All Dunn!!!
Browse files- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +1090 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 1.1316755048115336e+16,
|
| 4 |
+
"train_loss": 0.19905866315323814,
|
| 5 |
+
"train_runtime": 6850.4631,
|
| 6 |
+
"train_samples_per_second": 40.041,
|
| 7 |
+
"train_steps_per_second": 5.006
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 1.1316755048115336e+16,
|
| 4 |
+
"train_loss": 0.19905866315323814,
|
| 5 |
+
"train_runtime": 6850.4631,
|
| 6 |
+
"train_samples_per_second": 40.041,
|
| 7 |
+
"train_steps_per_second": 5.006
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1090 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3088998794555664,
|
| 3 |
+
"best_model_checkpoint": "bert-base-uncased-Regression-Edmunds_Car_Reviews/checkpoint-34290",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"global_step": 34290,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 1.999941673957422e-05,
|
| 13 |
+
"loss": 16.5207,
|
| 14 |
+
"step": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"learning_rate": 1.988334791484398e-05,
|
| 19 |
+
"loss": 1.4237,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.03,
|
| 24 |
+
"learning_rate": 1.9766695829687957e-05,
|
| 25 |
+
"loss": 0.3652,
|
| 26 |
+
"step": 400
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.05,
|
| 30 |
+
"learning_rate": 1.9650043744531936e-05,
|
| 31 |
+
"loss": 0.3355,
|
| 32 |
+
"step": 600
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.07,
|
| 36 |
+
"learning_rate": 1.9533391659375912e-05,
|
| 37 |
+
"loss": 0.2945,
|
| 38 |
+
"step": 800
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.09,
|
| 42 |
+
"learning_rate": 1.941673957421989e-05,
|
| 43 |
+
"loss": 0.2863,
|
| 44 |
+
"step": 1000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.1,
|
| 48 |
+
"learning_rate": 1.9300087489063867e-05,
|
| 49 |
+
"loss": 0.3029,
|
| 50 |
+
"step": 1200
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.12,
|
| 54 |
+
"learning_rate": 1.9183435403907846e-05,
|
| 55 |
+
"loss": 0.3111,
|
| 56 |
+
"step": 1400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.14,
|
| 60 |
+
"learning_rate": 1.9066783318751826e-05,
|
| 61 |
+
"loss": 0.2634,
|
| 62 |
+
"step": 1600
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.16,
|
| 66 |
+
"learning_rate": 1.89501312335958e-05,
|
| 67 |
+
"loss": 0.278,
|
| 68 |
+
"step": 1800
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.17,
|
| 72 |
+
"learning_rate": 1.883347914843978e-05,
|
| 73 |
+
"loss": 0.2729,
|
| 74 |
+
"step": 2000
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.19,
|
| 78 |
+
"learning_rate": 1.8716827063283757e-05,
|
| 79 |
+
"loss": 0.2829,
|
| 80 |
+
"step": 2200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.21,
|
| 84 |
+
"learning_rate": 1.8600174978127736e-05,
|
| 85 |
+
"loss": 0.2461,
|
| 86 |
+
"step": 2400
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.23,
|
| 90 |
+
"learning_rate": 1.8483522892971712e-05,
|
| 91 |
+
"loss": 0.2497,
|
| 92 |
+
"step": 2600
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.24,
|
| 96 |
+
"learning_rate": 1.836687080781569e-05,
|
| 97 |
+
"loss": 0.2592,
|
| 98 |
+
"step": 2800
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.26,
|
| 102 |
+
"learning_rate": 1.825021872265967e-05,
|
| 103 |
+
"loss": 0.2701,
|
| 104 |
+
"step": 3000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.28,
|
| 108 |
+
"learning_rate": 1.813356663750365e-05,
|
| 109 |
+
"loss": 0.2699,
|
| 110 |
+
"step": 3200
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.3,
|
| 114 |
+
"learning_rate": 1.8016914552347625e-05,
|
| 115 |
+
"loss": 0.2656,
|
| 116 |
+
"step": 3400
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.31,
|
| 120 |
+
"learning_rate": 1.79002624671916e-05,
|
| 121 |
+
"loss": 0.2521,
|
| 122 |
+
"step": 3600
