End of training
Browse files- all_results.json +16 -0
- eval_results.json +11 -0
- runs/May11_16-29-23_3384b47cb3ff/events.out.tfevents.1683826962.3384b47cb3ff.2104.2 +3 -0
- train_results.json +8 -0
- trainer_state.json +709 -0
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 22.0,
|
| 3 |
+
"eval_accuracy": 0.6718301778542742,
|
| 4 |
+
"eval_f1": 0.6702822325183047,
|
| 5 |
+
"eval_loss": 0.971169114112854,
|
| 6 |
+
"eval_precision": 0.6698287004001044,
|
| 7 |
+
"eval_recall": 0.6718301778542742,
|
| 8 |
+
"eval_runtime": 24.527,
|
| 9 |
+
"eval_samples_per_second": 71.064,
|
| 10 |
+
"eval_steps_per_second": 4.444,
|
| 11 |
+
"total_flos": 6.932186220913164e+18,
|
| 12 |
+
"train_loss": 0.7904366488483819,
|
| 13 |
+
"train_runtime": 4221.7797,
|
| 14 |
+
"train_samples_per_second": 21.188,
|
| 15 |
+
"train_steps_per_second": 0.167
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 22.0,
|
| 3 |
+
"eval_accuracy": 0.6718301778542742,
|
| 4 |
+
"eval_f1": 0.6702822325183047,
|
| 5 |
+
"eval_loss": 0.971169114112854,
|
| 6 |
+
"eval_precision": 0.6698287004001044,
|
| 7 |
+
"eval_recall": 0.6718301778542742,
|
| 8 |
+
"eval_runtime": 24.527,
|
| 9 |
+
"eval_samples_per_second": 71.064,
|
| 10 |
+
"eval_steps_per_second": 4.444
|
| 11 |
+
}
|
runs/May11_16-29-23_3384b47cb3ff/events.out.tfevents.1683826962.3384b47cb3ff.2104.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f8a14fb9b48b0518238709c561a7af5be57340a54a78e759ad606f6a7d901e13
|
| 3 |
+
size 560
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 22.0,
|
| 3 |
+
"total_flos": 6.932186220913164e+18,
|
| 4 |
+
"train_loss": 0.7904366488483819,
|
| 5 |
+
"train_runtime": 4221.7797,
|
| 6 |
+
"train_samples_per_second": 21.188,
|
| 7 |
+
"train_steps_per_second": 0.167
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,709 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6718301778542742,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-best-finetuned-on-affectnet_short/checkpoint-672",
|
| 4 |
+
"epoch": 22.0,
|
| 5 |
+
"global_step": 704,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.31,
|
| 12 |
+
"learning_rate": 7.042253521126762e-06,
|
| 13 |
+
"loss": 2.1884,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.62,
|
| 18 |
+
"learning_rate": 1.4084507042253523e-05,
|
| 19 |
+
"loss": 2.0969,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.94,
|
| 24 |
+
"learning_rate": 2.112676056338028e-05,
|
| 25 |
+
"loss": 1.9968,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 1.0,
|
| 30 |
+
"eval_accuracy": 0.27538726333907054,
|
| 31 |
+
"eval_f1": 0.22796602838082472,
|
| 32 |
+
"eval_loss": 1.9113303422927856,
|
| 33 |
+
"eval_precision": 0.2518389295969545,
|
| 34 |
+
"eval_recall": 0.27538726333907054,
|
| 35 |
+
"eval_runtime": 25.038,
|
| 36 |
+
"eval_samples_per_second": 69.614,
|
| 37 |
+
"eval_steps_per_second": 4.353,
|
| 38 |
+
"step": 32
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.25,
|
| 42 |
+
"learning_rate": 2.8169014084507046e-05,
|
| 43 |
+
"loss": 1.8758,
|
| 44 |
+
"step": 40
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.56,
|
| 48 |
+
"learning_rate": 3.5211267605633805e-05,
|
| 49 |
+
"loss": 1.6422,
|
| 50 |
+
"step": 50
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.88,
|
| 54 |
+
"learning_rate": 4.225352112676056e-05,
|
| 55 |
+
"loss": 1.4178,
|
| 56 |
+
"step": 60
