frostsolutions/frost-vision-v2-google_vit-base-patch16-224
Browse files- README.md +9 -9
- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Nov18_05-40-36_a7bbef788e81/events.out.tfevents.1731909195.a7bbef788e81.210.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +2181 -0
README.md
CHANGED
|
@@ -26,16 +26,16 @@ model-index:
|
|
| 26 |
metrics:
|
| 27 |
- name: Accuracy
|
| 28 |
type: accuracy
|
| 29 |
-
value: 0.
|
| 30 |
- name: F1
|
| 31 |
type: f1
|
| 32 |
-
value: 0.
|
| 33 |
- name: Precision
|
| 34 |
type: precision
|
| 35 |
-
value: 0.
|
| 36 |
- name: Recall
|
| 37 |
type: recall
|
| 38 |
-
value: 0.
|
| 39 |
---
|
| 40 |
|
| 41 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -45,11 +45,11 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 45 |
|
| 46 |
This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the webdataset dataset.
|
| 47 |
It achieves the following results on the evaluation set:
|
| 48 |
-
- Loss: 0.
|
| 49 |
-
- Accuracy: 0.
|
| 50 |
-
- F1: 0.
|
| 51 |
-
- Precision: 0.
|
| 52 |
-
- Recall: 0.
|
| 53 |
|
| 54 |
## Model description
|
| 55 |
|
|
|
|
| 26 |
metrics:
|
| 27 |
- name: Accuracy
|
| 28 |
type: accuracy
|
| 29 |
+
value: 0.9359420289855073
|
| 30 |
- name: F1
|
| 31 |
type: f1
|
| 32 |
+
value: 0.8380952380952381
|
| 33 |
- name: Precision
|
| 34 |
type: precision
|
| 35 |
+
value: 0.8895800933125972
|
| 36 |
- name: Recall
|
| 37 |
type: recall
|
| 38 |
+
value: 0.7922437673130194
|
| 39 |
---
|
| 40 |
|
| 41 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 45 |
|
| 46 |
This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the webdataset dataset.
|
| 47 |
It achieves the following results on the evaluation set:
|
| 48 |
+
- Loss: 0.1562
|
| 49 |
+
- Accuracy: 0.9359
|
| 50 |
+
- F1: 0.8381
|
| 51 |
+
- Precision: 0.8896
|
| 52 |
+
- Recall: 0.7922
|
| 53 |
|
| 54 |
## Model description
|
| 55 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"eval_accuracy": 0.9359420289855073,
|
| 4 |
+
"eval_f1": 0.8380952380952381,
|
| 5 |
+
"eval_loss": 0.15622803568840027,
|
| 6 |
+
"eval_precision": 0.8895800933125972,
|
| 7 |
+
"eval_recall": 0.7922437673130194,
|
| 8 |
+
"eval_runtime": 3.5438,
|
| 9 |
+
"eval_samples_per_second": 97.353,
|
| 10 |
+
"eval_steps_per_second": 12.416,
|
| 11 |
+
"total_flos": 3.2060734740537754e+18,
|
| 12 |
+
"train_loss": 0.09901393407606074,
|
| 13 |
+
"train_runtime": 730.7896,
|
| 14 |
+
"train_samples_per_second": 56.61,
|
| 15 |
+
"train_steps_per_second": 3.571
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"eval_accuracy": 0.9359420289855073,
|
| 4 |
+
"eval_f1": 0.8380952380952381,
|
| 5 |
+
"eval_loss": 0.15622803568840027,
|
| 6 |
+
"eval_precision": 0.8895800933125972,
|
| 7 |
+
"eval_recall": 0.7922437673130194,
|
| 8 |
+
"eval_runtime": 3.5438,
|
| 9 |
+
"eval_samples_per_second": 97.353,
|
| 10 |
+
"eval_steps_per_second": 12.416
|
| 11 |
+
}
|
runs/Nov18_05-40-36_a7bbef788e81/events.out.tfevents.1731909195.a7bbef788e81.210.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f388958303fa478a4bb3706b37197ac2ac93ec94389a4ce24113810635a398b
|
| 3 |
+
size 560
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"total_flos": 3.2060734740537754e+18,
|
| 4 |
+
"train_loss": 0.09901393407606074,
|
| 5 |
+
"train_runtime": 730.7896,
|
| 6 |
+
"train_samples_per_second": 56.61,
|
| 7 |
+
"train_steps_per_second": 3.571
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2181 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.15622803568840027,
|
| 3 |
+
"best_model_checkpoint": "frostsolutions/frost-vision-v2-google_vit-base-patch16-224/checkpoint-500",
|
| 4 |
+
"epoch": 30.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 2610,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.11494252873563218,
|
| 13 |
+
"grad_norm": 1.843792200088501,
|
| 14 |
+
"learning_rate": 1.9157088122605362e-06,
|
| 15 |
+
"loss": 0.7018,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.22988505747126436,
|
| 20 |
+
"grad_norm": 1.8868529796600342,
|
| 21 |
+
"learning_rate": 3.8314176245210725e-06,
|
| 22 |
+
"loss": 0.6818,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3448275862068966,
|
| 27 |
+
"grad_norm": 1.7420289516448975,
|
| 28 |
+
"learning_rate": 5.747126436781609e-06,
|
| 29 |
+
"loss": 0.6445,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.45977011494252873,
|
| 34 |
+
"grad_norm": 1.434963345527649,
|
| 35 |
+
"learning_rate": 7.662835249042145e-06,
|
| 36 |
+
"loss": 0.5892,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5747126436781609,
|
| 41 |
+
"grad_norm": 1.3582804203033447,
|
| 42 |
+
"learning_rate": 9.578544061302683e-06,
|
| 43 |
+
"loss": 0.5362,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6896551724137931,
|
| 48 |
+
"grad_norm": 1.2787097692489624,
|
| 49 |
+
"learning_rate": 1.1494252873563218e-05,
|
| 50 |
+
"loss": 0.4723,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.8045977011494253,
|
| 55 |
+
"grad_norm": 1.0611391067504883,
|
| 56 |
+
"learning_rate": 1.3409961685823755e-05,
|
| 57 |
+
"loss": 0.4433,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.9195402298850575,
|
| 62 |
+
"grad_norm": 1.0236091613769531,
|
| 63 |
+
"learning_rate": 1.532567049808429e-05,
|
| 64 |
+
"loss": 0.415,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.0344827586206897,
|
| 69 |
+
"grad_norm": 1.0555670261383057,
|
| 70 |
+
"learning_rate": 1.7241379310344828e-05,
|
| 71 |
+
"loss": 0.3879,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.1494252873563218,
|
| 76 |
+
"grad_norm": 1.0582798719406128,
|
| 77 |
+
"learning_rate": 1.9157088122605367e-05,
|
| 78 |
+
"loss": 0.3416,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.1494252873563218,
|
| 83 |
+
"eval_accuracy": 0.8771014492753623,
|
| 84 |
+
"eval_f1": 0.6124314442413162,
|
| 85 |
+
"eval_loss": 0.32730263471603394,
|
| 86 |
+
"eval_precision": 0.9005376344086021,
|
| 87 |
+
"eval_recall": 0.46398891966759004,
|
| 88 |
+
"eval_runtime": 2.5472,
|
| 89 |
+
"eval_samples_per_second": 135.44,
|
| 90 |
+
"eval_steps_per_second": 17.274,
|
| 91 |
+
"step": 100
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.264367816091954,
|
| 95 |
+
"grad_norm": 1.0698238611221313,
|
| 96 |
+
"learning_rate": 2.10727969348659e-05,
|
| 97 |
+
"loss": 0.3371,
|
| 98 |
+
"step": 110
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.3793103448275863,
|
| 102 |
+
"grad_norm": 1.0518220663070679,
|
| 103 |
+
"learning_rate": 2.2988505747126437e-05,
|
| 104 |
+
"loss": 0.3122,
|
| 105 |
+
"step": 120
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.4942528735632183,
|
| 109 |
+
"grad_norm": 1.3425720930099487,
|
| 110 |
+
"learning_rate": 2.4904214559386975e-05,
|
| 111 |
+
"loss": 0.2861,
|
| 112 |
+
"step": 130
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.6091954022988506,
|
| 116 |
+
"grad_norm": 0.789143979549408,
|
| 117 |
+
"learning_rate": 2.681992337164751e-05,
|
| 118 |
+
"loss": 0.2678,
|
| 119 |
+
"step": 140
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.7241379310344827,
|
| 123 |
+
"grad_norm": 0.8800056576728821,
|
| 124 |
+
"learning_rate": 2.8735632183908045e-05,
|
| 125 |
+
"loss": 0.2811,
|
| 126 |
+
"step": 150
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.839080459770115,
|
| 130 |
+
"grad_norm": 0.943029522895813,
|
| 131 |
+
"learning_rate": 3.065134099616858e-05,
|
| 132 |
+
"loss": 0.2767,
|
| 133 |
+
"step": 160
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.9540229885057472,
|
| 137 |
+
"grad_norm": 1.1474237442016602,
|
| 138 |
+
"learning_rate": 3.256704980842912e-05,
|
| 139 |
+
"loss": 0.266,
|
| 140 |
+
"step": 170
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.0689655172413794,
|
| 144 |
+
"grad_norm": 1.0653332471847534,
|
| 145 |
+
"learning_rate": 3.4482758620689657e-05,
|
| 146 |
+
"loss": 0.2381,
|
| 147 |
+
"step": 180
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 2.1839080459770113,
|
| 151 |
+
"grad_norm": 1.0384609699249268,
|
| 152 |
+
"learning_rate": 3.6398467432950195e-05,
|
| 153 |
+
"loss": 0.2266,
|
| 154 |
+
"step": 190
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 2.2988505747126435,
|
| 158 |
+
"grad_norm": 0.7939756512641907,
|
| 159 |
+
"learning_rate": 3.831417624521073e-05,
|
| 160 |
+
"loss": 0.2215,
|
| 161 |
+
"step": 200
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 2.2988505747126435,
|
| 165 |
+
"eval_accuracy": 0.9182608695652174,
|
| 166 |
+
"eval_f1": 0.7901785714285714,
|
| 167 |
+
"eval_loss": 0.21871700882911682,
|
| 168 |
+
"eval_precision": 0.8536977491961415,
|
| 169 |
+
"eval_recall": 0.7354570637119113,
|
| 170 |
+
"eval_runtime": 2.9768,
|
| 171 |
+
"eval_samples_per_second": 115.897,
|
| 172 |
+
"eval_steps_per_second": 14.781,
|
| 173 |
+
"step": 200
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 2.413793103448276,
|
| 177 |
+
"grad_norm": 0.8638312816619873,