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.33,
|
| 126 |
+
"learning_rate": 1.778361038203558e-05,
|
| 127 |
+
"loss": 0.2776,
|
| 128 |
+
"step": 3800
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.35,
|
| 132 |
+
"learning_rate": 1.7666958296879556e-05,
|
| 133 |
+
"loss": 0.2753,
|
| 134 |
+
"step": 4000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.37,
|
| 138 |
+
"learning_rate": 1.7550306211723536e-05,
|
| 139 |
+
"loss": 0.2268,
|
| 140 |
+
"step": 4200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.38,
|
| 144 |
+
"learning_rate": 1.7433654126567515e-05,
|
| 145 |
+
"loss": 0.2617,
|
| 146 |
+
"step": 4400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.4,
|
| 150 |
+
"learning_rate": 1.7317002041411494e-05,
|
| 151 |
+
"loss": 0.2425,
|
| 152 |
+
"step": 4600
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.42,
|
| 156 |
+
"learning_rate": 1.720034995625547e-05,
|
| 157 |
+
"loss": 0.2474,
|
| 158 |
+
"step": 4800
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.44,
|
| 162 |
+
"learning_rate": 1.7083697871099446e-05,
|
| 163 |
+
"loss": 0.2589,
|
| 164 |
+
"step": 5000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.45,
|
| 168 |
+
"learning_rate": 1.6967045785943425e-05,
|
| 169 |
+
"loss": 0.2389,
|
| 170 |
+
"step": 5200
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.47,
|
| 174 |
+
"learning_rate": 1.68503937007874e-05,
|
| 175 |
+
"loss": 0.2494,
|
| 176 |
+
"step": 5400
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.49,
|
| 180 |
+
"learning_rate": 1.673374161563138e-05,
|
| 181 |
+
"loss": 0.2558,
|
| 182 |
+
"step": 5600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.51,
|
| 186 |
+
"learning_rate": 1.661708953047536e-05,
|
| 187 |
+
"loss": 0.2524,
|
| 188 |
+
"step": 5800
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.52,
|
| 192 |
+
"learning_rate": 1.650043744531934e-05,
|
| 193 |
+
"loss": 0.2509,
|
| 194 |
+
"step": 6000
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.54,
|
| 198 |
+
"learning_rate": 1.6383785360163315e-05,
|
| 199 |
+
"loss": 0.2605,
|
| 200 |
+
"step": 6200
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.56,
|
| 204 |
+
"learning_rate": 1.626713327500729e-05,
|
| 205 |
+
"loss": 0.2354,
|
| 206 |
+
"step": 6400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.58,
|
| 210 |
+
"learning_rate": 1.615048118985127e-05,
|
| 211 |
+
"loss": 0.27,
|
| 212 |
+
"step": 6600
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.59,
|
| 216 |
+
"learning_rate": 1.603382910469525e-05,
|
| 217 |
+
"loss": 0.2453,
|
| 218 |
+
"step": 6800
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.61,
|
| 222 |
+
"learning_rate": 1.5917177019539225e-05,
|
| 223 |
+
"loss": 0.2589,
|
| 224 |
+
"step": 7000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.63,
|
| 228 |
+
"learning_rate": 1.5800524934383204e-05,
|
| 229 |
+
"loss": 0.2513,
|
| 230 |
+
"step": 7200
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.65,
|
| 234 |
+
"learning_rate": 1.5683872849227184e-05,
|
| 235 |
+
"loss": 0.2583,
|
| 236 |
+
"step": 7400
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.66,
|
| 240 |
+
"learning_rate": 1.556722076407116e-05,
|
| 241 |
+
"loss": 0.257,
|
| 242 |
+
"step": 7600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.68,
|
| 246 |
+
"learning_rate": 1.5450568678915135e-05,
|
| 247 |
+
"loss": 0.2363,
|
| 248 |
+
"step": 7800
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.7,
|
| 252 |
+
"learning_rate": 1.5333916593759115e-05,
|
| 253 |
+
"loss": 0.2378,
|
| 254 |
+
"step": 8000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.72,
|
| 258 |
+
"learning_rate": 1.5217264508603094e-05,
|
| 259 |
+
"loss": 0.2438,
|
| 260 |
+
"step": 8200
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.73,
|
| 264 |
+
"learning_rate": 1.510061242344707e-05,
