🍻 cheers
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"eval_accuracy": 0.8058823529411765,
|
| 4 |
+
"eval_loss": 0.6480256915092468,
|
| 5 |
+
"eval_runtime": 1.4895,
|
| 6 |
+
"eval_samples_per_second": 114.129,
|
| 7 |
+
"eval_steps_per_second": 14.77,
|
| 8 |
+
"total_flos": 1.6866433871220326e+18,
|
| 9 |
+
"train_loss": 0.11297949704276773,
|
| 10 |
+
"train_runtime": 502.5996,
|
| 11 |
+
"train_samples_per_second": 43.305,
|
| 12 |
+
"train_steps_per_second": 2.716
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"eval_accuracy": 0.8058823529411765,
|
| 4 |
+
"eval_loss": 0.6480256915092468,
|
| 5 |
+
"eval_runtime": 1.4895,
|
| 6 |
+
"eval_samples_per_second": 114.129,
|
| 7 |
+
"eval_steps_per_second": 14.77
|
| 8 |
+
}
|
runs/Aug02_05-24-55_cdbc38a7843c/events.out.tfevents.1659418442.cdbc38a7843c.73.7
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fa7f51636341d3b317d05b3181e8ca4960b3513cf9f8eea6322856753103e131
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"total_flos": 1.6866433871220326e+18,
|
| 4 |
+
"train_loss": 0.11297949704276773,
|
| 5 |
+
"train_runtime": 502.5996,
|
| 6 |
+
"train_samples_per_second": 43.305,
|
| 7 |
+
"train_steps_per_second": 2.716
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,958 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6480256915092468,
|
| 3 |
+
"best_model_checkpoint": "./croupier-creature-classifier/checkpoint-100",
|
| 4 |
+
"epoch": 15.0,
|
| 5 |
+
"global_step": 1365,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.11,
|
| 12 |
+
"learning_rate": 0.00019853479853479855,
|
| 13 |
+
"loss": 0.2843,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.22,
|
| 18 |
+
"learning_rate": 0.0001970695970695971,
|
| 19 |
+
"loss": 0.3895,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.33,
|
| 24 |
+
"learning_rate": 0.00019560439560439562,
|
| 25 |
+
"loss": 0.3592,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.44,
|
| 30 |
+
"learning_rate": 0.00019413919413919413,
|
| 31 |
+
"loss": 0.3891,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.55,
|
| 36 |
+
"learning_rate": 0.00019267399267399268,
|
| 37 |
+
"loss": 0.2862,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.66,
|
| 42 |
+
"learning_rate": 0.00019120879120879122,
|
| 43 |
+
"loss": 0.4721,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.77,
|
| 48 |
+
"learning_rate": 0.00018974358974358974,
|
| 49 |
+
"loss": 0.4275,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.88,
|
| 54 |
+
"learning_rate": 0.00018827838827838828,
|
| 55 |
+
"loss": 0.2297,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.99,
|
| 60 |
+
"learning_rate": 0.00018681318681318683,
|
| 61 |
+
"loss": 0.3175,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.1,
|
| 66 |
+
"learning_rate": 0.00018534798534798537,
|
| 67 |
+
"loss": 0.1967,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.1,
|
| 72 |
+
"eval_accuracy": 0.8058823529411765,
|
| 73 |
+
"eval_loss": 0.6480256915092468,
|
| 74 |
+
"eval_runtime": 1.3833,
|
| 75 |
+
"eval_samples_per_second": 122.894,
|
| 76 |
+
"eval_steps_per_second": 15.904,
|
| 77 |
+
"step": 100
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.21,
|
| 81 |
+
"learning_rate": 0.0001838827838827839,
|
| 82 |
+
"loss": 0.2783,
|
| 83 |
+
"step": 110
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 1.32,
|
| 87 |
+
"learning_rate": 0.0001824175824175824,
|
| 88 |
+
"loss": 0.3196,
|
| 89 |
+
"step": 120
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.43,