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 2.0,
|
| 60 |
+
"eval_accuracy": 0.5048766494549627,
|
| 61 |
+
"eval_f1": 0.4900447126711682,
|
| 62 |
+
"eval_loss": 1.2704347372055054,
|
| 63 |
+
"eval_precision": 0.5149366289457634,
|
| 64 |
+
"eval_recall": 0.5048766494549627,
|
| 65 |
+
"eval_runtime": 24.2957,
|
| 66 |
+
"eval_samples_per_second": 71.741,
|
| 67 |
+
"eval_steps_per_second": 4.486,
|
| 68 |
+
"step": 64
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.19,
|
| 72 |
+
"learning_rate": 4.929577464788733e-05,
|
| 73 |
+
"loss": 1.2352,
|
| 74 |
+
"step": 70
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 2.5,
|
| 78 |
+
"learning_rate": 4.928909952606635e-05,
|
| 79 |
+
"loss": 1.1754,
|
| 80 |
+
"step": 80
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 2.81,
|
| 84 |
+
"learning_rate": 4.849921011058452e-05,
|
| 85 |
+
"loss": 1.1751,
|
| 86 |
+
"step": 90
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 3.0,
|
| 90 |
+
"eval_accuracy": 0.5840504876649455,
|
| 91 |
+
"eval_f1": 0.5787373156833001,
|
| 92 |
+
"eval_loss": 1.111608624458313,
|
| 93 |
+
"eval_precision": 0.5891362431175351,
|
| 94 |
+
"eval_recall": 0.5840504876649455,
|
| 95 |
+
"eval_runtime": 24.5236,
|
| 96 |
+
"eval_samples_per_second": 71.074,
|
| 97 |
+
"eval_steps_per_second": 4.445,
|
| 98 |
+
"step": 96
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 3.12,
|
| 102 |
+
"learning_rate": 4.770932069510269e-05,
|
| 103 |
+
"loss": 1.1688,
|
| 104 |
+
"step": 100
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 3.44,
|
| 108 |
+
"learning_rate": 4.691943127962086e-05,
|
| 109 |
+
"loss": 1.0893,
|
| 110 |
+
"step": 110
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 3.75,
|
| 114 |
+
"learning_rate": 4.6129541864139027e-05,
|
| 115 |
+
"loss": 1.0127,
|
| 116 |
+
"step": 120
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 4.0,
|
| 120 |
+
"eval_accuracy": 0.6161790017211703,
|
| 121 |
+
"eval_f1": 0.6141128202096976,
|
| 122 |
+
"eval_loss": 1.0236890316009521,
|
| 123 |
+
"eval_precision": 0.633451007227001,
|
| 124 |
+
"eval_recall": 0.6161790017211703,
|
| 125 |
+
"eval_runtime": 24.3337,
|
| 126 |
+
"eval_samples_per_second": 71.629,
|
| 127 |
+
"eval_steps_per_second": 4.479,
|
| 128 |
+
"step": 128
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 4.06,
|
| 132 |
+
"learning_rate": 4.533965244865719e-05,
|
| 133 |
+
"loss": 1.0259,
|
| 134 |
+
"step": 130
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 4.38,
|
| 138 |
+
"learning_rate": 4.454976303317536e-05,
|
| 139 |
+
"loss": 0.9659,
|
| 140 |
+
"step": 140
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 4.69,
|
| 144 |
+
"learning_rate": 4.3759873617693526e-05,
|
| 145 |
+
"loss": 0.9996,
|
| 146 |
+
"step": 150
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 5.0,
|
| 150 |
+
"learning_rate": 4.2969984202211694e-05,
|
| 151 |
+
"loss": 0.9969,
|
| 152 |
+
"step": 160
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 5.0,
|
| 156 |
+
"eval_accuracy": 0.6259323006310958,
|
| 157 |
+
"eval_f1": 0.6150016278628547,
|
| 158 |
+
"eval_loss": 0.989003598690033,
|
| 159 |
+
"eval_precision": 0.6294369746617599,
|
| 160 |
+
"eval_recall": 0.6259323006310958,
|
| 161 |
+
"eval_runtime": 24.5638,
|
| 162 |
+
"eval_samples_per_second": 70.958,
|
| 163 |
+
"eval_steps_per_second": 4.437,
|
| 164 |
+
"step": 160