|
| 178 |
+
"learning_rate": 4.0229885057471265e-05,
|
| 179 |
+
"loss": 0.2274,
|
| 180 |
+
"step": 210
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 2.528735632183908,
|
| 184 |
+
"grad_norm": 1.0766372680664062,
|
| 185 |
+
"learning_rate": 4.21455938697318e-05,
|
| 186 |
+
"loss": 0.1967,
|
| 187 |
+
"step": 220
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 2.6436781609195403,
|
| 191 |
+
"grad_norm": 0.8049170970916748,
|
| 192 |
+
"learning_rate": 4.406130268199234e-05,
|
| 193 |
+
"loss": 0.2041,
|
| 194 |
+
"step": 230
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.7586206896551726,
|
| 198 |
+
"grad_norm": 0.8865797519683838,
|
| 199 |
+
"learning_rate": 4.597701149425287e-05,
|
| 200 |
+
"loss": 0.2288,
|
| 201 |
+
"step": 240
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.873563218390805,
|
| 205 |
+
"grad_norm": 0.906036376953125,
|
| 206 |
+
"learning_rate": 4.789272030651341e-05,
|
| 207 |
+
"loss": 0.2205,
|
| 208 |
+
"step": 250
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 2.9885057471264367,
|
| 212 |
+
"grad_norm": 1.1091452836990356,
|
| 213 |
+
"learning_rate": 4.980842911877395e-05,
|
| 214 |
+
"loss": 0.2075,
|
| 215 |
+
"step": 260
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.103448275862069,
|
| 219 |
+
"grad_norm": 1.1074674129486084,
|
| 220 |
+
"learning_rate": 4.980842911877395e-05,
|
| 221 |
+
"loss": 0.1914,
|
| 222 |
+
"step": 270
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 3.218390804597701,
|
| 226 |
+
"grad_norm": 1.6170417070388794,
|
| 227 |
+
"learning_rate": 4.959557258407833e-05,
|
| 228 |
+
"loss": 0.1792,
|
| 229 |
+
"step": 280
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 3.3333333333333335,
|
| 233 |
+
"grad_norm": 1.5800983905792236,
|
| 234 |
+
"learning_rate": 4.938271604938271e-05,
|
| 235 |
+
"loss": 0.1985,
|
| 236 |
+
"step": 290
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 3.4482758620689653,
|
| 240 |
+
"grad_norm": 1.2079609632492065,
|
| 241 |
+
"learning_rate": 4.916985951468711e-05,
|
| 242 |
+
"loss": 0.1753,
|
| 243 |
+
"step": 300
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 3.4482758620689653,
|
| 247 |
+
"eval_accuracy": 0.923768115942029,
|
| 248 |
+
"eval_f1": 0.8098336948662328,
|
| 249 |
+
"eval_loss": 0.1898600310087204,
|
| 250 |
+
"eval_precision": 0.8472012102874432,
|
| 251 |
+
"eval_recall": 0.775623268698061,
|
| 252 |
+
"eval_runtime": 3.0971,
|
| 253 |
+
"eval_samples_per_second": 111.395,
|
| 254 |
+
"eval_steps_per_second": 14.207,
|
| 255 |
+
"step": 300
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.5632183908045976,
|
| 259 |
+
"grad_norm": 0.70270836353302,
|
| 260 |
+
"learning_rate": 4.895700297999149e-05,
|
| 261 |
+
"loss": 0.2106,
|
| 262 |
+
"step": 310
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.67816091954023,
|
| 266 |
+
"grad_norm": 0.7896942496299744,
|
| 267 |
+
"learning_rate": 4.874414644529587e-05,
|
| 268 |
+
"loss": 0.168,
|
| 269 |
+
"step": 320
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.793103448275862,
|
| 273 |
+
"grad_norm": 1.1563091278076172,
|
| 274 |
+
"learning_rate": 4.853128991060026e-05,
|
| 275 |
+
"loss": 0.2003,
|
| 276 |
+
"step": 330
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.9080459770114944,
|
| 280 |
+
"grad_norm": 1.192246437072754,
|
| 281 |
+
"learning_rate": 4.831843337590464e-05,
|
| 282 |
+
"loss": 0.203,
|
| 283 |
+
"step": 340
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 4.022988505747127,
|
| 287 |
+
"grad_norm": 1.551155686378479,
|
| 288 |
+
"learning_rate": 4.810557684120903e-05,
|
| 289 |
+
"loss": 0.1936,
|
| 290 |
+
"step": 350
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.137931034482759,
|
| 294 |
+
"grad_norm": 1.4264142513275146,
|
| 295 |
+
"learning_rate": 4.789272030651341e-05,
|
| 296 |
+
"loss": 0.1783,
|
| 297 |
+
"step": 360
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 4.252873563218391,
|
| 301 |
+
"grad_norm": 0.6229875087738037,
|
| 302 |
+
"learning_rate": 4.767986377181779e-05,
|
| 303 |
+
"loss": 0.1797,
|
| 304 |
+
"step": 370
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 4.3678160919540225,
|
| 308 |
+
"grad_norm": 0.6375844478607178,
|
| 309 |
+
"learning_rate": 4.746700723712218e-05,
|
| 310 |
+
"loss": 0.1589,
|
| 311 |
+
"step": 380
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 4.482758620689655,
|
| 315 |
+
"grad_norm": 1.297573447227478,
|
| 316 |
+
"learning_rate": 4.725415070242657e-05,
|
| 317 |
+
"loss": 0.142,
|
| 318 |
+
"step": 390
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 4.597701149425287,
|
| 322 |
+
"grad_norm": 0.7838256359100342,
|
| 323 |
+
"learning_rate": 4.704129416773095e-05,
|
| 324 |
+
"loss": 0.1656,
|
| 325 |
+
"step": 400
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 4.597701149425287,
|
| 329 |
+
"eval_accuracy": 0.9272463768115942,
|
| 330 |
+
"eval_f1": 0.8174545454545454,
|
| 331 |
+
"eval_loss": 0.17321595549583435,
|
| 332 |
+
"eval_precision": 0.8606431852986217,
|
| 333 |
+
"eval_recall": 0.778393351800554,
|
| 334 |
+
"eval_runtime": 4.6227,
|
| 335 |
+
"eval_samples_per_second": 74.631,
|
| 336 |
+
"eval_steps_per_second": 9.518,
|
| 337 |
+
"step": 400
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 4.712643678160919,
|
| 341 |
+
"grad_norm": 0.8572351932525635,
|
| 342 |
+
"learning_rate": 4.682843763303534e-05,
|
| 343 |
+
"loss": 0.1843,
|
| 344 |
+
"step": 410
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.827586206896552,
|
| 348 |
+
"grad_norm": 0.9022108912467957,
|
| 349 |
+
"learning_rate": 4.661558109833972e-05,
|
| 350 |
+
"loss": 0.1638,
|
| 351 |
+
"step": 420
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 4.942528735632184,
|
| 355 |
+
"grad_norm": 1.1695079803466797,
|
| 356 |
+
"learning_rate": 4.640272456364411e-05,
|
| 357 |
+
"loss": 0.1447,
|
| 358 |
+
"step": 430
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 5.057471264367816,
|
| 362 |
+
"grad_norm": 0.9919353723526001,
|
| 363 |
+
"learning_rate": 4.618986802894849e-05,
|
| 364 |
+
"loss": 0.1796,
|
| 365 |
+
"step": 440
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 5.172413793103448,
|
| 369 |
+
"grad_norm": 0.9828383326530457,
|
| 370 |
+
"learning_rate": 4.597701149425287e-05,
|
| 371 |
+
"loss": 0.1376,
|
| 372 |
+
"step": 450
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 5.287356321839081,
|
| 376 |
+
"grad_norm": 0.7474183440208435,
|
| 377 |
+
"learning_rate": 4.576415495955726e-05,
|
| 378 |
+
"loss": 0.1245,
|
| 379 |
+
"step": 460
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 5.402298850574713,
|
| 383 |
+
"grad_norm": 0.9904183149337769,
|
| 384 |
+
"learning_rate": 4.555129842486164e-05,
|
| 385 |
+
"loss": 0.1531,
|
| 386 |
+
"step": 470
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 5.517241379310345,
|
| 390 |
+
"grad_norm": 0.7156611680984497,
|
| 391 |
+
"learning_rate": 4.5338441890166025e-05,
|
| 392 |
+
"loss": 0.1551,
|
| 393 |
+
"step": 480
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 5.6321839080459775,
|
| 397 |
+
"grad_norm": 0.7088082432746887,
|
| 398 |
+
"learning_rate": 4.512558535547042e-05,
|
| 399 |
+
"loss": 0.1444,
|
| 400 |
+
"step": 490
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 5.747126436781609,
|
| 404 |
+
"grad_norm": 0.8901511430740356,
|
| 405 |
+
"learning_rate": 4.49127288207748e-05,
|
| 406 |
+
"loss": 0.1288,
|
| 407 |
+
"step": 500
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 5.747126436781609,
|
| 411 |
+
"eval_accuracy": 0.9359420289855073,
|
| 412 |
+
"eval_f1": 0.8380952380952381,
|
| 413 |
+
"eval_loss": 0.15622803568840027,
|
| 414 |
+
"eval_precision": 0.8895800933125972,
|
| 415 |
+
"eval_recall": 0.7922437673130194,
|
| 416 |
+
"eval_runtime": 2.5149,
|
| 417 |
+
"eval_samples_per_second": 137.184,
|
| 418 |
+
"eval_steps_per_second": 17.496,
|
| 419 |
+
"step": 500
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.862068965517241,
|
| 423 |
+
"grad_norm": 0.8678961992263794,
|
| 424 |
+
"learning_rate": 4.469987228607918e-05,
|
| 425 |
+
"loss": 0.144,
|
| 426 |
+
"step": 510
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 5.977011494252873,
|
| 430 |
+
"grad_norm": 0.8695370554924011,
|
| 431 |
+
"learning_rate": 4.448701575138357e-05,
|
| 432 |
+
"loss": 0.1543,
|
| 433 |
+
"step": 520
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 6.091954022988506,
|
| 437 |
+
"grad_norm": 0.6334937810897827,
|
| 438 |
+
"learning_rate": 4.427415921668795e-05,
|
| 439 |
+
"loss": 0.118,
|
| 440 |
+
"step": 530
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 6.206896551724138,
|
| 444 |
+
"grad_norm": 0.8258758187294006,
|
| 445 |
+
"learning_rate": 4.406130268199234e-05,
|
| 446 |
+
"loss": 0.1127,
|
| 447 |
+
"step": 540
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 6.32183908045977,