|
| 93 |
+
"learning_rate": 0.00018095238095238095,
|
| 94 |
+
"loss": 0.2813,
|
| 95 |
+
"step": 130
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.54,
|
| 99 |
+
"learning_rate": 0.0001794871794871795,
|
| 100 |
+
"loss": 0.2686,
|
| 101 |
+
"step": 140
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.65,
|
| 105 |
+
"learning_rate": 0.00017802197802197802,
|
| 106 |
+
"loss": 0.2325,
|
| 107 |
+
"step": 150
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.76,
|
| 111 |
+
"learning_rate": 0.00017655677655677656,
|
| 112 |
+
"loss": 0.3969,
|
| 113 |
+
"step": 160
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.87,
|
| 117 |
+
"learning_rate": 0.0001750915750915751,
|
| 118 |
+
"loss": 0.1645,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.98,
|
| 123 |
+
"learning_rate": 0.00017362637362637365,
|
| 124 |
+
"loss": 0.3043,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 2.09,
|
| 129 |
+
"learning_rate": 0.00017216117216117217,
|
| 130 |
+
"loss": 0.1483,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 2.2,
|
| 135 |
+
"learning_rate": 0.00017069597069597068,
|
| 136 |
+
"loss": 0.1047,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.2,
|
| 141 |
+
"eval_accuracy": 0.7529411764705882,
|
| 142 |
+
"eval_loss": 0.8702762126922607,
|
| 143 |
+
"eval_runtime": 1.4808,
|
| 144 |
+
"eval_samples_per_second": 114.8,
|
| 145 |
+
"eval_steps_per_second": 14.856,
|
| 146 |
+
"step": 200
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 2.31,
|
| 150 |
+
"learning_rate": 0.00016923076923076923,
|
| 151 |
+
"loss": 0.1261,
|
| 152 |
+
"step": 210
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 2.42,
|
| 156 |
+
"learning_rate": 0.00016776556776556777,
|
| 157 |
+
"loss": 0.282,
|
| 158 |
+
"step": 220
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.53,
|
| 162 |
+
"learning_rate": 0.00016630036630036632,
|
| 163 |
+
"loss": 0.1773,
|
| 164 |
+
"step": 230
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.64,
|
| 168 |
+
"learning_rate": 0.00016483516483516484,
|
| 169 |
+
"loss": 0.1566,
|
| 170 |
+
"step": 240
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.75,
|
| 174 |
+
"learning_rate": 0.00016336996336996338,
|
| 175 |
+
"loss": 0.2896,
|
| 176 |
+
"step": 250
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.86,
|
| 180 |
+
"learning_rate": 0.00016190476190476192,
|
| 181 |
+
"loss": 0.2573,
|
| 182 |
+
"step": 260
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.97,
|
| 186 |
+
"learning_rate": 0.00016043956043956044,
|
| 187 |
+
"loss": 0.1943,
|
| 188 |
+
"step": 270
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.08,
|
| 192 |
+
"learning_rate": 0.00015897435897435896,
|
| 193 |
+
"loss": 0.1993,
|
| 194 |
+
"step": 280
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.19,
|
| 198 |
+
"learning_rate": 0.0001575091575091575,
|
| 199 |
+
"loss": 0.1106,
|
| 200 |
+
"step": 290
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 3.3,
|
| 204 |
+
"learning_rate": 0.00015604395604395605,
|
| 205 |
+
"loss": 0.2249,
|
| 206 |
+
"step": 300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.3,
|
| 210 |
+
"eval_accuracy": 0.7588235294117647,
|
| 211 |
+
"eval_loss": 0.9538877606391907,
|
| 212 |
+
"eval_runtime": 1.4679,
|
| 213 |
+
"eval_samples_per_second": 115.811,
|
| 214 |
+
"eval_steps_per_second": 14.987,
|
| 215 |
+
"step": 300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.41,
|
| 219 |
+
"learning_rate": 0.0001545787545787546,