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 5.31,
|
| 168 |
+
"learning_rate": 4.218009478672986e-05,
|
| 169 |
+
"loss": 0.9251,
|
| 170 |
+
"step": 170
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 5.62,
|
| 174 |
+
"learning_rate": 4.1390205371248025e-05,
|
| 175 |
+
"loss": 0.9081,
|
| 176 |
+
"step": 180
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 5.94,
|
| 180 |
+
"learning_rate": 4.060031595576619e-05,
|
| 181 |
+
"loss": 0.9376,
|
| 182 |
+
"step": 190
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 6.0,
|
| 186 |
+
"eval_accuracy": 0.6190476190476191,
|
| 187 |
+
"eval_f1": 0.6182633788848451,
|
| 188 |
+
"eval_loss": 0.9767513275146484,
|
| 189 |
+
"eval_precision": 0.6334597304884556,
|
| 190 |
+
"eval_recall": 0.6190476190476191,
|
| 191 |
+
"eval_runtime": 24.3243,
|
| 192 |
+
"eval_samples_per_second": 71.657,
|
| 193 |
+
"eval_steps_per_second": 4.481,
|
| 194 |
+
"step": 192
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 6.25,
|
| 198 |
+
"learning_rate": 3.981042654028436e-05,
|
| 199 |
+
"loss": 0.8888,
|
| 200 |
+
"step": 200
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 6.56,
|
| 204 |
+
"learning_rate": 3.902053712480253e-05,
|
| 205 |
+
"loss": 0.891,
|
| 206 |
+
"step": 210
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 6.88,
|
| 210 |
+
"learning_rate": 3.82306477093207e-05,
|
| 211 |
+
"loss": 0.8299,
|
| 212 |
+
"step": 220
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 7.0,
|
| 216 |
+
"eval_accuracy": 0.6356855995410212,
|
| 217 |
+
"eval_f1": 0.6282041804399863,
|
| 218 |
+
"eval_loss": 0.9579089283943176,
|
| 219 |
+
"eval_precision": 0.6339366993746611,
|
| 220 |
+
"eval_recall": 0.6356855995410212,
|
| 221 |
+
"eval_runtime": 24.4946,
|
| 222 |
+
"eval_samples_per_second": 71.159,
|
| 223 |
+
"eval_steps_per_second": 4.45,
|
| 224 |
+
"step": 224
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 7.19,
|
| 228 |
+
"learning_rate": 3.744075829383886e-05,
|
| 229 |
+
"loss": 0.8563,
|
| 230 |
+
"step": 230
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 7.5,
|
| 234 |
+
"learning_rate": 3.665086887835703e-05,
|
| 235 |
+
"loss": 0.8477,
|
| 236 |
+
"step": 240
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 7.81,
|
| 240 |
+
"learning_rate": 3.58609794628752e-05,
|
| 241 |
+
"loss": 0.7645,
|
| 242 |
+
"step": 250
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 8.0,
|
| 246 |
+
"eval_accuracy": 0.648881239242685,
|
| 247 |
+
"eval_f1": 0.647358676044992,
|
| 248 |
+
"eval_loss": 0.9366128444671631,
|
| 249 |
+
"eval_precision": 0.6559082514099844,
|
| 250 |
+
"eval_recall": 0.648881239242685,
|
| 251 |
+
"eval_runtime": 24.292,
|
| 252 |
+
"eval_samples_per_second": 71.752,
|
| 253 |
+
"eval_steps_per_second": 4.487,
|
| 254 |
+
"step": 256
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 8.12,
|
| 258 |
+
"learning_rate": 3.507109004739337e-05,
|
| 259 |
+
"loss": 0.8149,
|
| 260 |
+
"step": 260
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 8.44,
|
| 264 |
+
"learning_rate": 3.4281200631911535e-05,
|
| 265 |
+
"loss": 0.7755,
|
| 266 |
+
"step": 270
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 8.75,
|
| 270 |
+
"learning_rate": 3.34913112164297e-05,
|
| 271 |
+
"loss": 0.7944,
|
| 272 |
+
"step": 280
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 9.0,
|
| 276 |
+
"eval_accuracy": 0.6442914515203672,
|
| 277 |
+