|
| 265 |
+
"loss": 0.2489,
|
| 266 |
+
"step": 8400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.75,
|
| 270 |
+
"learning_rate": 1.4983960338291049e-05,
|
| 271 |
+
"loss": 0.2618,
|
| 272 |
+
"step": 8600
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.77,
|
| 276 |
+
"learning_rate": 1.4867308253135026e-05,
|
| 277 |
+
"loss": 0.2264,
|
| 278 |
+
"step": 8800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.79,
|
| 282 |
+
"learning_rate": 1.4750656167979002e-05,
|
| 283 |
+
"loss": 0.2595,
|
| 284 |
+
"step": 9000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.8,
|
| 288 |
+
"learning_rate": 1.4634004082822982e-05,
|
| 289 |
+
"loss": 0.2246,
|
| 290 |
+
"step": 9200
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.82,
|
| 294 |
+
"learning_rate": 1.451735199766696e-05,
|
| 295 |
+
"loss": 0.2281,
|
| 296 |
+
"step": 9400
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.84,
|
| 300 |
+
"learning_rate": 1.4400699912510938e-05,
|
| 301 |
+
"loss": 0.217,
|
| 302 |
+
"step": 9600
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.86,
|
| 306 |
+
"learning_rate": 1.4284047827354914e-05,
|
| 307 |
+
"loss": 0.2359,
|
| 308 |
+
"step": 9800
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.87,
|
| 312 |
+
"learning_rate": 1.4167395742198894e-05,
|
| 313 |
+
"loss": 0.2322,
|
| 314 |
+
"step": 10000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.89,
|
| 318 |
+
"learning_rate": 1.4050743657042871e-05,
|
| 319 |
+
"loss": 0.2055,
|
| 320 |
+
"step": 10200
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.91,
|
| 324 |
+
"learning_rate": 1.393409157188685e-05,
|
| 325 |
+
"loss": 0.2374,
|
| 326 |
+
"step": 10400
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.93,
|
| 330 |
+
"learning_rate": 1.3817439486730826e-05,
|
| 331 |
+
"loss": 0.2386,
|
| 332 |
+
"step": 10600
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.94,
|
| 336 |
+
"learning_rate": 1.3700787401574804e-05,
|
| 337 |
+
"loss": 0.2323,
|
| 338 |
+
"step": 10800
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.96,
|
| 342 |
+
"learning_rate": 1.3584135316418783e-05,
|
| 343 |
+
"loss": 0.2362,
|
| 344 |
+
"step": 11000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.98,
|
| 348 |
+
"learning_rate": 1.3467483231262759e-05,
|
| 349 |
+
"loss": 0.2211,
|
| 350 |
+
"step": 11200
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.0,
|
| 354 |
+
"learning_rate": 1.3350831146106738e-05,
|
| 355 |
+
"loss": 0.2142,
|
| 356 |
+
"step": 11400
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 1.0,
|
| 360 |
+
"eval_loss": 0.24207346141338348,
|
| 361 |
+
"eval_mae": 0.3126428425312042,
|
| 362 |
+
"eval_mse": 0.24207346141338348,
|
| 363 |
+
"eval_rmse": 0.49200960993766785,
|
| 364 |
+
"eval_runtime": 434.9139,
|
| 365 |
+
"eval_samples_per_second": 113.204,
|
| 366 |
+
"eval_steps_per_second": 14.152,
|
| 367 |
+
"step": 11430
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 1.01,
|
| 371 |
+
"learning_rate": 1.3234179060950716e-05,
|
| 372 |
+
"loss": 0.194,
|
| 373 |
+
"step": 11600
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 1.03,
|
| 377 |
+
"learning_rate": 1.3117526975794695e-05,
|
| 378 |
+
"loss": 0.1782,
|
| 379 |
+
"step": 11800
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 1.05,
|
| 383 |
+
"learning_rate": 1.3000874890638671e-05,
|
| 384 |
+
"loss": 0.1992,
|
| 385 |
+
"step": 12000
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 1.07,
|
| 389 |
+
"learning_rate": 1.2884222805482649e-05,
|
| 390 |
+
"loss": 0.1657,
|
| 391 |
+
"step": 12200
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 1.08,
|
| 395 |
+
"learning_rate": 1.2767570720326628e-05,
|
| 396 |
+
"loss": 0.2156,
|
| 397 |
+
"step": 12400
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.1,
|
| 401 |
+
"learning_rate": 1.2650918635170604e-05,
|
| 402 |