|
| 451 |
+
"grad_norm": 1.4297465085983276,
|
| 452 |
+
"learning_rate": 4.384844614729672e-05,
|
| 453 |
+
"loss": 0.124,
|
| 454 |
+
"step": 550
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 6.436781609195402,
|
| 458 |
+
"grad_norm": 0.9284784197807312,
|
| 459 |
+
"learning_rate": 4.3635589612601105e-05,
|
| 460 |
+
"loss": 0.1153,
|
| 461 |
+
"step": 560
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 6.551724137931035,
|
| 465 |
+
"grad_norm": 0.7226102352142334,
|
| 466 |
+
"learning_rate": 4.342273307790549e-05,
|
| 467 |
+
"loss": 0.1204,
|
| 468 |
+
"step": 570
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 6.666666666666667,
|
| 472 |
+
"grad_norm": 1.2235733270645142,
|
| 473 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 474 |
+
"loss": 0.1393,
|
| 475 |
+
"step": 580
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 6.781609195402299,
|
| 479 |
+
"grad_norm": 0.6442233324050903,
|
| 480 |
+
"learning_rate": 4.299702000851426e-05,
|
| 481 |
+
"loss": 0.1311,
|
| 482 |
+
"step": 590
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 6.896551724137931,
|
| 486 |
+
"grad_norm": 0.8887196183204651,
|
| 487 |
+
"learning_rate": 4.278416347381865e-05,
|
| 488 |
+
"loss": 0.1323,
|
| 489 |
+
"step": 600
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 6.896551724137931,
|
| 493 |
+
"eval_accuracy": 0.9321739130434783,
|
| 494 |
+
"eval_f1": 0.8326180257510729,
|
| 495 |
+
"eval_loss": 0.1597267985343933,
|
| 496 |
+
"eval_precision": 0.8609467455621301,
|
| 497 |
+
"eval_recall": 0.8060941828254847,
|
| 498 |
+
"eval_runtime": 2.5018,
|
| 499 |
+
"eval_samples_per_second": 137.898,
|
| 500 |
+
"eval_steps_per_second": 17.587,
|
| 501 |
+
"step": 600
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 7.011494252873563,
|
| 505 |
+
"grad_norm": 0.535347044467926,
|
| 506 |
+
"learning_rate": 4.257130693912303e-05,
|
| 507 |
+
"loss": 0.1151,
|
| 508 |
+
"step": 610
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 7.126436781609195,
|
| 512 |
+
"grad_norm": 1.1041584014892578,
|
| 513 |
+
"learning_rate": 4.235845040442742e-05,
|
| 514 |
+
"loss": 0.1196,
|
| 515 |
+
"step": 620
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 7.241379310344827,
|
| 519 |
+
"grad_norm": 1.098777413368225,
|
| 520 |
+
"learning_rate": 4.21455938697318e-05,
|
| 521 |
+
"loss": 0.1262,
|
| 522 |
+
"step": 630
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 7.35632183908046,
|
| 526 |
+
"grad_norm": 1.0132532119750977,
|
| 527 |
+
"learning_rate": 4.1932737335036185e-05,
|
| 528 |
+
"loss": 0.115,
|
| 529 |
+
"step": 640
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 7.471264367816092,
|
| 533 |
+
"grad_norm": 1.0600366592407227,
|
| 534 |
+
"learning_rate": 4.171988080034057e-05,
|
| 535 |
+
"loss": 0.1205,
|
| 536 |
+
"step": 650
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 7.586206896551724,
|
| 540 |
+
"grad_norm": 0.8362026810646057,
|
| 541 |
+
"learning_rate": 4.1507024265644955e-05,
|
| 542 |
+
"loss": 0.1294,
|
| 543 |
+
"step": 660
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 7.7011494252873565,
|
| 547 |
+
"grad_norm": 0.9791185855865479,
|
| 548 |
+
"learning_rate": 4.129416773094934e-05,
|
| 549 |
+
"loss": 0.1253,
|
| 550 |
+
"step": 670
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 7.816091954022989,
|
| 554 |
+
"grad_norm": 0.6780907511711121,
|
| 555 |
+
"learning_rate": 4.1081311196253725e-05,
|
| 556 |
+
"loss": 0.1072,
|
| 557 |
+
"step": 680
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 7.931034482758621,
|
| 561 |
+
"grad_norm": 0.9286981225013733,
|
| 562 |
+
"learning_rate": 4.086845466155811e-05,
|
| 563 |
+
"loss": 0.114,
|
| 564 |
+
"step": 690
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 8.045977011494253,
|
| 568 |
+
"grad_norm": 0.48207980394363403,
|
| 569 |
+
"learning_rate": 4.06555981268625e-05,
|
| 570 |
+
"loss": 0.1004,
|
| 571 |
+
"step": 700
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 8.045977011494253,
|
| 575 |
+
"eval_accuracy": 0.9315942028985508,
|
| 576 |
+
"eval_f1": 0.8323863636363636,
|
| 577 |
+
"eval_loss": 0.161317378282547,
|
| 578 |
+
"eval_precision": 0.8542274052478134,
|
| 579 |
+
"eval_recall": 0.8116343490304709,
|
| 580 |
+
"eval_runtime": 2.5046,
|
| 581 |
+
"eval_samples_per_second": 137.746,
|
| 582 |
+
"eval_steps_per_second": 17.568,
|
| 583 |
+
"step": 700
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 8.160919540229886,
|
| 587 |
+
"grad_norm": 1.1616382598876953,
|
| 588 |
+
"learning_rate": 4.044274159216688e-05,
|
| 589 |
+
"loss": 0.1209,
|
| 590 |
+
"step": 710
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 8.275862068965518,
|
| 594 |
+
"grad_norm": 0.7316162586212158,
|
| 595 |
+
"learning_rate": 4.0229885057471265e-05,
|
| 596 |
+
"loss": 0.096,
|
| 597 |
+
"step": 720
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 8.39080459770115,
|
| 601 |
+
"grad_norm": 0.8447398543357849,
|
| 602 |
+
"learning_rate": 4.001702852277565e-05,
|
| 603 |
+
"loss": 0.1163,
|
| 604 |
+
"step": 730
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 8.505747126436782,
|
| 608 |
+
"grad_norm": 0.6557740569114685,
|
| 609 |
+
"learning_rate": 3.9804171988080035e-05,
|
| 610 |
+
"loss": 0.0951,
|
| 611 |
+
"step": 740
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 8.620689655172415,
|
| 615 |
+
"grad_norm": 0.9132739901542664,
|
| 616 |
+
"learning_rate": 3.959131545338442e-05,
|
| 617 |
+
"loss": 0.1045,
|
| 618 |
+
"step": 750
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 8.735632183908045,
|
| 622 |
+
"grad_norm": 1.8383800983428955,
|
| 623 |
+
"learning_rate": 3.9378458918688805e-05,
|
| 624 |
+
"loss": 0.1007,
|
| 625 |
+
"step": 760
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 8.850574712643677,
|
| 629 |
+
"grad_norm": 1.6262847185134888,
|
| 630 |
+
"learning_rate": 3.9165602383993187e-05,
|
| 631 |
+
"loss": 0.0886,
|
| 632 |
+
"step": 770
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 8.96551724137931,
|
| 636 |
+
"grad_norm": 1.1226928234100342,
|
| 637 |
+
"learning_rate": 3.8952745849297575e-05,
|
| 638 |
+
"loss": 0.1136,
|
| 639 |
+
"step": 780
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 9.080459770114942,
|
| 643 |
+
"grad_norm": 0.8685470223426819,
|
| 644 |
+
"learning_rate": 3.873988931460196e-05,
|
| 645 |
+
"loss": 0.086,
|
| 646 |
+
"step": 790
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 9.195402298850574,
|
| 650 |
+
"grad_norm": 0.662932276725769,
|
| 651 |
+
"learning_rate": 3.8527032779906345e-05,
|
| 652 |
+
"loss": 0.0956,
|
| 653 |
+
"step": 800
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 9.195402298850574,
|
| 657 |
+
"eval_accuracy": 0.933623188405797,
|
| 658 |
+
"eval_f1": 0.8367783321454028,
|
| 659 |
+
"eval_loss": 0.16119635105133057,
|
| 660 |
+
"eval_precision": 0.8619676945668135,
|
| 661 |
+
"eval_recall": 0.8130193905817175,
|
| 662 |
+
"eval_runtime": 3.6353,
|
| 663 |
+
"eval_samples_per_second": 94.902,
|
| 664 |
+
"eval_steps_per_second": 12.103,
|
| 665 |
+
"step": 800
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 9.310344827586206,
|
| 669 |
+
"grad_norm": 0.9096585512161255,
|
| 670 |
+
"learning_rate": 3.831417624521073e-05,
|
| 671 |
+
"loss": 0.0939,
|
| 672 |
+
"step": 810
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 9.425287356321839,
|
| 676 |
+
"grad_norm": 0.6489679217338562,
|
| 677 |
+
"learning_rate": 3.8101319710515115e-05,
|
| 678 |
+
"loss": 0.0894,
|
| 679 |
+
"step": 820
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 9.540229885057471,
|
| 683 |
+
"grad_norm": 1.230733871459961,
|
| 684 |
+
"learning_rate": 3.7888463175819497e-05,
|
| 685 |
+
"loss": 0.1027,
|
| 686 |
+
"step": 830
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 9.655172413793103,
|
| 690 |
+
"grad_norm": 0.5888535976409912,
|
| 691 |
+
"learning_rate": 3.7675606641123885e-05,
|
| 692 |
+
"loss": 0.1034,
|
| 693 |
+
"step": 840
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 9.770114942528735,
|
| 697 |
+
"grad_norm": 1.329034447669983,
|
| 698 |
+
"learning_rate": 3.7462750106428267e-05,
|
| 699 |
+
"loss": 0.0785,
|
| 700 |
+
"step": 850
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 9.885057471264368,
|
| 704 |
+
"grad_norm": 0.7995342016220093,
|
| 705 |
+
"learning_rate": 3.7249893571732655e-05,
|
| 706 |
+
"loss": 0.1072,
|
| 707 |
+
"step": 860
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 10.0,
|
| 711 |
+
"grad_norm": 2.6494181156158447,
|
| 712 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 713 |
+
"loss": 0.1007,
|
| 714 |
+
"step": 870
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 10.114942528735632,
|
| 718 |
+
"grad_norm": 0.9277469515800476,
|
| 719 |
+
"learning_rate": 3.682418050234142e-05,
|
| 720 |
+
"loss": 0.0835,
|
| 721 |
+
"step": 880