"eval_f1": 0.6447425675895946,
|
| 278 |
+
"eval_loss": 0.9302862286567688,
|
| 279 |
+
"eval_precision": 0.6494435537086374,
|
| 280 |
+
"eval_recall": 0.6442914515203672,
|
| 281 |
+
"eval_runtime": 24.5079,
|
| 282 |
+
"eval_samples_per_second": 71.12,
|
| 283 |
+
"eval_steps_per_second": 4.448,
|
| 284 |
+
"step": 288
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 9.06,
|
| 288 |
+
"learning_rate": 3.2701421800947866e-05,
|
| 289 |
+
"loss": 0.74,
|
| 290 |
+
"step": 290
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 9.38,
|
| 294 |
+
"learning_rate": 3.1911532385466034e-05,
|
| 295 |
+
"loss": 0.7196,
|
| 296 |
+
"step": 300
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 9.69,
|
| 300 |
+
"learning_rate": 3.11216429699842e-05,
|
| 301 |
+
"loss": 0.7218,
|
| 302 |
+
"step": 310
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 10.0,
|
| 306 |
+
"learning_rate": 3.0331753554502375e-05,
|
| 307 |
+
"loss": 0.7334,
|
| 308 |
+
"step": 320
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 10.0,
|
| 312 |
+
"eval_accuracy": 0.6546184738955824,
|
| 313 |
+
"eval_f1": 0.6523112247053579,
|
| 314 |
+
"eval_loss": 0.9510300755500793,
|
| 315 |
+
"eval_precision": 0.6634128950611891,
|
| 316 |
+
"eval_recall": 0.6546184738955824,
|
| 317 |
+
"eval_runtime": 24.2151,
|
| 318 |
+
"eval_samples_per_second": 71.98,
|
| 319 |
+
"eval_steps_per_second": 4.501,
|
| 320 |
+
"step": 320
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 10.31,
|
| 324 |
+
"learning_rate": 2.9541864139020537e-05,
|
| 325 |
+
"loss": 0.7152,
|
| 326 |
+
"step": 330
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 10.62,
|
| 330 |
+
"learning_rate": 2.8751974723538705e-05,
|
| 331 |
+
"loss": 0.6944,
|
| 332 |
+
"step": 340
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 10.94,
|
| 336 |
+
"learning_rate": 2.7962085308056874e-05,
|
| 337 |
+
"loss": 0.6596,
|
| 338 |
+
"step": 350
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 11.0,
|
| 342 |
+
"eval_accuracy": 0.6448651749856569,
|
| 343 |
+
"eval_f1": 0.6428493928535022,
|
| 344 |
+
"eval_loss": 0.936934232711792,
|
| 345 |
+
"eval_precision": 0.6528079277533798,
|
| 346 |
+
"eval_recall": 0.6448651749856569,
|
| 347 |
+
"eval_runtime": 24.2671,
|
| 348 |
+
"eval_samples_per_second": 71.826,
|
| 349 |
+
"eval_steps_per_second": 4.492,
|
| 350 |
+
"step": 352
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 11.25,
|
| 354 |
+
"learning_rate": 2.7172195892575043e-05,
|
| 355 |
+
"loss": 0.6759,
|
| 356 |
+
"step": 360
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 11.56,
|
| 360 |
+
"learning_rate": 2.638230647709321e-05,
|
| 361 |
+
"loss": 0.6274,
|
| 362 |
+
"step": 370
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 11.88,
|
| 366 |
+
"learning_rate": 2.5592417061611373e-05,
|
| 367 |
+
"loss": 0.6781,
|
| 368 |
+
"step": 380
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 12.0,
|
| 372 |
+
"eval_accuracy": 0.6368330464716007,
|
| 373 |
+
"eval_f1": 0.6359635726412315,
|
| 374 |
+
"eval_loss": 0.9716911911964417,
|
| 375 |
+
"eval_precision": 0.6513038081890579,
|
| 376 |
+
"eval_recall": 0.6368330464716007,
|
| 377 |
+
"eval_runtime": 24.1568,
|
| 378 |
+
"eval_samples_per_second": 72.154,
|
| 379 |
+
"eval_steps_per_second": 4.512,
|
| 380 |
+
"step": 384
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 12.19,
|
| 384 |
+