+
"loss": 0.1742,
|
| 403 |
+
"step": 12600
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 1.12,
|
| 407 |
+
"learning_rate": 1.2534266550014583e-05,
|
| 408 |
+
"loss": 0.156,
|
| 409 |
+
"step": 12800
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.14,
|
| 413 |
+
"learning_rate": 1.241761446485856e-05,
|
| 414 |
+
"loss": 0.2079,
|
| 415 |
+
"step": 13000
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.15,
|
| 419 |
+
"learning_rate": 1.230096237970254e-05,
|
| 420 |
+
"loss": 0.1635,
|
| 421 |
+
"step": 13200
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 1.17,
|
| 425 |
+
"learning_rate": 1.2184310294546516e-05,
|
| 426 |
+
"loss": 0.1987,
|
| 427 |
+
"step": 13400
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.19,
|
| 431 |
+
"learning_rate": 1.2067658209390493e-05,
|
| 432 |
+
"loss": 0.1683,
|
| 433 |
+
"step": 13600
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 1.21,
|
| 437 |
+
"learning_rate": 1.1951006124234472e-05,
|
| 438 |
+
"loss": 0.1735,
|
| 439 |
+
"step": 13800
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 1.22,
|
| 443 |
+
"learning_rate": 1.183435403907845e-05,
|
| 444 |
+
"loss": 0.1971,
|
| 445 |
+
"step": 14000
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.24,
|
| 449 |
+
"learning_rate": 1.1717701953922428e-05,
|
| 450 |
+
"loss": 0.2014,
|
| 451 |
+
"step": 14200
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.26,
|
| 455 |
+
"learning_rate": 1.1601049868766405e-05,
|
| 456 |
+
"loss": 0.2114,
|
| 457 |
+
"step": 14400
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.28,
|
| 461 |
+
"learning_rate": 1.1484397783610384e-05,
|
| 462 |
+
"loss": 0.1707,
|
| 463 |
+
"step": 14600
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.29,
|
| 467 |
+
"learning_rate": 1.136774569845436e-05,
|
| 468 |
+
"loss": 0.194,
|
| 469 |
+
"step": 14800
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 1.31,
|
| 473 |
+
"learning_rate": 1.1251093613298338e-05,
|
| 474 |
+
"loss": 0.1714,
|
| 475 |
+
"step": 15000
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.33,
|
| 479 |
+
"learning_rate": 1.1134441528142317e-05,
|
| 480 |
+
"loss": 0.1973,
|
| 481 |
+
"step": 15200
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 1.35,
|
| 485 |
+
"learning_rate": 1.1017789442986295e-05,
|
| 486 |
+
"loss": 0.1705,
|
| 487 |
+
"step": 15400
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.36,
|
| 491 |
+
"learning_rate": 1.0901137357830272e-05,
|
| 492 |
+
"loss": 0.1842,
|
| 493 |
+
"step": 15600
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.38,
|
| 497 |
+
"learning_rate": 1.078448527267425e-05,
|
| 498 |
+
"loss": 0.1578,
|
| 499 |
+
"step": 15800
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.4,
|
| 503 |
+
"learning_rate": 1.0667833187518229e-05,
|
| 504 |
+
"loss": 0.1598,
|
| 505 |
+
"step": 16000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 1.42,
|
| 509 |
+
"learning_rate": 1.0551181102362205e-05,
|
| 510 |
+
"loss": 0.1895,
|
| 511 |
+
"step": 16200
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 1.43,
|
| 515 |
+
"learning_rate": 1.0434529017206183e-05,
|
| 516 |
+
"loss": 0.1846,
|
| 517 |
+
"step": 16400
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 1.45,
|
| 521 |
+
"learning_rate": 1.0317876932050162e-05,
|
| 522 |
+
"loss": 0.1875,
|
| 523 |
+
"step": 16600
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 1.47,
|
| 527 |
+
"learning_rate": 1.020122484689414e-05,
|
| 528 |
+
"loss": 0.1748,
|
| 529 |
+
"step": 16800
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.49,
|
| 533 |
+
"learning_rate": 1.0084572761738117e-05,
|
| 534 |
+
"loss": 0.1837,
|
| 535 |
+
"step": 17000
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.5,
|
| 539 |
+
"learning_rate": 9.967920676582095e-06,
|
| 540 |
+
"loss": 0.1888,
|
| 541 |
+
"step": 17200