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 10.229885057471265,
|
| 725 |
+
"grad_norm": 0.8300806879997253,
|
| 726 |
+
"learning_rate": 3.661132396764581e-05,
|
| 727 |
+
"loss": 0.0853,
|
| 728 |
+
"step": 890
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 10.344827586206897,
|
| 732 |
+
"grad_norm": 0.49981820583343506,
|
| 733 |
+
"learning_rate": 3.6398467432950195e-05,
|
| 734 |
+
"loss": 0.0841,
|
| 735 |
+
"step": 900
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 10.344827586206897,
|
| 739 |
+
"eval_accuracy": 0.9344927536231884,
|
| 740 |
+
"eval_f1": 0.8383404864091559,
|
| 741 |
+
"eval_loss": 0.16212987899780273,
|
| 742 |
+
"eval_precision": 0.8668639053254438,
|
| 743 |
+
"eval_recall": 0.8116343490304709,
|
| 744 |
+
"eval_runtime": 2.4973,
|
| 745 |
+
"eval_samples_per_second": 138.151,
|
| 746 |
+
"eval_steps_per_second": 17.619,
|
| 747 |
+
"step": 900
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 10.459770114942529,
|
| 751 |
+
"grad_norm": 1.2328234910964966,
|
| 752 |
+
"learning_rate": 3.6185610898254577e-05,
|
| 753 |
+
"loss": 0.0825,
|
| 754 |
+
"step": 910
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 10.574712643678161,
|
| 758 |
+
"grad_norm": 0.48059162497520447,
|
| 759 |
+
"learning_rate": 3.5972754363558965e-05,
|
| 760 |
+
"loss": 0.0765,
|
| 761 |
+
"step": 920
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 10.689655172413794,
|
| 765 |
+
"grad_norm": 1.0247180461883545,
|
| 766 |
+
"learning_rate": 3.5759897828863347e-05,
|
| 767 |
+
"loss": 0.0876,
|
| 768 |
+
"step": 930
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 10.804597701149426,
|
| 772 |
+
"grad_norm": 0.6012043356895447,
|
| 773 |
+
"learning_rate": 3.5547041294167735e-05,
|
| 774 |
+
"loss": 0.08,
|
| 775 |
+
"step": 940
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 10.919540229885058,
|
| 779 |
+
"grad_norm": 0.8350706696510315,
|
| 780 |
+
"learning_rate": 3.5334184759472117e-05,
|
| 781 |
+
"loss": 0.0889,
|
| 782 |
+
"step": 950
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 11.03448275862069,
|
| 786 |
+
"grad_norm": 0.46032950282096863,
|
| 787 |
+
"learning_rate": 3.51213282247765e-05,
|
| 788 |
+
"loss": 0.0815,
|
| 789 |
+
"step": 960
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 11.149425287356323,
|
| 793 |
+
"grad_norm": 0.8548628687858582,
|
| 794 |
+
"learning_rate": 3.4908471690080887e-05,
|
| 795 |
+
"loss": 0.09,
|
| 796 |
+
"step": 970
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 11.264367816091955,
|
| 800 |
+
"grad_norm": 0.9966102242469788,
|
| 801 |
+
"learning_rate": 3.469561515538527e-05,
|
| 802 |
+
"loss": 0.0707,
|
| 803 |
+
"step": 980
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 11.379310344827585,
|
| 807 |
+
"grad_norm": 1.062538981437683,
|
| 808 |
+
"learning_rate": 3.4482758620689657e-05,
|
| 809 |
+
"loss": 0.0652,
|
| 810 |
+
"step": 990
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 11.494252873563218,
|
| 814 |
+
"grad_norm": 0.6467556953430176,
|
| 815 |
+
"learning_rate": 3.4269902085994045e-05,
|
| 816 |
+
"loss": 0.0764,
|
| 817 |
+
"step": 1000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 11.494252873563218,
|
| 821 |
+
"eval_accuracy": 0.9359420289855073,
|
| 822 |
+
"eval_f1": 0.8438162544169612,
|
| 823 |
+
"eval_loss": 0.15857626497745514,
|
| 824 |
+
"eval_precision": 0.8614718614718615,
|
| 825 |
+
"eval_recall": 0.8268698060941828,
|
| 826 |
+
"eval_runtime": 2.4932,
|
| 827 |
+
"eval_samples_per_second": 138.379,
|
| 828 |
+
"eval_steps_per_second": 17.648,
|
| 829 |
+
"step": 1000
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 11.60919540229885,
|
| 833 |
+
"grad_norm": 0.9217528700828552,
|
| 834 |
+
"learning_rate": 3.4057045551298427e-05,
|
| 835 |
+
"loss": 0.0884,
|
| 836 |
+
"step": 1010
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 11.724137931034482,
|
| 840 |
+
"grad_norm": 0.9330977201461792,
|
| 841 |
+
"learning_rate": 3.3844189016602815e-05,
|
| 842 |
+
"loss": 0.0869,
|
| 843 |
+
"step": 1020
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 11.839080459770114,
|
| 847 |
+
"grad_norm": 1.4273611307144165,
|
| 848 |
+
"learning_rate": 3.3631332481907197e-05,
|
| 849 |
+
"loss": 0.0846,
|
| 850 |
+
"step": 1030
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 11.954022988505747,
|
| 854 |
+
"grad_norm": 1.0101591348648071,
|
| 855 |
+
"learning_rate": 3.341847594721158e-05,
|
| 856 |
+
"loss": 0.0773,
|
| 857 |
+
"step": 1040
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 12.068965517241379,
|
| 861 |
+
"grad_norm": 1.1758484840393066,
|
| 862 |
+
"learning_rate": 3.3205619412515967e-05,
|
| 863 |
+
"loss": 0.0744,
|
| 864 |
+
"step": 1050
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 12.183908045977011,
|
| 868 |
+
"grad_norm": 0.8224564790725708,
|
| 869 |
+
"learning_rate": 3.299276287782035e-05,
|
| 870 |
+
"loss": 0.0795,
|
| 871 |
+
"step": 1060
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 12.298850574712644,
|
| 875 |
+
"grad_norm": 0.468171626329422,
|
| 876 |
+
"learning_rate": 3.277990634312474e-05,
|
| 877 |
+
"loss": 0.0637,
|
| 878 |
+
"step": 1070
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 12.413793103448276,
|
| 882 |
+
"grad_norm": 1.1701269149780273,
|
| 883 |
+
"learning_rate": 3.256704980842912e-05,
|
| 884 |
+
"loss": 0.0689,
|
| 885 |
+
"step": 1080
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 12.528735632183908,
|
| 889 |
+
"grad_norm": 1.3815059661865234,
|
| 890 |
+
"learning_rate": 3.235419327373351e-05,
|
| 891 |
+
"loss": 0.0861,
|
| 892 |
+
"step": 1090
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 12.64367816091954,
|
| 896 |
+
"grad_norm": 1.2272127866744995,
|
| 897 |
+
"learning_rate": 3.2141336739037895e-05,
|
| 898 |
+
"loss": 0.0726,
|
| 899 |
+
"step": 1100
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 12.64367816091954,
|
| 903 |
+
"eval_accuracy": 0.9420289855072463,
|
| 904 |
+
"eval_f1": 0.8593530239099859,
|
| 905 |
+
"eval_loss": 0.15456102788448334,
|
| 906 |
+
"eval_precision": 0.8728571428571429,
|
| 907 |
+
"eval_recall": 0.8462603878116344,
|
| 908 |
+
"eval_runtime": 2.5301,
|
| 909 |
+
"eval_samples_per_second": 136.358,
|
| 910 |
+
"eval_steps_per_second": 17.391,
|
| 911 |
+
"step": 1100
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 12.758620689655173,
|
| 915 |
+
"grad_norm": 0.9871719479560852,
|
| 916 |
+
"learning_rate": 3.192848020434228e-05,
|
| 917 |
+
"loss": 0.0683,
|
| 918 |
+
"step": 1110
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 12.873563218390805,
|
| 922 |
+
"grad_norm": 0.49420595169067383,
|
| 923 |
+
"learning_rate": 3.171562366964666e-05,
|
| 924 |
+
"loss": 0.0721,
|
| 925 |
+
"step": 1120
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 12.988505747126437,
|
| 929 |
+
"grad_norm": 0.25016915798187256,
|
| 930 |
+
"learning_rate": 3.150276713495105e-05,
|
| 931 |
+
"loss": 0.0785,
|
| 932 |
+
"step": 1130
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 13.10344827586207,
|
| 936 |
+
"grad_norm": 1.1972845792770386,
|
| 937 |
+
"learning_rate": 3.128991060025543e-05,
|
| 938 |
+
"loss": 0.0732,
|
| 939 |
+
"step": 1140
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 13.218390804597702,
|
| 943 |
+
"grad_norm": 0.6661795973777771,
|
| 944 |
+
"learning_rate": 3.107705406555981e-05,
|
| 945 |
+
"loss": 0.0709,
|
| 946 |
+
"step": 1150
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 13.333333333333334,
|
| 950 |
+
"grad_norm": 0.5791137218475342,
|
| 951 |
+
"learning_rate": 3.08641975308642e-05,
|
| 952 |
+
"loss": 0.0805,
|
| 953 |
+
"step": 1160
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 13.448275862068966,
|
| 957 |
+
"grad_norm": 0.6590031385421753,
|
| 958 |
+
"learning_rate": 3.065134099616858e-05,
|
| 959 |
+
"loss": 0.0665,
|
| 960 |
+
"step": 1170
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 13.563218390804598,
|
| 964 |
+
"grad_norm": 0.5040526986122131,
|
| 965 |
+
"learning_rate": 3.043848446147297e-05,
|
| 966 |
+
"loss": 0.0766,
|
| 967 |
+
"step": 1180
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 13.678160919540229,
|
| 971 |
+
"grad_norm": 0.785590648651123,
|
| 972 |
+
"learning_rate": 3.0225627926777357e-05,
|
| 973 |
+
"loss": 0.0557,
|
| 974 |
+
"step": 1190
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 13.793103448275861,
|
| 978 |
+
"grad_norm": 0.7830068469047546,
|
| 979 |
+
"learning_rate": 3.0012771392081738e-05,
|
| 980 |
+
"loss": 0.0732,
|
| 981 |
+
"step": 1200
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 13.793103448275861,
|
| 985 |
+
"eval_accuracy": 0.9408695652173913,
|
| 986 |
+
"eval_f1": 0.8565400843881856,
|
| 987 |
+
"eval_loss": 0.15285652875900269,
|
| 988 |
+
"eval_precision": 0.87,
|
| 989 |
+
"eval_recall": 0.8434903047091413,
|
| 990 |
+
"eval_runtime": 2.5569,
