": 340
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 23.73,
|
| 424 |
+
"learning_rate": 1.9736842105263158e-05,
|
| 425 |
+
"loss": 0.0799,
|
| 426 |
+
"step": 350
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 24.0,
|
| 430 |
+
"eval_accuracy": 0.75,
|
| 431 |
+
"eval_loss": 0.9426201581954956,
|
| 432 |
+
"eval_runtime": 1.0401,
|
| 433 |
+
"eval_samples_per_second": 57.685,
|
| 434 |
+
"eval_steps_per_second": 1.923,
|
| 435 |
+
"step": 354
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 24.41,
|
| 439 |
+
"learning_rate": 1.8796992481203007e-05,
|
| 440 |
+
"loss": 0.0921,
|
| 441 |
+
"step": 360
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 24.95,
|
| 445 |
+
"eval_accuracy": 0.75,
|
| 446 |
+
"eval_loss": 0.9211530089378357,
|
| 447 |
+
"eval_runtime": 1.0219,
|
| 448 |
+
"eval_samples_per_second": 58.712,
|
| 449 |
+
"eval_steps_per_second": 1.957,
|
| 450 |
+
"step": 368
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 25.08,
|
| 454 |
+
"learning_rate": 1.785714285714286e-05,
|
| 455 |
+
"loss": 0.0699,
|
| 456 |
+
"step": 370
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 25.76,
|
| 460 |
+
"learning_rate": 1.6917293233082707e-05,
|
| 461 |
+
"loss": 0.0602,
|
| 462 |
+
"step": 380
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 25.97,
|
| 466 |
+
"eval_accuracy": 0.75,
|
| 467 |
+
"eval_loss": 0.9827994704246521,
|
| 468 |
+
"eval_runtime": 1.0441,
|
| 469 |
+
"eval_samples_per_second": 57.465,
|
| 470 |
+
"eval_steps_per_second": 1.916,
|
| 471 |
+
"step": 383
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 26.44,
|
| 475 |
+
"learning_rate": 1.597744360902256e-05,
|
| 476 |
+
"loss": 0.059,
|
| 477 |
+
"step": 390
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 26.98,
|
| 481 |
+
"eval_accuracy": 0.8,
|
| 482 |
+
"eval_loss": 0.8860684633255005,
|
| 483 |
+
"eval_runtime": 1.0631,
|
| 484 |
+
"eval_samples_per_second": 56.44,
|
| 485 |
+
"eval_steps_per_second": 1.881,
|
| 486 |
+
"step": 398
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 27.12,
|
| 490 |
+
"learning_rate": 1.5037593984962406e-05,
|
| 491 |
+
"loss": 0.0544,
|
| 492 |
+
"step": 400
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 27.8,
|
| 496 |
+
"learning_rate": 1.4097744360902257e-05,
|
| 497 |
+
"loss": 0.0669,
|
| 498 |
+
"step": 410
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 28.0,
|
| 502 |
+
"eval_accuracy": 0.7333333333333333,
|
| 503 |
+
"eval_loss": 0.9302221536636353,
|
| 504 |
+
"eval_runtime": 1.0241,
|
| 505 |
+
"eval_samples_per_second": 58.591,
|
| 506 |
+
"eval_steps_per_second": 1.953,
|
| 507 |
+
"step": 413
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 28.47,
|
| 511 |
+
"learning_rate": 1.3157894736842106e-05,
|
| 512 |
+
"loss": 0.0508,
|
| 513 |
+
"step": 420
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 28.95,
|
| 517 |
+
"eval_accuracy": 0.7166666666666667,
|
| 518 |
+
"eval_loss": 1.0306384563446045,
|
| 519 |
+
"eval_runtime": 1.0394,
|
| 520 |
+
"eval_samples_per_second": 57.723,
|
| 521 |
+
"eval_steps_per_second": 1.924,
|
| 522 |
+
"step": 427
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 29.15,
|
| 526 |
+
"learning_rate": 1.2218045112781954e-05,
|
| 527 |
+
"loss": 0.0658,
|
| 528 |
+
"step": 430
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 29.83,
|
| 532 |
+
"learning_rate": 1.1278195488721805e-05,
|
| 533 |
+
"loss": 0.0585,
|
| 534 |
+
"step": 440
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 29.97,
|
| 538 |
+
"eval_accuracy": 0.75,
|
| 539 |
+
"eval_loss": 0.9148640036582947,
|
| 540 |
+
"eval_runtime": 1.0382,
|
| 541 |
+
"eval_samples_per_second": 57.79,
|
| 542 |
+
"eval_steps_per_second": 1.926,
|
| 543 |
+
"step": 442
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 30.51,
|
| 547 |
+
"learning_rate": 1.0338345864661655e-05,
|
| 548 |
+
"loss": 0.0619,
|
| 549 |
+
"step": 450
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 30.98,
|
| 553 |
+
"eval_accuracy": 0.7833333333333333,
|
| 554 |
+
"eval_loss": 0.8941699862480164,
|
| 555 |
+
"eval_runtime": 1.0414,
|
| 556 |
+
"eval_samples_per_second": 57.616,
|
| 557 |
+
"eval_steps_per_second": 1.921,
|
| 558 |
+
"step": 457
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 31.19,
|
| 562 |
+
"learning_rate": 9.398496240601503e-06,
|
| 563 |
+
"loss": 0.048,
|
| 564 |
+
"step": 460