|
| 220 |
+
"loss": 0.2232,
|
| 221 |
+
"step": 310
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 3.52,
|
| 225 |
+
"learning_rate": 0.0001531135531135531,
|
| 226 |
+
"loss": 0.084,
|
| 227 |
+
"step": 320
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 3.63,
|
| 231 |
+
"learning_rate": 0.00015164835164835165,
|
| 232 |
+
"loss": 0.1352,
|
| 233 |
+
"step": 330
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 3.74,
|
| 237 |
+
"learning_rate": 0.0001501831501831502,
|
| 238 |
+
"loss": 0.1632,
|
| 239 |
+
"step": 340
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.85,
|
| 243 |
+
"learning_rate": 0.00014871794871794872,
|
| 244 |
+
"loss": 0.2161,
|
| 245 |
+
"step": 350
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 3.96,
|
| 249 |
+
"learning_rate": 0.00014725274725274726,
|
| 250 |
+
"loss": 0.1823,
|
| 251 |
+
"step": 360
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 4.07,
|
| 255 |
+
"learning_rate": 0.00014578754578754578,
|
| 256 |
+
"loss": 0.0965,
|
| 257 |
+
"step": 370
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 4.18,
|
| 261 |
+
"learning_rate": 0.00014432234432234432,
|
| 262 |
+
"loss": 0.135,
|
| 263 |
+
"step": 380
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 4.29,
|
| 267 |
+
"learning_rate": 0.00014285714285714287,
|
| 268 |
+
"loss": 0.0782,
|
| 269 |
+
"step": 390
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 4.4,
|
| 273 |
+
"learning_rate": 0.0001413919413919414,
|
| 274 |
+
"loss": 0.0984,
|
| 275 |
+
"step": 400
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 4.4,
|
| 279 |
+
"eval_accuracy": 0.7529411764705882,
|
| 280 |
+
"eval_loss": 0.9318849444389343,
|
| 281 |
+
"eval_runtime": 1.4691,
|
| 282 |
+
"eval_samples_per_second": 115.719,
|
| 283 |
+
"eval_steps_per_second": 14.975,
|
| 284 |
+
"step": 400
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 4.51,
|
| 288 |
+
"learning_rate": 0.00013992673992673993,
|
| 289 |
+
"loss": 0.1553,
|
| 290 |
+
"step": 410
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.62,
|
| 294 |
+
"learning_rate": 0.00013846153846153847,
|
| 295 |
+
"loss": 0.0954,
|
| 296 |
+
"step": 420
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 4.73,
|
| 300 |
+
"learning_rate": 0.000136996336996337,
|
| 301 |
+
"loss": 0.1225,
|
| 302 |
+
"step": 430
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 4.84,
|
| 306 |
+
"learning_rate": 0.00013553113553113554,
|
| 307 |
+
"loss": 0.2155,
|
| 308 |
+
"step": 440
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 4.95,
|
| 312 |
+
"learning_rate": 0.00013406593406593405,
|
| 313 |
+
"loss": 0.1496,
|
| 314 |
+
"step": 450
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 5.05,
|
| 318 |
+
"learning_rate": 0.0001326007326007326,
|
| 319 |
+
"loss": 0.0837,
|
| 320 |
+
"step": 460
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 5.16,
|
| 324 |
+
"learning_rate": 0.00013113553113553114,
|
| 325 |
+
"loss": 0.0693,
|
| 326 |
+
"step": 470
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 5.27,
|
| 330 |
+
"learning_rate": 0.0001296703296703297,
|
| 331 |
+
"loss": 0.08,
|
| 332 |
+
"step": 480
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 5.38,
|
| 336 |
+
"learning_rate": 0.00012820512820512823,
|
| 337 |
+
"loss": 0.0989,
|
| 338 |
+
"step": 490
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.49,
|
| 342 |
+
"learning_rate": 0.00012673992673992675,
|
| 343 |
+
"loss": 0.086,
|
| 344 |
+
"step": 500