|
| 991 |
+
"eval_samples_per_second": 134.928,
|
| 992 |
+
"eval_steps_per_second": 17.208,
|
| 993 |
+
"step": 1200
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 13.908045977011493,
|
| 997 |
+
"grad_norm": 0.5470532178878784,
|
| 998 |
+
"learning_rate": 2.9799914857386123e-05,
|
| 999 |
+
"loss": 0.0665,
|
| 1000 |
+
"step": 1210
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 14.022988505747126,
|
| 1004 |
+
"grad_norm": 0.9854804873466492,
|
| 1005 |
+
"learning_rate": 2.9587058322690508e-05,
|
| 1006 |
+
"loss": 0.063,
|
| 1007 |
+
"step": 1220
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 14.137931034482758,
|
| 1011 |
+
"grad_norm": 0.562602162361145,
|
| 1012 |
+
"learning_rate": 2.9374201787994893e-05,
|
| 1013 |
+
"loss": 0.0683,
|
| 1014 |
+
"step": 1230
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 14.25287356321839,
|
| 1018 |
+
"grad_norm": 0.5807965993881226,
|
| 1019 |
+
"learning_rate": 2.9161345253299278e-05,
|
| 1020 |
+
"loss": 0.0695,
|
| 1021 |
+
"step": 1240
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 14.367816091954023,
|
| 1025 |
+
"grad_norm": 0.659021258354187,
|
| 1026 |
+
"learning_rate": 2.894848871860366e-05,
|
| 1027 |
+
"loss": 0.069,
|
| 1028 |
+
"step": 1250
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 14.482758620689655,
|
| 1032 |
+
"grad_norm": 0.42295771837234497,
|
| 1033 |
+
"learning_rate": 2.8735632183908045e-05,
|
| 1034 |
+
"loss": 0.0692,
|
| 1035 |
+
"step": 1260
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 14.597701149425287,
|
| 1039 |
+
"grad_norm": 1.1786874532699585,
|
| 1040 |
+
"learning_rate": 2.852277564921243e-05,
|
| 1041 |
+
"loss": 0.0564,
|
| 1042 |
+
"step": 1270
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 14.71264367816092,
|
| 1046 |
+
"grad_norm": 0.7675245404243469,
|
| 1047 |
+
"learning_rate": 2.8309919114516818e-05,
|
| 1048 |
+
"loss": 0.063,
|
| 1049 |
+
"step": 1280
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 14.827586206896552,
|
| 1053 |
+
"grad_norm": 0.6836528182029724,
|
| 1054 |
+
"learning_rate": 2.8097062579821203e-05,
|
| 1055 |
+
"loss": 0.0711,
|
| 1056 |
+
"step": 1290
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 14.942528735632184,
|
| 1060 |
+
"grad_norm": 0.6420086026191711,
|
| 1061 |
+
"learning_rate": 2.7884206045125588e-05,
|
| 1062 |
+
"loss": 0.0626,
|
| 1063 |
+
"step": 1300
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 14.942528735632184,
|
| 1067 |
+
"eval_accuracy": 0.9376811594202898,
|
| 1068 |
+
"eval_f1": 0.8484848484848485,
|
| 1069 |
+
"eval_loss": 0.1589481383562088,
|
| 1070 |
+
"eval_precision": 0.8637015781922525,
|
| 1071 |
+
"eval_recall": 0.8337950138504155,
|
| 1072 |
+
"eval_runtime": 2.7864,
|
| 1073 |
+
"eval_samples_per_second": 123.815,
|
| 1074 |
+
"eval_steps_per_second": 15.791,
|
| 1075 |
+
"step": 1300
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 15.057471264367816,
|
| 1079 |
+
"grad_norm": 0.5504136681556702,
|
| 1080 |
+
"learning_rate": 2.7671349510429973e-05,
|
| 1081 |
+
"loss": 0.0451,
|
| 1082 |
+
"step": 1310
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 15.172413793103448,
|
| 1086 |
+
"grad_norm": 1.0383538007736206,
|
| 1087 |
+
"learning_rate": 2.745849297573436e-05,
|
| 1088 |
+
"loss": 0.0569,
|
| 1089 |
+
"step": 1320
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 15.28735632183908,
|
| 1093 |
+
"grad_norm": 0.42968350648880005,
|
| 1094 |
+
"learning_rate": 2.724563644103874e-05,
|
| 1095 |
+
"loss": 0.0551,
|
| 1096 |
+
"step": 1330
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 15.402298850574713,
|
| 1100 |
+
"grad_norm": 0.8487522602081299,
|
| 1101 |
+
"learning_rate": 2.7032779906343125e-05,
|
| 1102 |
+
"loss": 0.0501,
|
| 1103 |
+
"step": 1340
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 15.517241379310345,
|
| 1107 |
+
"grad_norm": 0.8882860541343689,
|
| 1108 |
+
"learning_rate": 2.681992337164751e-05,
|
| 1109 |
+
"loss": 0.0467,
|
| 1110 |
+
"step": 1350
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 15.632183908045977,
|
| 1114 |
+
"grad_norm": 1.2020477056503296,
|
| 1115 |
+
"learning_rate": 2.6607066836951895e-05,
|
| 1116 |
+
"loss": 0.0649,
|
| 1117 |
+
"step": 1360
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 15.74712643678161,
|
| 1121 |
+
"grad_norm": 0.6664167046546936,
|
| 1122 |
+
"learning_rate": 2.6394210302256277e-05,
|
| 1123 |
+
"loss": 0.0462,
|
| 1124 |
+
"step": 1370
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 15.862068965517242,
|
| 1128 |
+
"grad_norm": 0.6712772250175476,
|
| 1129 |
+
"learning_rate": 2.618135376756067e-05,
|
| 1130 |
+
"loss": 0.0729,
|
| 1131 |
+
"step": 1380
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 15.977011494252874,
|
| 1135 |
+
"grad_norm": 0.5807361602783203,
|
| 1136 |
+
"learning_rate": 2.5968497232865053e-05,
|
| 1137 |
+
"loss": 0.0532,
|
| 1138 |
+
"step": 1390
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 16.091954022988507,
|
| 1142 |
+
"grad_norm": 0.4083256423473358,
|
| 1143 |
+
"learning_rate": 2.5755640698169435e-05,
|
| 1144 |
+
"loss": 0.0481,
|
| 1145 |
+
"step": 1400
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 16.091954022988507,
|
| 1149 |
+
"eval_accuracy": 0.9394202898550724,
|
| 1150 |
+
"eval_f1": 0.8510334996436208,
|
| 1151 |
+
"eval_loss": 0.16117151081562042,
|
| 1152 |
+
"eval_precision": 0.8766519823788547,
|
| 1153 |
+
"eval_recall": 0.8268698060941828,
|
| 1154 |
+
"eval_runtime": 3.3334,
|
| 1155 |
+
"eval_samples_per_second": 103.498,
|
| 1156 |
+
"eval_steps_per_second": 13.2,
|
| 1157 |
+
"step": 1400
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 16.20689655172414,
|
| 1161 |
+
"grad_norm": 0.8922817707061768,
|
| 1162 |
+
"learning_rate": 2.554278416347382e-05,
|
| 1163 |
+
"loss": 0.0674,
|
| 1164 |
+
"step": 1410
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 16.32183908045977,
|
| 1168 |
+
"grad_norm": 0.9631970524787903,
|
| 1169 |
+
"learning_rate": 2.5329927628778205e-05,
|
| 1170 |
+
"loss": 0.0503,
|
| 1171 |
+
"step": 1420
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 16.436781609195403,
|
| 1175 |
+
"grad_norm": 0.8879241943359375,
|
| 1176 |
+
"learning_rate": 2.511707109408259e-05,
|
| 1177 |
+
"loss": 0.0482,
|
| 1178 |
+
"step": 1430
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 16.551724137931036,
|
| 1182 |
+
"grad_norm": 0.7775533199310303,
|
| 1183 |
+
"learning_rate": 2.4904214559386975e-05,
|
| 1184 |
+
"loss": 0.0636,
|
| 1185 |
+
"step": 1440
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 16.666666666666668,
|
| 1189 |
+
"grad_norm": 0.9835919737815857,
|
| 1190 |
+
"learning_rate": 2.4691358024691357e-05,
|
| 1191 |
+
"loss": 0.0568,
|
| 1192 |
+
"step": 1450
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 16.7816091954023,
|
| 1196 |
+
"grad_norm": 0.7925294041633606,
|
| 1197 |
+
"learning_rate": 2.4478501489995745e-05,
|
| 1198 |
+
"loss": 0.0529,
|
| 1199 |
+
"step": 1460
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 16.896551724137932,
|
| 1203 |
+
"grad_norm": 0.6245427131652832,
|
| 1204 |
+
"learning_rate": 2.426564495530013e-05,
|
| 1205 |
+
"loss": 0.0641,
|
| 1206 |
+
"step": 1470
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 17.011494252873565,
|
| 1210 |
+
"grad_norm": 0.5181954503059387,
|
| 1211 |
+
"learning_rate": 2.4052788420604515e-05,
|
| 1212 |
+
"loss": 0.046,
|
| 1213 |
+
"step": 1480
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 17.126436781609197,
|
| 1217 |
+
"grad_norm": 0.40600207448005676,
|
| 1218 |
+
"learning_rate": 2.3839931885908897e-05,
|
| 1219 |
+
"loss": 0.039,
|
| 1220 |
+
"step": 1490
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 17.24137931034483,
|
| 1224 |
+
"grad_norm": 0.7081565260887146,
|
| 1225 |
+
"learning_rate": 2.3627075351213285e-05,
|
| 1226 |
+
"loss": 0.0507,
|
| 1227 |
+
"step": 1500
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 17.24137931034483,
|
| 1231 |
+
"eval_accuracy": 0.9339130434782609,
|
| 1232 |
+
"eval_f1": 0.8394366197183099,
|
| 1233 |
+
"eval_loss": 0.1679152399301529,
|
| 1234 |
+
"eval_precision": 0.8538681948424068,
|
| 1235 |
+
"eval_recall": 0.8254847645429363,
|
| 1236 |
+
"eval_runtime": 2.512,
|
| 1237 |
+
"eval_samples_per_second": 137.339,
|
| 1238 |
+
"eval_steps_per_second": 17.516,
|
| 1239 |
+
"step": 1500
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 17.35632183908046,
|
| 1243 |
+
"grad_norm": 0.9278731346130371,
|
| 1244 |
+
"learning_rate": 2.341421881651767e-05,
|
| 1245 |
+
"loss": 0.0546,
|
| 1246 |
+
"step": 1510
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 17.47126436781609,
|
| 1250 |
+
"grad_norm": 1.362691044807434,
|
| 1251 |
+
"learning_rate": 2.3201362281822055e-05,
|
| 1252 |
+
"loss": 0.0729,
|
| 1253 |
+
"step": 1520
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 17.586206896551722,