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 1.52,
|
| 545 |
+
"learning_rate": 9.851268591426074e-06,
|
| 546 |
+
"loss": 0.2038,
|
| 547 |
+
"step": 17400
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 1.54,
|
| 551 |
+
"learning_rate": 9.73461650627005e-06,
|
| 552 |
+
"loss": 0.1736,
|
| 553 |
+
"step": 17600
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 1.56,
|
| 557 |
+
"learning_rate": 9.617964421114027e-06,
|
| 558 |
+
"loss": 0.1853,
|
| 559 |
+
"step": 17800
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.57,
|
| 563 |
+
"learning_rate": 9.501312335958006e-06,
|
| 564 |
+
"loss": 0.1762,
|
| 565 |
+
"step": 18000
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 1.59,
|
| 569 |
+
"learning_rate": 9.384660250801984e-06,
|
| 570 |
+
"loss": 0.1979,
|
| 571 |
+
"step": 18200
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.61,
|
| 575 |
+
"learning_rate": 9.268008165645962e-06,
|
| 576 |
+
"loss": 0.1747,
|
| 577 |
+
"step": 18400
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.63,
|
| 581 |
+
"learning_rate": 9.15135608048994e-06,
|
| 582 |
+
"loss": 0.1793,
|
| 583 |
+
"step": 18600
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 1.64,
|
| 587 |
+
"learning_rate": 9.034703995333918e-06,
|
| 588 |
+
"loss": 0.1788,
|
| 589 |
+
"step": 18800
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.66,
|
| 593 |
+
"learning_rate": 8.918051910177894e-06,
|
| 594 |
+
"loss": 0.1901,
|
| 595 |
+
"step": 19000
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 1.68,
|
| 599 |
+
"learning_rate": 8.801399825021874e-06,
|
| 600 |
+
"loss": 0.1755,
|
| 601 |
+
"step": 19200
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.7,
|
| 605 |
+
"learning_rate": 8.684747739865851e-06,
|
| 606 |
+
"loss": 0.1782,
|
| 607 |
+
"step": 19400
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.71,
|
| 611 |
+
"learning_rate": 8.568095654709829e-06,
|
| 612 |
+
"loss": 0.1928,
|
| 613 |
+
"step": 19600
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.73,
|
| 617 |
+
"learning_rate": 8.451443569553806e-06,
|
| 618 |
+
"loss": 0.1821,
|
| 619 |
+
"step": 19800
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.75,
|
| 623 |
+
"learning_rate": 8.334791484397784e-06,
|
| 624 |
+
"loss": 0.1836,
|
| 625 |
+
"step": 20000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 1.77,
|
| 629 |
+
"learning_rate": 8.218139399241763e-06,
|
| 630 |
+
"loss": 0.1803,
|
| 631 |
+
"step": 20200
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 1.78,
|
| 635 |
+
"learning_rate": 8.101487314085739e-06,
|
| 636 |
+
"loss": 0.176,
|
| 637 |
+
"step": 20400
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.8,
|
| 641 |
+
"learning_rate": 7.984835228929718e-06,
|
| 642 |
+
"loss": 0.2068,
|
| 643 |
+
"step": 20600
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.82,
|
| 647 |
+
"learning_rate": 7.868183143773696e-06,
|
| 648 |
+
"loss": 0.2005,
|
| 649 |
+
"step": 20800
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 1.84,
|
| 653 |
+
"learning_rate": 7.751531058617673e-06,
|
| 654 |
+
"loss": 0.1706,
|
| 655 |
+
"step": 21000
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.85,
|
| 659 |
+
"learning_rate": 7.634878973461651e-06,
|
| 660 |
+
"loss": 0.1821,
|
| 661 |
+
"step": 21200
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.87,
|
| 665 |
+
"learning_rate": 7.5182268883056285e-06,
|
| 666 |
+
"loss": 0.2003,
|
| 667 |
+
"step": 21400
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 1.89,
|
| 671 |
+
"learning_rate": 7.401574803149607e-06,
|
| 672 |
+
"loss": 0.1839,
|
| 673 |
+
"step": 21600
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 1.91,
|
| 677 |
+
"learning_rate": 7.2849227179935845e-06,
|
| 678 |
+
"loss": 0.191,
|
| 679 |
+
"step": 21800
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.92,
|
| 683 |
+