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 5.49,
|
| 348 |
+
"eval_accuracy": 0.7705882352941177,
|
| 349 |
+
"eval_loss": 0.9061105251312256,
|
| 350 |
+
"eval_runtime": 1.4763,
|
| 351 |
+
"eval_samples_per_second": 115.156,
|
| 352 |
+
"eval_steps_per_second": 14.903,
|
| 353 |
+
"step": 500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 5.6,
|
| 357 |
+
"learning_rate": 0.00012527472527472527,
|
| 358 |
+
"loss": 0.1005,
|
| 359 |
+
"step": 510
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 5.71,
|
| 363 |
+
"learning_rate": 0.0001238095238095238,
|
| 364 |
+
"loss": 0.0685,
|
| 365 |
+
"step": 520
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 5.82,
|
| 369 |
+
"learning_rate": 0.00012234432234432236,
|
| 370 |
+
"loss": 0.1174,
|
| 371 |
+
"step": 530
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 5.93,
|
| 375 |
+
"learning_rate": 0.00012087912087912087,
|
| 376 |
+
"loss": 0.1728,
|
| 377 |
+
"step": 540
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 6.04,
|
| 381 |
+
"learning_rate": 0.00011941391941391942,
|
| 382 |
+
"loss": 0.086,
|
| 383 |
+
"step": 550
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 6.15,
|
| 387 |
+
"learning_rate": 0.00011794871794871796,
|
| 388 |
+
"loss": 0.0738,
|
| 389 |
+
"step": 560
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 6.26,
|
| 393 |
+
"learning_rate": 0.0001164835164835165,
|
| 394 |
+
"loss": 0.04,
|
| 395 |
+
"step": 570
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 6.37,
|
| 399 |
+
"learning_rate": 0.00011501831501831501,
|
| 400 |
+
"loss": 0.1145,
|
| 401 |
+
"step": 580
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 6.48,
|
| 405 |
+
"learning_rate": 0.00011355311355311356,
|
| 406 |
+
"loss": 0.0571,
|
| 407 |
+
"step": 590
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 6.59,
|
| 411 |
+
"learning_rate": 0.0001120879120879121,
|
| 412 |
+
"loss": 0.1164,
|
| 413 |
+
"step": 600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 6.59,
|
| 417 |
+
"eval_accuracy": 0.8176470588235294,
|
| 418 |
+
"eval_loss": 0.7493334412574768,
|
| 419 |
+
"eval_runtime": 1.4541,
|
| 420 |
+
"eval_samples_per_second": 116.908,
|
| 421 |
+
"eval_steps_per_second": 15.129,
|
| 422 |
+
"step": 600
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 6.7,
|
| 426 |
+
"learning_rate": 0.00011062271062271063,
|
| 427 |
+
"loss": 0.0479,
|
| 428 |
+
"step": 610
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 6.81,
|
| 432 |
+
"learning_rate": 0.00010915750915750915,
|
| 433 |
+
"loss": 0.0899,
|
| 434 |
+
"step": 620
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 6.92,
|
| 438 |
+
"learning_rate": 0.0001076923076923077,
|
| 439 |
+
"loss": 0.0746,
|
| 440 |
+
"step": 630
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 7.03,
|
| 444 |
+
"learning_rate": 0.00010622710622710624,
|
| 445 |
+
"loss": 0.0826,
|
| 446 |
+
"step": 640
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 7.14,
|
| 450 |
+
"learning_rate": 0.00010476190476190477,
|
| 451 |
+
"loss": 0.0382,
|
| 452 |
+
"step": 650
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 7.25,
|
| 456 |
+
"learning_rate": 0.00010329670329670331,
|
| 457 |
+
"loss": 0.0724,
|
| 458 |
+
"step": 660
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 7.36,
|
| 462 |
+
"learning_rate": 0.00010183150183150183,
|
| 463 |
+
"loss": 0.036,
|
| 464 |
+
"step": 670
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 7.47,
|
| 468 |
+
"learning_rate": 0.00010036630036630038,