|
| 1257 |
+
"grad_norm": 0.6156861782073975,
|
| 1258 |
+
"learning_rate": 2.2988505747126437e-05,
|
| 1259 |
+
"loss": 0.0587,
|
| 1260 |
+
"step": 1530
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 17.701149425287355,
|
| 1264 |
+
"grad_norm": 0.530103862285614,
|
| 1265 |
+
"learning_rate": 2.277564921243082e-05,
|
| 1266 |
+
"loss": 0.0488,
|
| 1267 |
+
"step": 1540
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 17.816091954022987,
|
| 1271 |
+
"grad_norm": 0.6204285025596619,
|
| 1272 |
+
"learning_rate": 2.256279267773521e-05,
|
| 1273 |
+
"loss": 0.0499,
|
| 1274 |
+
"step": 1550
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 17.93103448275862,
|
| 1278 |
+
"grad_norm": 0.4868471920490265,
|
| 1279 |
+
"learning_rate": 2.234993614303959e-05,
|
| 1280 |
+
"loss": 0.0536,
|
| 1281 |
+
"step": 1560
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 18.04597701149425,
|
| 1285 |
+
"grad_norm": 0.5951109528541565,
|
| 1286 |
+
"learning_rate": 2.2137079608343977e-05,
|
| 1287 |
+
"loss": 0.0436,
|
| 1288 |
+
"step": 1570
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 18.160919540229884,
|
| 1292 |
+
"grad_norm": 1.3129435777664185,
|
| 1293 |
+
"learning_rate": 2.192422307364836e-05,
|
| 1294 |
+
"loss": 0.0455,
|
| 1295 |
+
"step": 1580
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 18.275862068965516,
|
| 1299 |
+
"grad_norm": 0.4130817651748657,
|
| 1300 |
+
"learning_rate": 2.1711366538952747e-05,
|
| 1301 |
+
"loss": 0.0539,
|
| 1302 |
+
"step": 1590
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 18.39080459770115,
|
| 1306 |
+
"grad_norm": 0.6867007613182068,
|
| 1307 |
+
"learning_rate": 2.149851000425713e-05,
|
| 1308 |
+
"loss": 0.0446,
|
| 1309 |
+
"step": 1600
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 18.39080459770115,
|
| 1313 |
+
"eval_accuracy": 0.9417391304347826,
|
| 1314 |
+
"eval_f1": 0.8597348220516399,
|
| 1315 |
+
"eval_loss": 0.16227415204048157,
|
| 1316 |
+
"eval_precision": 0.8663853727144867,
|
| 1317 |
+
"eval_recall": 0.853185595567867,
|
| 1318 |
+
"eval_runtime": 2.7697,
|
| 1319 |
+
"eval_samples_per_second": 124.564,
|
| 1320 |
+
"eval_steps_per_second": 15.886,
|
| 1321 |
+
"step": 1600
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 18.50574712643678,
|
| 1325 |
+
"grad_norm": 0.44690120220184326,
|
| 1326 |
+
"learning_rate": 2.1285653469561517e-05,
|
| 1327 |
+
"loss": 0.0411,
|
| 1328 |
+
"step": 1610
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 18.620689655172413,
|
| 1332 |
+
"grad_norm": 0.47931063175201416,
|
| 1333 |
+
"learning_rate": 2.10727969348659e-05,
|
| 1334 |
+
"loss": 0.0487,
|
| 1335 |
+
"step": 1620
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 18.735632183908045,
|
| 1339 |
+
"grad_norm": 0.5977224707603455,
|
| 1340 |
+
"learning_rate": 2.0859940400170287e-05,
|
| 1341 |
+
"loss": 0.0501,
|
| 1342 |
+
"step": 1630
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 18.850574712643677,
|
| 1346 |
+
"grad_norm": 0.5454962253570557,
|
| 1347 |
+
"learning_rate": 2.064708386547467e-05,
|
| 1348 |
+
"loss": 0.0516,
|
| 1349 |
+
"step": 1640
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 18.96551724137931,
|
| 1353 |
+
"grad_norm": 0.5072124004364014,
|
| 1354 |
+
"learning_rate": 2.0434227330779057e-05,
|
| 1355 |
+
"loss": 0.0407,
|
| 1356 |
+
"step": 1650
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 19.080459770114942,
|
| 1360 |
+
"grad_norm": 0.8368410468101501,
|
| 1361 |
+
"learning_rate": 2.022137079608344e-05,
|
| 1362 |
+
"loss": 0.0481,
|
| 1363 |
+
"step": 1660
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 19.195402298850574,
|
| 1367 |
+
"grad_norm": 1.143817663192749,
|
| 1368 |
+
"learning_rate": 2.0008514261387827e-05,
|
| 1369 |
+
"loss": 0.0406,
|
| 1370 |
+
"step": 1670
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 19.310344827586206,
|
| 1374 |
+
"grad_norm": 0.8621155023574829,
|
| 1375 |
+
"learning_rate": 1.979565772669221e-05,
|
| 1376 |
+
"loss": 0.0472,
|
| 1377 |
+
"step": 1680
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 19.42528735632184,
|
| 1381 |
+
"grad_norm": 0.8814654350280762,
|
| 1382 |
+
"learning_rate": 1.9582801191996593e-05,
|
| 1383 |
+
"loss": 0.0403,
|
| 1384 |
+
"step": 1690
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 19.54022988505747,
|
| 1388 |
+
"grad_norm": 0.37576496601104736,
|
| 1389 |
+
"learning_rate": 1.936994465730098e-05,
|
| 1390 |
+
"loss": 0.0498,
|
| 1391 |
+
"step": 1700
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 19.54022988505747,
|
| 1395 |
+
"eval_accuracy": 0.9417391304347826,
|
| 1396 |
+
"eval_f1": 0.860125260960334,
|
| 1397 |
+
"eval_loss": 0.16253642737865448,
|
| 1398 |
+
"eval_precision": 0.8643356643356643,
|
| 1399 |
+
"eval_recall": 0.8559556786703602,
|
| 1400 |
+
"eval_runtime": 3.3471,
|
| 1401 |
+
"eval_samples_per_second": 103.074,
|
| 1402 |
+
"eval_steps_per_second": 13.146,
|
| 1403 |
+
"step": 1700
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 19.655172413793103,
|
| 1407 |
+
"grad_norm": 1.052512764930725,
|
| 1408 |
+
"learning_rate": 1.9157088122605367e-05,
|
| 1409 |
+
"loss": 0.0569,
|
| 1410 |
+
"step": 1710
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 19.770114942528735,
|
| 1414 |
+
"grad_norm": 0.47423890233039856,
|
| 1415 |
+
"learning_rate": 1.8944231587909748e-05,
|
| 1416 |
+
"loss": 0.039,
|
| 1417 |
+
"step": 1720
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 19.885057471264368,
|
| 1421 |
+
"grad_norm": 0.922591507434845,
|
| 1422 |
+
"learning_rate": 1.8731375053214133e-05,
|
| 1423 |
+
"loss": 0.0376,
|
| 1424 |
+
"step": 1730
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 20.0,
|
| 1428 |
+
"grad_norm": 1.3589046001434326,
|
| 1429 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 1430 |
+
"loss": 0.0496,
|
| 1431 |
+
"step": 1740
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 20.114942528735632,
|
| 1435 |
+
"grad_norm": 0.8147189617156982,
|
| 1436 |
+
"learning_rate": 1.8305661983822907e-05,
|
| 1437 |
+
"loss": 0.0355,
|
| 1438 |
+
"step": 1750
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 20.229885057471265,
|
| 1442 |
+
"grad_norm": 0.2483612447977066,
|
| 1443 |
+
"learning_rate": 1.8092805449127288e-05,
|
| 1444 |
+
"loss": 0.0432,
|
| 1445 |
+
"step": 1760
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 20.344827586206897,
|
| 1449 |
+
"grad_norm": 0.29129087924957275,
|
| 1450 |
+
"learning_rate": 1.7879948914431673e-05,
|
| 1451 |
+
"loss": 0.0394,
|
| 1452 |
+
"step": 1770
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 20.45977011494253,
|
| 1456 |
+
"grad_norm": 0.6358793377876282,
|
| 1457 |
+
"learning_rate": 1.7667092379736058e-05,
|
| 1458 |
+
"loss": 0.0453,
|
| 1459 |
+
"step": 1780
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 20.57471264367816,
|
| 1463 |
+
"grad_norm": 0.3951164782047272,
|
| 1464 |
+
"learning_rate": 1.7454235845040443e-05,
|
| 1465 |
+
"loss": 0.0499,
|
| 1466 |
+
"step": 1790
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 20.689655172413794,
|
| 1470 |
+
"grad_norm": 0.5246742367744446,
|
| 1471 |
+
"learning_rate": 1.7241379310344828e-05,
|
| 1472 |
+
"loss": 0.0458,
|
| 1473 |
+
"step": 1800
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 20.689655172413794,
|
| 1477 |
+
"eval_accuracy": 0.9397101449275362,
|
| 1478 |
+
"eval_f1": 0.8533145275035261,
|
| 1479 |
+
"eval_loss": 0.16005827486515045,
|
| 1480 |
+
"eval_precision": 0.8692528735632183,
|
| 1481 |
+
"eval_recall": 0.8379501385041551,
|
| 1482 |
+
"eval_runtime": 2.5451,
|
| 1483 |
+
"eval_samples_per_second": 135.555,
|
| 1484 |
+
"eval_steps_per_second": 17.288,
|
| 1485 |
+
"step": 1800
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 20.804597701149426,
|
| 1489 |
+
"grad_norm": 0.4585019648075104,
|
| 1490 |
+
"learning_rate": 1.7028522775649213e-05,
|
| 1491 |
+
"loss": 0.0466,
|
| 1492 |
+
"step": 1810
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 20.919540229885058,
|
| 1496 |
+
"grad_norm": 0.5193343162536621,
|
| 1497 |
+
"learning_rate": 1.6815666240953598e-05,
|
| 1498 |
+
"loss": 0.0321,
|
| 1499 |
+
"step": 1820
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 21.03448275862069,
|
| 1503 |
+
"grad_norm": 0.3044123649597168,
|
| 1504 |
+
"learning_rate": 1.6602809706257983e-05,
|
| 1505 |
+
"loss": 0.0391,
|
| 1506 |
+
"step": 1830
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 21.149425287356323,
|
| 1510 |
+
"grad_norm": 0.3363288342952728,
|
| 1511 |
+
"learning_rate": 1.638995317156237e-05,
|
| 1512 |
+
"loss": 0.0344,
|
| 1513 |
+
"step": 1840
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 21.264367816091955,
|
| 1517 |
+
"grad_norm": 0.38722899556159973,
|
| 1518 |
+
"learning_rate": 1.6177096636866753e-05,
|
| 1519 |
+
"loss": 0.0352,