"learning_rate": 7.168270632837563e-06,
|
| 684 |
+
"loss": 0.2031,
|
| 685 |
+
"step": 22000
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.94,
|
| 689 |
+
"learning_rate": 7.0516185476815405e-06,
|
| 690 |
+
"loss": 0.2054,
|
| 691 |
+
"step": 22200
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 1.96,
|
| 695 |
+
"learning_rate": 6.934966462525519e-06,
|
| 696 |
+
"loss": 0.1687,
|
| 697 |
+
"step": 22400
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.98,
|
| 701 |
+
"learning_rate": 6.818314377369496e-06,
|
| 702 |
+
"loss": 0.176,
|
| 703 |
+
"step": 22600
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.99,
|
| 707 |
+
"learning_rate": 6.701662292213474e-06,
|
| 708 |
+
"loss": 0.1931,
|
| 709 |
+
"step": 22800
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 2.0,
|
| 713 |
+
"eval_loss": 0.2529958486557007,
|
| 714 |
+
"eval_mae": 0.3336344063282013,
|
| 715 |
+
"eval_mse": 0.2529958486557007,
|
| 716 |
+
"eval_rmse": 0.5029869079589844,
|
| 717 |
+
"eval_runtime": 433.6838,
|
| 718 |
+
"eval_samples_per_second": 113.525,
|
| 719 |
+
"eval_steps_per_second": 14.192,
|
| 720 |
+
"step": 22860
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 2.01,
|
| 724 |
+
"learning_rate": 6.585010207057452e-06,
|
| 725 |
+
"loss": 0.1951,
|
| 726 |
+
"step": 23000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 2.03,
|
| 730 |
+
"learning_rate": 6.468358121901429e-06,
|
| 731 |
+
"loss": 0.127,
|
| 732 |
+
"step": 23200
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 2.05,
|
| 736 |
+
"learning_rate": 6.351706036745408e-06,
|
| 737 |
+
"loss": 0.1387,
|
| 738 |
+
"step": 23400
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 2.06,
|
| 742 |
+
"learning_rate": 6.235053951589385e-06,
|
| 743 |
+
"loss": 0.1436,
|
| 744 |
+
"step": 23600
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 2.08,
|
| 748 |
+
"learning_rate": 6.118401866433364e-06,
|
| 749 |
+
"loss": 0.1446,
|
| 750 |
+
"step": 23800
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 2.1,
|
| 754 |
+
"learning_rate": 6.00174978127734e-06,
|
| 755 |
+
"loss": 0.1315,
|
| 756 |
+
"step": 24000
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 2.12,
|
| 760 |
+
"learning_rate": 5.885097696121319e-06,
|
| 761 |
+
"loss": 0.1489,
|
| 762 |
+
"step": 24200
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 2.13,
|
| 766 |
+
"learning_rate": 5.768445610965296e-06,
|
| 767 |
+
"loss": 0.1527,
|
| 768 |
+
"step": 24400
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 2.15,
|
| 772 |
+
"learning_rate": 5.651793525809274e-06,
|
| 773 |
+
"loss": 0.1516,
|
| 774 |
+
"step": 24600
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 2.17,
|
| 778 |
+
"learning_rate": 5.535141440653252e-06,
|
| 779 |
+
"loss": 0.1203,
|
| 780 |
+
"step": 24800
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 2.19,
|
| 784 |
+
"learning_rate": 5.41848935549723e-06,
|
| 785 |
+
"loss": 0.1204,
|
| 786 |
+
"step": 25000
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 2.2,
|
| 790 |
+
"learning_rate": 5.301837270341208e-06,
|
| 791 |
+
"loss": 0.15,
|
| 792 |
+
"step": 25200
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 2.22,
|
| 796 |
+
"learning_rate": 5.185185185185185e-06,
|
| 797 |
+
"loss": 0.1453,
|
| 798 |
+
"step": 25400
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 2.24,
|
| 802 |
+
"learning_rate": 5.068533100029163e-06,
|
| 803 |
+
"loss": 0.1406,
|
| 804 |
+
"step": 25600
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 2.26,
|
| 808 |
+
"learning_rate": 4.951881014873141e-06,
|
| 809 |
+
"loss": 0.1263,
|
| 810 |
+
"step": 25800
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 2.27,
|
| 814 |
+
"learning_rate": 4.835228929717119e-06,
|
| 815 |
+
"loss": 0.1187,
|
| 816 |
+
"step": 26000
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 2.29,
|
| 820 |
+