"learning_rate": 2.480252764612954e-05,
|
| 385 |
+
"loss": 0.6889,
|
| 386 |
+
"step": 390
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 12.5,
|
| 390 |
+
"learning_rate": 2.401263823064771e-05,
|
| 391 |
+
"loss": 0.6189,
|
| 392 |
+
"step": 400
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 12.81,
|
| 396 |
+
"learning_rate": 2.322274881516588e-05,
|
| 397 |
+
"loss": 0.5688,
|
| 398 |
+
"step": 410
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 13.0,
|
| 402 |
+
"eval_accuracy": 0.6540447504302926,
|
| 403 |
+
"eval_f1": 0.6495260815424722,
|
| 404 |
+
"eval_loss": 0.95087730884552,
|
| 405 |
+
"eval_precision": 0.6531237727480816,
|
| 406 |
+
"eval_recall": 0.6540447504302926,
|
| 407 |
+
"eval_runtime": 24.3359,
|
| 408 |
+
"eval_samples_per_second": 71.623,
|
| 409 |
+
"eval_steps_per_second": 4.479,
|
| 410 |
+
"step": 416
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 13.12,
|
| 414 |
+
"learning_rate": 2.2432859399684044e-05,
|
| 415 |
+
"loss": 0.6399,
|
| 416 |
+
"step": 420
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 13.44,
|
| 420 |
+
"learning_rate": 2.1642969984202213e-05,
|
| 421 |
+
"loss": 0.5852,
|
| 422 |
+
"step": 430
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 13.75,
|
| 426 |
+
"learning_rate": 2.0853080568720378e-05,
|
| 427 |
+
"loss": 0.5766,
|
| 428 |
+
"step": 440
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 14.0,
|
| 432 |
+
"eval_accuracy": 0.6615031554790591,
|
| 433 |
+
"eval_f1": 0.6601098289619065,
|
| 434 |
+
"eval_loss": 0.9484531283378601,
|
| 435 |
+
"eval_precision": 0.6655227009747305,
|
| 436 |
+
"eval_recall": 0.6615031554790591,
|
| 437 |
+
"eval_runtime": 24.0407,
|
| 438 |
+
"eval_samples_per_second": 72.502,
|
| 439 |
+
"eval_steps_per_second": 4.534,
|
| 440 |
+
"step": 448
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 14.06,
|
| 444 |
+
"learning_rate": 2.0063191153238547e-05,
|
| 445 |
+
"loss": 0.5559,
|
| 446 |
+
"step": 450
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 14.38,
|
| 450 |
+
"learning_rate": 1.9273301737756715e-05,
|
| 451 |
+
"loss": 0.5602,
|
| 452 |
+
"step": 460
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 14.69,
|
| 456 |
+
"learning_rate": 1.848341232227488e-05,
|
| 457 |
+
"loss": 0.5529,
|
| 458 |
+
"step": 470
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 15.0,
|
| 462 |
+
"learning_rate": 1.769352290679305e-05,
|
| 463 |
+
"loss": 0.5529,
|
| 464 |
+
"step": 480
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 15.0,
|
| 468 |
+
"eval_accuracy": 0.6569133677567413,
|
| 469 |
+
"eval_f1": 0.6537964772465467,
|
| 470 |
+
"eval_loss": 0.9590263962745667,
|
| 471 |
+
"eval_precision": 0.6561006700640171,
|
| 472 |
+
"eval_recall": 0.6569133677567413,
|
| 473 |
+
"eval_runtime": 24.2565,
|
| 474 |
+
"eval_samples_per_second": 71.857,
|
| 475 |
+
"eval_steps_per_second": 4.494,
|
| 476 |
+
"step": 480
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 15.31,
|
| 480 |
+
"learning_rate": 1.6903633491311218e-05,
|
| 481 |
+
"loss": 0.5288,
|
| 482 |
+
"step": 490
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 15.62,
|
| 486 |
+
"learning_rate": 1.6113744075829386e-05,
|
| 487 |
+
"loss": 0.5355,
|
| 488 |
+
"step": 500
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 15.94,
|
| 492 |
+
"learning_rate": 1.532385466034755e-05,
|
| 493 |
+
"loss": 0.4998,
|
| 494 |
+
"step": 510