|
| 1520 |
+
"step": 1850
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 21.379310344827587,
|
| 1524 |
+
"grad_norm": 0.5276902914047241,
|
| 1525 |
+
"learning_rate": 1.596424010217114e-05,
|
| 1526 |
+
"loss": 0.0354,
|
| 1527 |
+
"step": 1860
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 21.49425287356322,
|
| 1531 |
+
"grad_norm": 0.37170952558517456,
|
| 1532 |
+
"learning_rate": 1.5751383567475523e-05,
|
| 1533 |
+
"loss": 0.0369,
|
| 1534 |
+
"step": 1870
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 21.60919540229885,
|
| 1538 |
+
"grad_norm": 1.0527151823043823,
|
| 1539 |
+
"learning_rate": 1.5538527032779905e-05,
|
| 1540 |
+
"loss": 0.032,
|
| 1541 |
+
"step": 1880
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 21.724137931034484,
|
| 1545 |
+
"grad_norm": 0.42286577820777893,
|
| 1546 |
+
"learning_rate": 1.532567049808429e-05,
|
| 1547 |
+
"loss": 0.0417,
|
| 1548 |
+
"step": 1890
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 21.839080459770116,
|
| 1552 |
+
"grad_norm": 0.4092840254306793,
|
| 1553 |
+
"learning_rate": 1.5112813963388678e-05,
|
| 1554 |
+
"loss": 0.0307,
|
| 1555 |
+
"step": 1900
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 21.839080459770116,
|
| 1559 |
+
"eval_accuracy": 0.9431884057971015,
|
| 1560 |
+
"eval_f1": 0.8636995827538247,
|
| 1561 |
+
"eval_loss": 0.16264864802360535,
|
| 1562 |
+
"eval_precision": 0.86731843575419,
|
| 1563 |
+
"eval_recall": 0.8601108033240997,
|
| 1564 |
+
"eval_runtime": 2.5207,
|
| 1565 |
+
"eval_samples_per_second": 136.867,
|
| 1566 |
+
"eval_steps_per_second": 17.455,
|
| 1567 |
+
"step": 1900
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 21.95402298850575,
|
| 1571 |
+
"grad_norm": 0.4764251708984375,
|
| 1572 |
+
"learning_rate": 1.4899957428693062e-05,
|
| 1573 |
+
"loss": 0.042,
|
| 1574 |
+
"step": 1910
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 22.06896551724138,
|
| 1578 |
+
"grad_norm": 0.925857424736023,
|
| 1579 |
+
"learning_rate": 1.4687100893997447e-05,
|
| 1580 |
+
"loss": 0.0342,
|
| 1581 |
+
"step": 1920
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 22.183908045977013,
|
| 1585 |
+
"grad_norm": 0.6081854701042175,
|
| 1586 |
+
"learning_rate": 1.447424435930183e-05,
|
| 1587 |
+
"loss": 0.0443,
|
| 1588 |
+
"step": 1930
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 22.298850574712645,
|
| 1592 |
+
"grad_norm": 0.35534176230430603,
|
| 1593 |
+
"learning_rate": 1.4261387824606215e-05,
|
| 1594 |
+
"loss": 0.0296,
|
| 1595 |
+
"step": 1940
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 22.413793103448278,
|
| 1599 |
+
"grad_norm": 0.5677205324172974,
|
| 1600 |
+
"learning_rate": 1.4048531289910602e-05,
|
| 1601 |
+
"loss": 0.0319,
|
| 1602 |
+
"step": 1950
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 22.52873563218391,
|
| 1606 |
+
"grad_norm": 0.9339087009429932,
|
| 1607 |
+
"learning_rate": 1.3835674755214987e-05,
|
| 1608 |
+
"loss": 0.0367,
|
| 1609 |
+
"step": 1960
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 22.64367816091954,
|
| 1613 |
+
"grad_norm": 0.30769288539886475,
|
| 1614 |
+
"learning_rate": 1.362281822051937e-05,
|
| 1615 |
+
"loss": 0.0372,
|
| 1616 |
+
"step": 1970
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 22.75862068965517,
|
| 1620 |
+
"grad_norm": 0.4518299698829651,
|
| 1621 |
+
"learning_rate": 1.3409961685823755e-05,
|
| 1622 |
+
"loss": 0.0501,
|
| 1623 |
+
"step": 1980
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 22.873563218390803,
|
| 1627 |
+
"grad_norm": 0.3183022737503052,
|
| 1628 |
+
"learning_rate": 1.3197105151128138e-05,
|
| 1629 |
+
"loss": 0.0338,
|
| 1630 |
+
"step": 1990
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 22.988505747126435,
|
| 1634 |
+
"grad_norm": 0.7140583395957947,
|
| 1635 |
+
"learning_rate": 1.2984248616432527e-05,
|
| 1636 |
+
"loss": 0.0334,
|
| 1637 |
+
"step": 2000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 22.988505747126435,
|
| 1641 |
+
"eval_accuracy": 0.9443478260869566,
|
| 1642 |
+
"eval_f1": 0.8642149929278642,
|
| 1643 |
+
"eval_loss": 0.16209261119365692,
|
| 1644 |
+
"eval_precision": 0.8829479768786127,
|
| 1645 |
+
"eval_recall": 0.8462603878116344,
|
| 1646 |
+
"eval_runtime": 3.2664,
|
| 1647 |
+
"eval_samples_per_second": 105.621,
|
| 1648 |
+
"eval_steps_per_second": 13.471,
|
| 1649 |
+
"step": 2000
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 23.103448275862068,
|
| 1653 |
+
"grad_norm": 0.647758424282074,
|
| 1654 |
+
"learning_rate": 1.277139208173691e-05,
|
| 1655 |
+
"loss": 0.036,
|
| 1656 |
+
"step": 2010
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 23.2183908045977,
|
| 1660 |
+
"grad_norm": 0.7994110584259033,
|
| 1661 |
+
"learning_rate": 1.2558535547041295e-05,
|
| 1662 |
+
"loss": 0.0313,
|
| 1663 |
+
"step": 2020
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 23.333333333333332,
|
| 1667 |
+
"grad_norm": 0.33477237820625305,
|
| 1668 |
+
"learning_rate": 1.2345679012345678e-05,
|
| 1669 |
+
"loss": 0.034,
|
| 1670 |
+
"step": 2030
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 23.448275862068964,
|
| 1674 |
+
"grad_norm": 0.17108239233493805,
|
| 1675 |
+
"learning_rate": 1.2132822477650065e-05,
|
| 1676 |
+
"loss": 0.0262,
|
| 1677 |
+
"step": 2040
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 23.563218390804597,
|
| 1681 |
+
"grad_norm": 0.40771496295928955,
|
| 1682 |
+
"learning_rate": 1.1919965942954448e-05,
|
| 1683 |
+
"loss": 0.0375,
|
| 1684 |
+
"step": 2050
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 23.67816091954023,
|
| 1688 |
+
"grad_norm": 1.7193657159805298,
|
| 1689 |
+
"learning_rate": 1.1707109408258835e-05,
|
| 1690 |
+
"loss": 0.0348,
|
| 1691 |
+
"step": 2060
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 23.79310344827586,
|
| 1695 |
+
"grad_norm": 0.2730618715286255,
|
| 1696 |
+
"learning_rate": 1.1494252873563218e-05,
|
| 1697 |
+
"loss": 0.0335,
|
| 1698 |
+
"step": 2070
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 23.908045977011493,
|
| 1702 |
+
"grad_norm": 0.12540611624717712,
|
| 1703 |
+
"learning_rate": 1.1281396338867605e-05,
|
| 1704 |
+
"loss": 0.0338,
|
| 1705 |
+
"step": 2080
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 24.022988505747126,
|
| 1709 |
+
"grad_norm": 0.6330766081809998,
|
| 1710 |
+
"learning_rate": 1.1068539804171988e-05,
|
| 1711 |
+
"loss": 0.0409,
|
| 1712 |
+
"step": 2090
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 24.137931034482758,
|
| 1716 |
+
"grad_norm": 0.5552668571472168,
|
| 1717 |
+
"learning_rate": 1.0855683269476373e-05,
|
| 1718 |
+
"loss": 0.0339,
|
| 1719 |
+
"step": 2100
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 24.137931034482758,
|
| 1723 |
+
"eval_accuracy": 0.9434782608695652,
|
| 1724 |
+
"eval_f1": 0.8644892286309938,
|
| 1725 |
+
"eval_loss": 0.1680324375629425,
|
| 1726 |
+
"eval_precision": 0.8675034867503487,
|
| 1727 |
+
"eval_recall": 0.8614958448753463,
|
| 1728 |
+
"eval_runtime": 2.5297,
|
| 1729 |
+
"eval_samples_per_second": 136.381,
|
| 1730 |
+
"eval_steps_per_second": 17.394,
|
| 1731 |
+
"step": 2100
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 24.25287356321839,
|
| 1735 |
+
"grad_norm": 0.8327674269676208,
|
| 1736 |
+
"learning_rate": 1.0642826734780758e-05,
|
| 1737 |
+
"loss": 0.0307,
|
| 1738 |
+
"step": 2110
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 24.367816091954023,
|
| 1742 |
+
"grad_norm": 0.33239325881004333,
|
| 1743 |
+
"learning_rate": 1.0429970200085143e-05,
|
| 1744 |
+
"loss": 0.0333,
|
| 1745 |
+
"step": 2120
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 24.482758620689655,
|
| 1749 |
+
"grad_norm": 0.44145530462265015,
|
| 1750 |
+
"learning_rate": 1.0217113665389528e-05,
|
| 1751 |
+
"loss": 0.037,
|
| 1752 |
+
"step": 2130
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 24.597701149425287,
|
| 1756 |
+
"grad_norm": 0.7869518995285034,
|
| 1757 |
+
"learning_rate": 1.0004257130693913e-05,
|
| 1758 |
+
"loss": 0.0364,
|
| 1759 |
+
"step": 2140
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 24.71264367816092,
|
| 1763 |
+
"grad_norm": 0.31890323758125305,
|
| 1764 |
+
"learning_rate": 9.791400595998297e-06,
|
| 1765 |
+
"loss": 0.0329,
|
| 1766 |
+
"step": 2150
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 24.82758620689655,
|
| 1770 |
+
"grad_norm": 0.53341144323349,
|
| 1771 |
+
"learning_rate": 9.578544061302683e-06,
|
| 1772 |
+
"loss": 0.0481,
|
| 1773 |
+
"step": 2160
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 24.942528735632184,
|
| 1777 |
+
"grad_norm": 0.4981224834918976,
|
| 1778 |
+
"learning_rate": 9.365687526607067e-06,
|
| 1779 |
+
"loss": 0.0368,
|
| 1780 |
+
"step": 2170
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 25.057471264367816,
|
| 1784 |
+
"grad_norm": 1.0361753702163696,