|
| 469 |
+
"loss": 0.0579,
|
| 470 |
+
"step": 680
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 7.58,
|
| 474 |
+
"learning_rate": 9.89010989010989e-05,
|
| 475 |
+
"loss": 0.089,
|
| 476 |
+
"step": 690
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 7.69,
|
| 480 |
+
"learning_rate": 9.743589743589744e-05,
|
| 481 |
+
"loss": 0.0518,
|
| 482 |
+
"step": 700
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 7.69,
|
| 486 |
+
"eval_accuracy": 0.7764705882352941,
|
| 487 |
+
"eval_loss": 0.8781054615974426,
|
| 488 |
+
"eval_runtime": 1.4674,
|
| 489 |
+
"eval_samples_per_second": 115.852,
|
| 490 |
+
"eval_steps_per_second": 14.993,
|
| 491 |
+
"step": 700
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 7.8,
|
| 495 |
+
"learning_rate": 9.597069597069598e-05,
|
| 496 |
+
"loss": 0.0603,
|
| 497 |
+
"step": 710
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 7.91,
|
| 501 |
+
"learning_rate": 9.450549450549451e-05,
|
| 502 |
+
"loss": 0.0565,
|
| 503 |
+
"step": 720
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 8.02,
|
| 507 |
+
"learning_rate": 9.304029304029304e-05,
|
| 508 |
+
"loss": 0.0689,
|
| 509 |
+
"step": 730
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 8.13,
|
| 513 |
+
"learning_rate": 9.157509157509158e-05,
|
| 514 |
+
"loss": 0.0295,
|
| 515 |
+
"step": 740
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 8.24,
|
| 519 |
+
"learning_rate": 9.010989010989012e-05,
|
| 520 |
+
"loss": 0.0555,
|
| 521 |
+
"step": 750
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 8.35,
|
| 525 |
+
"learning_rate": 8.864468864468865e-05,
|
| 526 |
+
"loss": 0.0501,
|
| 527 |
+
"step": 760
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 8.46,
|
| 531 |
+
"learning_rate": 8.717948717948718e-05,
|
| 532 |
+
"loss": 0.0657,
|
| 533 |
+
"step": 770
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 8.57,
|
| 537 |
+
"learning_rate": 8.571428571428571e-05,
|
| 538 |
+
"loss": 0.0479,
|
| 539 |
+
"step": 780
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 8.68,
|
| 543 |
+
"learning_rate": 8.424908424908426e-05,
|
| 544 |
+
"loss": 0.0665,
|
| 545 |
+
"step": 790
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 8.79,
|
| 549 |
+
"learning_rate": 8.278388278388279e-05,
|
| 550 |
+
"loss": 0.0458,
|
| 551 |
+
"step": 800
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 8.79,
|
| 555 |
+
"eval_accuracy": 0.7823529411764706,
|
| 556 |
+
"eval_loss": 0.8851077556610107,
|
| 557 |
+
"eval_runtime": 1.4711,
|
| 558 |
+
"eval_samples_per_second": 115.561,
|
| 559 |
+
"eval_steps_per_second": 14.955,
|
| 560 |
+
"step": 800
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 8.9,
|
| 564 |
+
"learning_rate": 8.131868131868132e-05,
|
| 565 |
+
"loss": 0.0704,
|
| 566 |
+
"step": 810
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 9.01,
|
| 570 |
+
"learning_rate": 7.985347985347986e-05,
|
| 571 |
+
"loss": 0.0779,
|
| 572 |
+
"step": 820
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 9.12,
|
| 576 |
+
"learning_rate": 7.83882783882784e-05,
|
| 577 |
+
"loss": 0.0575,
|
| 578 |
+
"step": 830
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 9.23,
|
| 582 |
+
"learning_rate": 7.692307692307693e-05,
|
| 583 |
+
"loss": 0.0541,
|
| 584 |
+
"step": 840
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 9.34,
|
| 588 |
+
"learning_rate": 7.545787545787546e-05,
|
| 589 |
+
"loss": 0.0413,
|
| 590 |
+
"step": 850
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 9.45,