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 16.0,
|
| 498 |
+
"eval_accuracy": 0.6511761331038439,
|
| 499 |
+
"eval_f1": 0.6487852957891203,
|
| 500 |
+
"eval_loss": 0.9676968455314636,
|
| 501 |
+
"eval_precision": 0.6513647898316787,
|
| 502 |
+
"eval_recall": 0.6511761331038439,
|
| 503 |
+
"eval_runtime": 23.9784,
|
| 504 |
+
"eval_samples_per_second": 72.69,
|
| 505 |
+
"eval_steps_per_second": 4.546,
|
| 506 |
+
"step": 512
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 16.25,
|
| 510 |
+
"learning_rate": 1.4533965244865718e-05,
|
| 511 |
+
"loss": 0.4864,
|
| 512 |
+
"step": 520
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 16.56,
|
| 516 |
+
"learning_rate": 1.3744075829383887e-05,
|
| 517 |
+
"loss": 0.5188,
|
| 518 |
+
"step": 530
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 16.88,
|
| 522 |
+
"learning_rate": 1.2954186413902054e-05,
|
| 523 |
+
"loss": 0.4908,
|
| 524 |
+
"step": 540
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 17.0,
|
| 528 |
+
"eval_accuracy": 0.663798049340218,
|
| 529 |
+
"eval_f1": 0.6615693208028428,
|
| 530 |
+
"eval_loss": 0.9670336842536926,
|
| 531 |
+
"eval_precision": 0.6645442403159969,
|
| 532 |
+
"eval_recall": 0.663798049340218,
|
| 533 |
+
"eval_runtime": 24.1891,
|
| 534 |
+
"eval_samples_per_second": 72.057,
|
| 535 |
+
"eval_steps_per_second": 4.506,
|
| 536 |
+
"step": 544
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 17.19,
|
| 540 |
+
"learning_rate": 1.216429699842022e-05,
|
| 541 |
+
"loss": 0.4697,
|
| 542 |
+
"step": 550
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 17.5,
|
| 546 |
+
"learning_rate": 1.137440758293839e-05,
|
| 547 |
+
"loss": 0.4791,
|
| 548 |
+
"step": 560
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 17.81,
|
| 552 |
+
"learning_rate": 1.0584518167456558e-05,
|
| 553 |
+
"loss": 0.4682,
|
| 554 |
+
"step": 570
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 18.0,
|
| 558 |
+
"eval_accuracy": 0.6678141135972461,
|
| 559 |
+
"eval_f1": 0.668417951352874,
|
| 560 |
+
"eval_loss": 0.9634829163551331,
|
| 561 |
+
"eval_precision": 0.6707246062463791,
|
| 562 |
+
"eval_recall": 0.6678141135972461,
|
| 563 |
+
"eval_runtime": 24.0376,
|
| 564 |
+
"eval_samples_per_second": 72.511,
|
| 565 |
+
"eval_steps_per_second": 4.535,
|
| 566 |
+
"step": 576
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 18.12,
|
| 570 |
+
"learning_rate": 9.794628751974725e-06,
|
| 571 |
+
"loss": 0.5067,
|
| 572 |
+
"step": 580
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 18.44,
|
| 576 |
+
"learning_rate": 9.004739336492892e-06,
|
| 577 |
+
"loss": 0.4401,
|
| 578 |
+
"step": 590
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 18.75,
|
| 582 |
+
"learning_rate": 8.214849921011059e-06,
|
| 583 |
+
"loss": 0.4761,
|
| 584 |
+
"step": 600
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 19.0,
|
| 588 |
+
"eval_accuracy": 0.6666666666666666,
|
| 589 |
+
"eval_f1": 0.6658035749901843,
|
| 590 |
+
"eval_loss": 0.968035876750946,
|
| 591 |
+
"eval_precision": 0.6673820855481607,
|
| 592 |
+
"eval_recall": 0.6666666666666666,
|
| 593 |
+
"eval_runtime": 24.4115,
|
| 594 |
+
"eval_samples_per_second": 71.401,
|
| 595 |
+
"eval_steps_per_second": 4.465,
|
| 596 |
+
"step": 608
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 19.06,
|
| 600 |
+
"learning_rate": 7.424960505529226e-06,
|
| 601 |
+