"learning_rate": 4.718576844561097e-06,
|
| 821 |
+
"loss": 0.1363,
|
| 822 |
+
"step": 26200
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 2.31,
|
| 826 |
+
"learning_rate": 4.6019247594050745e-06,
|
| 827 |
+
"loss": 0.1284,
|
| 828 |
+
"step": 26400
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 2.33,
|
| 832 |
+
"learning_rate": 4.485272674249053e-06,
|
| 833 |
+
"loss": 0.1371,
|
| 834 |
+
"step": 26600
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.34,
|
| 838 |
+
"learning_rate": 4.3686205890930305e-06,
|
| 839 |
+
"loss": 0.128,
|
| 840 |
+
"step": 26800
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 2.36,
|
| 844 |
+
"learning_rate": 4.251968503937008e-06,
|
| 845 |
+
"loss": 0.1246,
|
| 846 |
+
"step": 27000
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 2.38,
|
| 850 |
+
"learning_rate": 4.135316418780986e-06,
|
| 851 |
+
"loss": 0.1283,
|
| 852 |
+
"step": 27200
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 2.4,
|
| 856 |
+
"learning_rate": 4.018664333624964e-06,
|
| 857 |
+
"loss": 0.1382,
|
| 858 |
+
"step": 27400
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 2.41,
|
| 862 |
+
"learning_rate": 3.902012248468942e-06,
|
| 863 |
+
"loss": 0.1324,
|
| 864 |
+
"step": 27600
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 2.43,
|
| 868 |
+
"learning_rate": 3.7853601633129196e-06,
|
| 869 |
+
"loss": 0.1305,
|
| 870 |
+
"step": 27800
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 2.45,
|
| 874 |
+
"learning_rate": 3.668708078156897e-06,
|
| 875 |
+
"loss": 0.1319,
|
| 876 |
+
"step": 28000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 2.47,
|
| 880 |
+
"learning_rate": 3.552055993000875e-06,
|
| 881 |
+
"loss": 0.1322,
|
| 882 |
+
"step": 28200
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 2.48,
|
| 886 |
+
"learning_rate": 3.435403907844853e-06,
|
| 887 |
+
"loss": 0.1424,
|
| 888 |
+
"step": 28400
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 2.5,
|
| 892 |
+
"learning_rate": 3.318751822688831e-06,
|
| 893 |
+
"loss": 0.1473,
|
| 894 |
+
"step": 28600
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 2.52,
|
| 898 |
+
"learning_rate": 3.2020997375328087e-06,
|
| 899 |
+
"loss": 0.1361,
|
| 900 |
+
"step": 28800
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 2.54,
|
| 904 |
+
"learning_rate": 3.0854476523767863e-06,
|
| 905 |
+
"loss": 0.128,
|
| 906 |
+
"step": 29000
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.55,
|
| 910 |
+
"learning_rate": 2.9687955672207643e-06,
|
| 911 |
+
"loss": 0.1212,
|
| 912 |
+
"step": 29200
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 2.57,
|
| 916 |
+
"learning_rate": 2.852143482064742e-06,
|
| 917 |
+
"loss": 0.13,
|
| 918 |
+
"step": 29400
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 2.59,
|
| 922 |
+
"learning_rate": 2.73549139690872e-06,
|
| 923 |
+
"loss": 0.1401,
|
| 924 |
+
"step": 29600
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 2.61,
|
| 928 |
+
"learning_rate": 2.618839311752698e-06,
|
| 929 |
+
"loss": 0.1412,
|
| 930 |
+
"step": 29800
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 2.62,
|
| 934 |
+
"learning_rate": 2.502187226596676e-06,
|
| 935 |
+
"loss": 0.1262,
|
| 936 |
+
"step": 30000
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 2.64,
|
| 940 |
+
"learning_rate": 2.3855351414406534e-06,
|
| 941 |
+
"loss": 0.1412,
|
| 942 |
+
"step": 30200
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 2.66,
|
| 946 |
+
"learning_rate": 2.2688830562846314e-06,
|
| 947 |
+
"loss": 0.1441,
|
| 948 |
+
"step": 30400
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.68,
|
| 952 |
+
"learning_rate": 2.152230971128609e-06,
|
| 953 |
+
"loss": 0.1443,
|
| 954 |
+
"step": 30600
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 2.69,
|
| 958 |
+
"learning_rate": 2.035578885972587e-06,
|