|
| 594 |
+
"learning_rate": 7.3992673992674e-05,
|
| 595 |
+
"loss": 0.0367,
|
| 596 |
+
"step": 860
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 9.56,
|
| 600 |
+
"learning_rate": 7.252747252747253e-05,
|
| 601 |
+
"loss": 0.0857,
|
| 602 |
+
"step": 870
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 9.67,
|
| 606 |
+
"learning_rate": 7.106227106227106e-05,
|
| 607 |
+
"loss": 0.061,
|
| 608 |
+
"step": 880
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 9.78,
|
| 612 |
+
"learning_rate": 6.95970695970696e-05,
|
| 613 |
+
"loss": 0.0408,
|
| 614 |
+
"step": 890
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 9.89,
|
| 618 |
+
"learning_rate": 6.813186813186814e-05,
|
| 619 |
+
"loss": 0.0521,
|
| 620 |
+
"step": 900
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 9.89,
|
| 624 |
+
"eval_accuracy": 0.788235294117647,
|
| 625 |
+
"eval_loss": 0.9448409080505371,
|
| 626 |
+
"eval_runtime": 1.3677,
|
| 627 |
+
"eval_samples_per_second": 124.3,
|
| 628 |
+
"eval_steps_per_second": 16.086,
|
| 629 |
+
"step": 900
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 10.0,
|
| 633 |
+
"learning_rate": 6.666666666666667e-05,
|
| 634 |
+
"loss": 0.0452,
|
| 635 |
+
"step": 910
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 10.11,
|
| 639 |
+
"learning_rate": 6.52014652014652e-05,
|
| 640 |
+
"loss": 0.0573,
|
| 641 |
+
"step": 920
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 10.22,
|
| 645 |
+
"learning_rate": 6.373626373626373e-05,
|
| 646 |
+
"loss": 0.0333,
|
| 647 |
+
"step": 930
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 10.33,
|
| 651 |
+
"learning_rate": 6.227106227106228e-05,
|
| 652 |
+
"loss": 0.035,
|
| 653 |
+
"step": 940
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 10.44,
|
| 657 |
+
"learning_rate": 6.08058608058608e-05,
|
| 658 |
+
"loss": 0.0513,
|
| 659 |
+
"step": 950
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 10.55,
|
| 663 |
+
"learning_rate": 5.9340659340659345e-05,
|
| 664 |
+
"loss": 0.0412,
|
| 665 |
+
"step": 960
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 10.66,
|
| 669 |
+
"learning_rate": 5.787545787545788e-05,
|
| 670 |
+
"loss": 0.044,
|
| 671 |
+
"step": 970
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 10.77,
|
| 675 |
+
"learning_rate": 5.6410256410256414e-05,
|
| 676 |
+
"loss": 0.0559,
|
| 677 |
+
"step": 980
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 10.88,
|
| 681 |
+
"learning_rate": 5.494505494505495e-05,
|
| 682 |
+
"loss": 0.0889,
|
| 683 |
+
"step": 990
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 10.99,
|
| 687 |
+
"learning_rate": 5.347985347985348e-05,
|
| 688 |
+
"loss": 0.0576,
|
| 689 |
+
"step": 1000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 10.99,
|
| 693 |
+
"eval_accuracy": 0.7823529411764706,
|
| 694 |
+
"eval_loss": 0.8883835673332214,
|
| 695 |
+
"eval_runtime": 1.3691,
|
| 696 |
+
"eval_samples_per_second": 124.171,
|
| 697 |
+
"eval_steps_per_second": 16.069,
|
| 698 |
+
"step": 1000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 11.1,
|
| 702 |
+
"learning_rate": 5.201465201465202e-05,
|
| 703 |
+
"loss": 0.0591,
|
| 704 |
+
"step": 1010
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 11.21,
|
| 708 |
+
"learning_rate": 5.054945054945055e-05,
|
| 709 |
+
"loss": 0.0649,
|
| 710 |
+
"step": 1020
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 11.32,
|
| 714 |
+
"learning_rate": 4.908424908424908e-05,
|
| 715 |