"loss": 0.4684,
|
| 602 |
+
"step": 610
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 19.38,
|
| 606 |
+
"learning_rate": 6.635071090047394e-06,
|
| 607 |
+
"loss": 0.4466,
|
| 608 |
+
"step": 620
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 19.69,
|
| 612 |
+
"learning_rate": 5.845181674565561e-06,
|
| 613 |
+
"loss": 0.4507,
|
| 614 |
+
"step": 630
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 20.0,
|
| 618 |
+
"learning_rate": 5.055292259083728e-06,
|
| 619 |
+
"loss": 0.4161,
|
| 620 |
+
"step": 640
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 20.0,
|
| 624 |
+
"eval_accuracy": 0.6712564543889845,
|
| 625 |
+
"eval_f1": 0.6700650267836397,
|
| 626 |
+
"eval_loss": 0.9700806736946106,
|
| 627 |
+
"eval_precision": 0.6719394542069612,
|
| 628 |
+
"eval_recall": 0.6712564543889845,
|
| 629 |
+
"eval_runtime": 24.0864,
|
| 630 |
+
"eval_samples_per_second": 72.364,
|
| 631 |
+
"eval_steps_per_second": 4.525,
|
| 632 |
+
"step": 640
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 20.31,
|
| 636 |
+
"learning_rate": 4.265402843601897e-06,
|
| 637 |
+
"loss": 0.4724,
|
| 638 |
+
"step": 650
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 20.62,
|
| 642 |
+
"learning_rate": 3.4755134281200636e-06,
|
| 643 |
+
"loss": 0.4127,
|
| 644 |
+
"step": 660
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 20.94,
|
| 648 |
+
"learning_rate": 2.685624012638231e-06,
|
| 649 |
+
"loss": 0.4295,
|
| 650 |
+
"step": 670
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 21.0,
|
| 654 |
+
"eval_accuracy": 0.6718301778542742,
|
| 655 |
+
"eval_f1": 0.6702822325183047,
|
| 656 |
+
"eval_loss": 0.971169114112854,
|
| 657 |
+
"eval_precision": 0.6698287004001044,
|
| 658 |
+
"eval_recall": 0.6718301778542742,
|
| 659 |
+
"eval_runtime": 24.5456,
|
| 660 |
+
"eval_samples_per_second": 71.011,
|
| 661 |
+
"eval_steps_per_second": 4.441,
|
| 662 |
+
"step": 672
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 21.25,
|
| 666 |
+
"learning_rate": 1.8957345971563984e-06,
|
| 667 |
+
"loss": 0.4618,
|
| 668 |
+
"step": 680
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 21.56,
|
| 672 |
+
"learning_rate": 1.1058451816745656e-06,
|
| 673 |
+
"loss": 0.434,
|
| 674 |
+
"step": 690
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 21.88,
|
| 678 |
+
"learning_rate": 3.15955766192733e-07,
|
| 679 |
+
"loss": 0.434,
|
| 680 |
+
"step": 700
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 22.0,
|
| 684 |
+
"eval_accuracy": 0.6706827309236948,
|
| 685 |
+
"eval_f1": 0.6690376760558668,
|
| 686 |
+
"eval_loss": 0.9755037426948547,
|
| 687 |
+
"eval_precision": 0.6705174221778998,
|
| 688 |
+
"eval_recall": 0.6706827309236948,
|
| 689 |
+
"eval_runtime": 24.2573,
|
| 690 |
+
"eval_samples_per_second": 71.855,
|
| 691 |
+
"eval_steps_per_second": 4.493,
|
| 692 |
+
"step": 704
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 22.0,
|
| 696 |
+
"step": 704,
|
| 697 |
+
"total_flos": 6.932186220913164e+18,
|
| 698 |
+
"train_loss": 0.7904366488483819,
|
| 699 |
+
"train_runtime": 4221.7797,
|
| 700 |
+
"train_samples_per_second": 21.188,
|
| 701 |
+
"train_steps_per_second": 0.167
|
| 702 |
+
}
|
| 703 |
+
],
|
| 704 |
+
"max_steps": 704,
|
| 705 |
+
"num_train_epochs": 22,
|
| 706 |
+
"total_flos": 6.932186220913164e+18,
|
| 707 |
+
"trial_name": null,
|
| 708 |
+
"trial_params": null
|
| 709 |
+
}
|