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 31.86,
|
| 568 |
+
"learning_rate": 8.458646616541353e-06,
|
| 569 |
+
"loss": 0.0626,
|
| 570 |
+
"step": 470
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 32.0,
|
| 574 |
+
"eval_accuracy": 0.7666666666666667,
|
| 575 |
+
"eval_loss": 0.9069377779960632,
|
| 576 |
+
"eval_runtime": 1.0588,
|
| 577 |
+
"eval_samples_per_second": 56.665,
|
| 578 |
+
"eval_steps_per_second": 1.889,
|
| 579 |
+
"step": 472
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 32.54,
|
| 583 |
+
"learning_rate": 7.518796992481203e-06,
|
| 584 |
+
"loss": 0.0575,
|
| 585 |
+
"step": 480
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 32.95,
|
| 589 |
+
"eval_accuracy": 0.8,
|
| 590 |
+
"eval_loss": 0.8656221628189087,
|
| 591 |
+
"eval_runtime": 1.0337,
|
| 592 |
+
"eval_samples_per_second": 58.042,
|
| 593 |
+
"eval_steps_per_second": 1.935,
|
| 594 |
+
"step": 486
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 33.22,
|
| 598 |
+
"learning_rate": 6.578947368421053e-06,
|
| 599 |
+
"loss": 0.052,
|
| 600 |
+
"step": 490
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 33.9,
|
| 604 |
+
"learning_rate": 5.639097744360902e-06,
|
| 605 |
+
"loss": 0.0483,
|
| 606 |
+
"step": 500
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 33.97,
|
| 610 |
+
"eval_accuracy": 0.8166666666666667,
|
| 611 |
+
"eval_loss": 0.8778695464134216,
|
| 612 |
+
"eval_runtime": 1.0357,
|
| 613 |
+
"eval_samples_per_second": 57.935,
|
| 614 |
+
"eval_steps_per_second": 1.931,
|
| 615 |
+
"step": 501
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 34.58,
|
| 619 |
+
"learning_rate": 4.699248120300752e-06,
|
| 620 |
+
"loss": 0.0576,
|
| 621 |
+
"step": 510
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 34.98,
|
| 625 |
+
"eval_accuracy": 0.7833333333333333,
|
| 626 |
+
"eval_loss": 0.9077943563461304,
|
| 627 |
+
"eval_runtime": 1.0205,
|
| 628 |
+
"eval_samples_per_second": 58.793,
|
| 629 |
+
"eval_steps_per_second": 1.96,
|
| 630 |
+
"step": 516
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 35.25,
|
| 634 |
+
"learning_rate": 3.7593984962406014e-06,
|
| 635 |
+
"loss": 0.0429,
|
| 636 |
+
"step": 520
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 35.93,
|
| 640 |
+
"learning_rate": 2.819548872180451e-06,
|
| 641 |
+
"loss": 0.0633,
|
| 642 |
+
"step": 530
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 36.0,
|
| 646 |
+
"eval_accuracy": 0.8,
|
| 647 |
+
"eval_loss": 0.8879902958869934,
|
| 648 |
+
"eval_runtime": 1.0817,
|
| 649 |
+
"eval_samples_per_second": 55.469,
|
| 650 |
+
"eval_steps_per_second": 1.849,
|
| 651 |
+
"step": 531
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 36.61,
|
| 655 |
+
"learning_rate": 1.8796992481203007e-06,
|
| 656 |
+
"loss": 0.0511,
|
| 657 |
+
"step": 540
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 36.95,
|
| 661 |
+
"eval_accuracy": 0.8,
|
| 662 |
+
"eval_loss": 0.8573408722877502,
|
| 663 |
+
"eval_runtime": 1.0249,
|
| 664 |
+
"eval_samples_per_second": 58.542,
|
| 665 |
+
"eval_steps_per_second": 1.951,
|
| 666 |
+
"step": 545
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 37.29,
|
| 670 |
+
"learning_rate": 9.398496240601504e-07,
|
| 671 |
+
"loss": 0.0402,
|
| 672 |
+
"step": 550
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 37.97,
|
| 676 |
+
"learning_rate": 0.0,
|
| 677 |
+
"loss": 0.049,
|
| 678 |
+
"step": 560
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 37.97,
|
| 682 |
+
"eval_accuracy": 0.8,
|
| 683 |
+
"eval_loss": 0.8563874363899231,
|
| 684 |
+
"eval_runtime": 1.0455,
|
| 685 |
+
"eval_samples_per_second": 57.387,
|
| 686 |
+
"eval_steps_per_second": 1.913,
|
| 687 |
+
"step": 560
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 37.97,
|
| 691 |
+
"step": 560,
|
| 692 |
+
"total_flos": 5.526052834168259e+18,
|
| 693 |
+
"train_loss": 0.23852362962705748,
|
| 694 |
+
"train_runtime": 1111.7542,
|
| 695 |
+
"train_samples_per_second": 67.569,
|
| 696 |
+
"train_steps_per_second": 0.504
|
| 697 |
+
}
|
| 698 |
+
],
|
| 699 |
+
"logging_steps": 10,
|
| 700 |
+
"max_steps": 560,
|
| 701 |
+
"num_input_tokens_seen": 0,
|
| 702 |
+
"num_train_epochs": 40,
|
| 703 |
+
"save_steps": 500,
|
| 704 |
+
"total_flos": 5.526052834168259e+18,
|
| 705 |
+
"train_batch_size": 32,
|
| 706 |
+
"trial_name": null,
|
| 707 |
+
"trial_params": null
|
| 708 |
+
}
|