+
"loss": 0.0425,
|
| 716 |
+
"step": 1030
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 11.43,
|
| 720 |
+
"learning_rate": 4.761904761904762e-05,
|
| 721 |
+
"loss": 0.0471,
|
| 722 |
+
"step": 1040
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 11.54,
|
| 726 |
+
"learning_rate": 4.615384615384616e-05,
|
| 727 |
+
"loss": 0.0366,
|
| 728 |
+
"step": 1050
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 11.65,
|
| 732 |
+
"learning_rate": 4.4688644688644696e-05,
|
| 733 |
+
"loss": 0.0414,
|
| 734 |
+
"step": 1060
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 11.76,
|
| 738 |
+
"learning_rate": 4.322344322344323e-05,
|
| 739 |
+
"loss": 0.0393,
|
| 740 |
+
"step": 1070
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 11.87,
|
| 744 |
+
"learning_rate": 4.1758241758241765e-05,
|
| 745 |
+
"loss": 0.0551,
|
| 746 |
+
"step": 1080
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 11.98,
|
| 750 |
+
"learning_rate": 4.0293040293040296e-05,
|
| 751 |
+
"loss": 0.0503,
|
| 752 |
+
"step": 1090
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 12.09,
|
| 756 |
+
"learning_rate": 3.8827838827838833e-05,
|
| 757 |
+
"loss": 0.0442,
|
| 758 |
+
"step": 1100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 12.09,
|
| 762 |
+
"eval_accuracy": 0.788235294117647,
|
| 763 |
+
"eval_loss": 0.896532416343689,
|
| 764 |
+
"eval_runtime": 1.5128,
|
| 765 |
+
"eval_samples_per_second": 112.377,
|
| 766 |
+
"eval_steps_per_second": 14.543,
|
| 767 |
+
"step": 1100
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 12.2,
|
| 771 |
+
"learning_rate": 3.7362637362637365e-05,
|
| 772 |
+
"loss": 0.0595,
|
| 773 |
+
"step": 1110
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 12.31,
|
| 777 |
+
"learning_rate": 3.58974358974359e-05,
|
| 778 |
+
"loss": 0.0304,
|
| 779 |
+
"step": 1120
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 12.42,
|
| 783 |
+
"learning_rate": 3.443223443223443e-05,
|
| 784 |
+
"loss": 0.0365,
|
| 785 |
+
"step": 1130
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 12.53,
|
| 789 |
+
"learning_rate": 3.296703296703297e-05,
|
| 790 |
+
"loss": 0.0917,
|
| 791 |
+
"step": 1140
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 12.64,
|
| 795 |
+
"learning_rate": 3.15018315018315e-05,
|
| 796 |
+
"loss": 0.0549,
|
| 797 |
+
"step": 1150
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 12.75,
|
| 801 |
+
"learning_rate": 3.0036630036630036e-05,
|
| 802 |
+
"loss": 0.0616,
|
| 803 |
+
"step": 1160
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 12.86,
|
| 807 |
+
"learning_rate": 2.857142857142857e-05,
|
| 808 |
+
"loss": 0.0526,
|
| 809 |
+
"step": 1170
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 12.97,
|
| 813 |
+
"learning_rate": 2.7106227106227105e-05,
|
| 814 |
+
"loss": 0.0297,
|
| 815 |
+
"step": 1180
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 13.08,
|
| 819 |
+
"learning_rate": 2.564102564102564e-05,
|
| 820 |
+
"loss": 0.0409,
|
| 821 |
+
"step": 1190
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 13.19,
|
| 825 |
+
"learning_rate": 2.4175824175824177e-05,
|
| 826 |
+
"loss": 0.0254,
|
| 827 |
+
"step": 1200
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 13.19,
|
| 831 |
+
"eval_accuracy": 0.788235294117647,
|
| 832 |
+
"eval_loss": 0.9140186905860901,
|
| 833 |
+
"eval_runtime": 1.3923,
|
| 834 |
+
"eval_samples_per_second": 122.1,
|
| 835 |
+
"eval_steps_per_second": 15.801,
|
| 836 |
+
"step": 1200