| 959 |
+
"loss": 0.1541,
|
| 960 |
+
"step": 30800
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 2.71,
|
| 964 |
+
"learning_rate": 1.918926800816565e-06,
|
| 965 |
+
"loss": 0.1187,
|
| 966 |
+
"step": 31000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 2.73,
|
| 970 |
+
"learning_rate": 1.8022747156605425e-06,
|
| 971 |
+
"loss": 0.1254,
|
| 972 |
+
"step": 31200
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 2.75,
|
| 976 |
+
"learning_rate": 1.6856226305045203e-06,
|
| 977 |
+
"loss": 0.1403,
|
| 978 |
+
"step": 31400
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 2.76,
|
| 982 |
+
"learning_rate": 1.5689705453484983e-06,
|
| 983 |
+
"loss": 0.1272,
|
| 984 |
+
"step": 31600
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.78,
|
| 988 |
+
"learning_rate": 1.452318460192476e-06,
|
| 989 |
+
"loss": 0.1223,
|
| 990 |
+
"step": 31800
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 2.8,
|
| 994 |
+
"learning_rate": 1.335666375036454e-06,
|
| 995 |
+
"loss": 0.1448,
|
| 996 |
+
"step": 32000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 2.82,
|
| 1000 |
+
"learning_rate": 1.2190142898804318e-06,
|
| 1001 |
+
"loss": 0.1376,
|
| 1002 |
+
"step": 32200
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 2.83,
|
| 1006 |
+
"learning_rate": 1.1023622047244096e-06,
|
| 1007 |
+
"loss": 0.1246,
|
| 1008 |
+
"step": 32400
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 2.85,
|
| 1012 |
+
"learning_rate": 9.857101195683874e-07,
|
| 1013 |
+
"loss": 0.1548,
|
| 1014 |
+
"step": 32600
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.87,
|
| 1018 |
+
"learning_rate": 8.690580344123652e-07,
|
| 1019 |
+
"loss": 0.1408,
|
| 1020 |
+
"step": 32800
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 2.89,
|
| 1024 |
+
"learning_rate": 7.524059492563431e-07,
|
| 1025 |
+
"loss": 0.1282,
|
| 1026 |
+
"step": 33000
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 2.9,
|
| 1030 |
+
"learning_rate": 6.357538641003208e-07,
|
| 1031 |
+
"loss": 0.1552,
|
| 1032 |
+
"step": 33200
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 2.92,
|
| 1036 |
+
"learning_rate": 5.191017789442987e-07,
|
| 1037 |
+
"loss": 0.1272,
|
| 1038 |
+
"step": 33400
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 2.94,
|
| 1042 |
+
"learning_rate": 4.024496937882765e-07,
|
| 1043 |
+
"loss": 0.1166,
|
| 1044 |
+
"step": 33600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.96,
|
| 1048 |
+
"learning_rate": 2.8579760863225433e-07,
|
| 1049 |
+
"loss": 0.1245,
|
| 1050 |
+
"step": 33800
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 2.97,
|
| 1054 |
+
"learning_rate": 1.6914552347623217e-07,
|
| 1055 |
+
"loss": 0.1233,
|
| 1056 |
+
"step": 34000
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 2.99,
|
| 1060 |
+
"learning_rate": 5.249343832020998e-08,
|
| 1061 |
+
"loss": 0.1192,
|
| 1062 |
+
"step": 34200
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 3.0,
|
| 1066 |
+
"eval_loss": 0.23235057294368744,
|
| 1067 |
+
"eval_mae": 0.3088998794555664,
|
| 1068 |
+
"eval_mse": 0.23235058784484863,
|
| 1069 |
+
"eval_rmse": 0.4820275902748108,
|
| 1070 |
+
"eval_runtime": 434.9142,
|
| 1071 |
+
"eval_samples_per_second": 113.204,
|
| 1072 |
+
"eval_steps_per_second": 14.152,
|
| 1073 |
+
"step": 34290
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 3.0,
|
| 1077 |
+
"step": 34290,
|
| 1078 |
+
"total_flos": 1.1316755048115336e+16,
|
| 1079 |
+
"train_loss": 0.19905866315323814,
|
| 1080 |
+
"train_runtime": 6850.4631,
|
| 1081 |
+
"train_samples_per_second": 40.041,
|
| 1082 |
+
"train_steps_per_second": 5.006
|
| 1083 |
+
}
|
| 1084 |
+
],
|
| 1085 |
+
"max_steps": 34290,
|
| 1086 |
+
"num_train_epochs": 3,
|
| 1087 |
+
"total_flos": 1.1316755048115336e+16,
|
| 1088 |
+
"trial_name": null,
|
| 1089 |
+
"trial_params": null
|
| 1090 |
+
}
|