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 13.3,
|
| 840 |
+
"learning_rate": 2.2710622710622712e-05,
|
| 841 |
+
"loss": 0.0373,
|
| 842 |
+
"step": 1210
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 13.41,
|
| 846 |
+
"learning_rate": 2.1245421245421246e-05,
|
| 847 |
+
"loss": 0.0325,
|
| 848 |
+
"step": 1220
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 13.52,
|
| 852 |
+
"learning_rate": 1.978021978021978e-05,
|
| 853 |
+
"loss": 0.0373,
|
| 854 |
+
"step": 1230
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 13.63,
|
| 858 |
+
"learning_rate": 1.8315018315018315e-05,
|
| 859 |
+
"loss": 0.0545,
|
| 860 |
+
"step": 1240
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 13.74,
|
| 864 |
+
"learning_rate": 1.6849816849816853e-05,
|
| 865 |
+
"loss": 0.0613,
|
| 866 |
+
"step": 1250
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 13.85,
|
| 870 |
+
"learning_rate": 1.5384615384615387e-05,
|
| 871 |
+
"loss": 0.0515,
|
| 872 |
+
"step": 1260
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 13.96,
|
| 876 |
+
"learning_rate": 1.391941391941392e-05,
|
| 877 |
+
"loss": 0.0708,
|
| 878 |
+
"step": 1270
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 14.07,
|
| 882 |
+
"learning_rate": 1.2454212454212454e-05,
|
| 883 |
+
"loss": 0.0615,
|
| 884 |
+
"step": 1280
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 14.18,
|
| 888 |
+
"learning_rate": 1.0989010989010989e-05,
|
| 889 |
+
"loss": 0.0495,
|
| 890 |
+
"step": 1290
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 14.29,
|
| 894 |
+
"learning_rate": 9.523809523809523e-06,
|
| 895 |
+
"loss": 0.0426,
|
| 896 |
+
"step": 1300
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 14.29,
|
| 900 |
+
"eval_accuracy": 0.788235294117647,
|
| 901 |
+
"eval_loss": 0.927388072013855,
|
| 902 |
+
"eval_runtime": 1.3816,
|
| 903 |
+
"eval_samples_per_second": 123.046,
|
| 904 |
+
"eval_steps_per_second": 15.924,
|
| 905 |
+
"step": 1300
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 14.4,
|
| 909 |
+
"learning_rate": 8.058608058608059e-06,
|
| 910 |
+
"loss": 0.0437,
|
| 911 |
+
"step": 1310
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 14.51,
|
| 915 |
+
"learning_rate": 6.5934065934065935e-06,
|
| 916 |
+
"loss": 0.028,
|
| 917 |
+
"step": 1320
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 14.62,
|
| 921 |
+
"learning_rate": 5.128205128205128e-06,
|
| 922 |
+
"loss": 0.0345,
|
| 923 |
+
"step": 1330
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 14.73,
|
| 927 |
+
"learning_rate": 3.663003663003663e-06,
|
| 928 |
+
"loss": 0.0565,
|
| 929 |
+
"step": 1340
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 14.84,
|
| 933 |
+
"learning_rate": 2.197802197802198e-06,
|
| 934 |
+
"loss": 0.0758,
|
| 935 |
+
"step": 1350
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 14.95,
|
| 939 |
+
"learning_rate": 7.326007326007326e-07,
|
| 940 |
+
"loss": 0.047,
|
| 941 |
+
"step": 1360
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 15.0,
|
| 945 |
+
"step": 1365,
|
| 946 |
+
"total_flos": 1.6866433871220326e+18,
|
| 947 |
+
"train_loss": 0.11297949704276773,
|
| 948 |
+
"train_runtime": 502.5996,
|
| 949 |
+
"train_samples_per_second": 43.305,
|
| 950 |
+
"train_steps_per_second": 2.716
|
| 951 |
+
}
|
| 952 |
+
],
|
| 953 |
+
"max_steps": 1365,
|
| 954 |
+
"num_train_epochs": 15,
|
| 955 |
+
"total_flos": 1.6866433871220326e+18,
|
| 956 |
+
"trial_name": null,
|
| 957 |
+
"trial_params": null
|
| 958 |
+
}
|