|
| 1785 |
+
"learning_rate": 9.152830991911453e-06,
|
| 1786 |
+
"loss": 0.0464,
|
| 1787 |
+
"step": 2180
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 25.17241379310345,
|
| 1791 |
+
"grad_norm": 0.1470736712217331,
|
| 1792 |
+
"learning_rate": 8.939974457215837e-06,
|
| 1793 |
+
"loss": 0.0372,
|
| 1794 |
+
"step": 2190
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 25.28735632183908,
|
| 1798 |
+
"grad_norm": 0.5003727078437805,
|
| 1799 |
+
"learning_rate": 8.727117922520222e-06,
|
| 1800 |
+
"loss": 0.0222,
|
| 1801 |
+
"step": 2200
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 25.28735632183908,
|
| 1805 |
+
"eval_accuracy": 0.9394202898550724,
|
| 1806 |
+
"eval_f1": 0.853743876836949,
|
| 1807 |
+
"eval_loss": 0.16558243334293365,
|
| 1808 |
+
"eval_precision": 0.8628005657708628,
|
| 1809 |
+
"eval_recall": 0.8448753462603878,
|
| 1810 |
+
"eval_runtime": 2.7223,
|
| 1811 |
+
"eval_samples_per_second": 126.732,
|
| 1812 |
+
"eval_steps_per_second": 16.163,
|
| 1813 |
+
"step": 2200
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 25.402298850574713,
|
| 1817 |
+
"grad_norm": 1.032658576965332,
|
| 1818 |
+
"learning_rate": 8.514261387824607e-06,
|
| 1819 |
+
"loss": 0.036,
|
| 1820 |
+
"step": 2210
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 25.517241379310345,
|
| 1824 |
+
"grad_norm": 0.3298107087612152,
|
| 1825 |
+
"learning_rate": 8.301404853128992e-06,
|
| 1826 |
+
"loss": 0.0311,
|
| 1827 |
+
"step": 2220
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 25.632183908045977,
|
| 1831 |
+
"grad_norm": 1.8201491832733154,
|
| 1832 |
+
"learning_rate": 8.088548318433377e-06,
|
| 1833 |
+
"loss": 0.0373,
|
| 1834 |
+
"step": 2230
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 25.74712643678161,
|
| 1838 |
+
"grad_norm": 0.5777905583381653,
|
| 1839 |
+
"learning_rate": 7.875691783737762e-06,
|
| 1840 |
+
"loss": 0.0323,
|
| 1841 |
+
"step": 2240
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 25.862068965517242,
|
| 1845 |
+
"grad_norm": 0.3212037980556488,
|
| 1846 |
+
"learning_rate": 7.662835249042145e-06,
|
| 1847 |
+
"loss": 0.0337,
|
| 1848 |
+
"step": 2250
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 25.977011494252874,
|
| 1852 |
+
"grad_norm": 1.0763506889343262,
|
| 1853 |
+
"learning_rate": 7.449978714346531e-06,
|
| 1854 |
+
"loss": 0.0455,
|
| 1855 |
+
"step": 2260
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 26.091954022988507,
|
| 1859 |
+
"grad_norm": 0.7833639979362488,
|
| 1860 |
+
"learning_rate": 7.237122179650915e-06,
|
| 1861 |
+
"loss": 0.0289,
|
| 1862 |
+
"step": 2270
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 26.20689655172414,
|
| 1866 |
+
"grad_norm": 1.0893276929855347,
|
| 1867 |
+
"learning_rate": 7.024265644955301e-06,
|
| 1868 |
+
"loss": 0.037,
|
| 1869 |
+
"step": 2280
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 26.32183908045977,
|
| 1873 |
+
"grad_norm": 0.534814715385437,
|
| 1874 |
+
"learning_rate": 6.811409110259685e-06,
|
| 1875 |
+
"loss": 0.0309,
|
| 1876 |
+
"step": 2290
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 26.436781609195403,
|
| 1880 |
+
"grad_norm": 1.0022532939910889,
|
| 1881 |
+
"learning_rate": 6.598552575564069e-06,
|
| 1882 |
+
"loss": 0.026,
|
| 1883 |
+
"step": 2300
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 26.436781609195403,
|
| 1887 |
+
"eval_accuracy": 0.9385507246376812,
|
| 1888 |
+
"eval_f1": 0.8515406162464986,
|
| 1889 |
+
"eval_loss": 0.16867290437221527,
|
| 1890 |
+
"eval_precision": 0.8611898016997167,
|
| 1891 |
+
"eval_recall": 0.8421052631578947,
|
| 1892 |
+
"eval_runtime": 3.2071,
|
| 1893 |
+
"eval_samples_per_second": 107.575,
|
| 1894 |
+
"eval_steps_per_second": 13.72,
|
| 1895 |
+
"step": 2300
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 26.551724137931036,
|
| 1899 |
+
"grad_norm": 0.396712988615036,
|
| 1900 |
+
"learning_rate": 6.385696040868455e-06,
|
| 1901 |
+
"loss": 0.0422,
|
| 1902 |
+
"step": 2310
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 26.666666666666668,
|
| 1906 |
+
"grad_norm": 0.6816790103912354,
|
| 1907 |
+
"learning_rate": 6.172839506172839e-06,
|
| 1908 |
+
"loss": 0.0259,
|
| 1909 |
+
"step": 2320
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 26.7816091954023,
|
| 1913 |
+
"grad_norm": 0.44069159030914307,
|
| 1914 |
+
"learning_rate": 5.959982971477224e-06,
|
| 1915 |
+
"loss": 0.0314,
|
| 1916 |
+
"step": 2330
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 26.896551724137932,
|
| 1920 |
+
"grad_norm": 0.8928768634796143,
|
| 1921 |
+
"learning_rate": 5.747126436781609e-06,
|
| 1922 |
+
"loss": 0.0451,
|
| 1923 |
+
"step": 2340
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 27.011494252873565,
|
| 1927 |
+
"grad_norm": 0.15559493005275726,
|
| 1928 |
+
"learning_rate": 5.534269902085994e-06,
|
| 1929 |
+
"loss": 0.0292,
|
| 1930 |
+
"step": 2350
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 27.126436781609197,
|
| 1934 |
+
"grad_norm": 0.57392817735672,
|
| 1935 |
+
"learning_rate": 5.321413367390379e-06,
|
| 1936 |
+
"loss": 0.0339,
|
| 1937 |
+
"step": 2360
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 27.24137931034483,
|
| 1941 |
+
"grad_norm": 0.4282006621360779,
|
| 1942 |
+
"learning_rate": 5.108556832694764e-06,
|
| 1943 |
+
"loss": 0.0272,
|
| 1944 |
+
"step": 2370
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 27.35632183908046,
|
| 1948 |
+
"grad_norm": 0.3765332102775574,
|
| 1949 |
+
"learning_rate": 4.895700297999148e-06,
|
| 1950 |
+
"loss": 0.0224,
|
| 1951 |
+
"step": 2380
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 27.47126436781609,
|
| 1955 |
+
"grad_norm": 0.26194748282432556,
|
| 1956 |
+
"learning_rate": 4.682843763303533e-06,
|
| 1957 |
+
"loss": 0.025,
|
| 1958 |
+
"step": 2390
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 27.586206896551722,
|
| 1962 |
+
"grad_norm": 0.6783250570297241,
|
| 1963 |
+
"learning_rate": 4.469987228607918e-06,
|
| 1964 |
+
"loss": 0.0353,
|
| 1965 |
+
"step": 2400
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 27.586206896551722,
|
| 1969 |
+
"eval_accuracy": 0.9402898550724638,
|
| 1970 |
+
"eval_f1": 0.8555399719495091,
|
| 1971 |
+
"eval_loss": 0.16655248403549194,
|
| 1972 |
+
"eval_precision": 0.8664772727272727,
|
| 1973 |
+
"eval_recall": 0.8448753462603878,
|
| 1974 |
+
"eval_runtime": 2.5159,
|
| 1975 |
+
"eval_samples_per_second": 137.126,
|
| 1976 |
+
"eval_steps_per_second": 17.489,
|
| 1977 |
+
"step": 2400
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 27.701149425287355,
|
| 1981 |
+
"grad_norm": 0.7945173382759094,
|
| 1982 |
+
"learning_rate": 4.257130693912303e-06,
|
| 1983 |
+
"loss": 0.0346,
|
| 1984 |
+
"step": 2410
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 27.816091954022987,
|
| 1988 |
+
"grad_norm": 0.34426242113113403,
|
| 1989 |
+
"learning_rate": 4.044274159216688e-06,
|
| 1990 |
+
"loss": 0.0354,
|
| 1991 |
+
"step": 2420
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 27.93103448275862,
|
| 1995 |
+
"grad_norm": 0.2353845238685608,
|
| 1996 |
+
"learning_rate": 3.8314176245210725e-06,
|
| 1997 |
+
"loss": 0.0306,
|
| 1998 |
+
"step": 2430
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 28.04597701149425,
|
| 2002 |
+
"grad_norm": 0.7292235493659973,
|
| 2003 |
+
"learning_rate": 3.6185610898254575e-06,
|
| 2004 |
+
"loss": 0.0268,
|
| 2005 |
+
"step": 2440
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 28.160919540229884,
|
| 2009 |
+
"grad_norm": 0.65985107421875,
|
| 2010 |
+
"learning_rate": 3.4057045551298425e-06,
|
| 2011 |
+
"loss": 0.0235,
|
| 2012 |
+
"step": 2450
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 28.275862068965516,
|
| 2016 |
+
"grad_norm": 0.36814385652542114,
|
| 2017 |
+
"learning_rate": 3.1928480204342275e-06,
|
| 2018 |
+
"loss": 0.0234,
|
| 2019 |
+
"step": 2460
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 28.39080459770115,
|
| 2023 |
+
"grad_norm": 0.2845553159713745,
|
| 2024 |
+
"learning_rate": 2.979991485738612e-06,
|
| 2025 |
+
"loss": 0.0249,
|
| 2026 |
+
"step": 2470
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 28.50574712643678,
|
| 2030 |
+
"grad_norm": 0.36483725905418396,
|
| 2031 |
+
"learning_rate": 2.767134951042997e-06,
|
| 2032 |
+
"loss": 0.0267,
|
| 2033 |
+
"step": 2480
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 28.620689655172413,
|
| 2037 |
+
"grad_norm": 0.5152461528778076,
|
| 2038 |
+
"learning_rate": 2.554278416347382e-06,
|
| 2039 |
+
"loss": 0.0325,
|
| 2040 |
+
"step": 2490
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 28.735632183908045,
|
| 2044 |
+
"grad_norm": 0.775412380695343,
|
| 2045 |
+
"learning_rate": 2.3414218816517667e-06,
|
| 2046 |
+
"loss": 0.0294,
|
| 2047 |
+
"step": 2500
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 28.735632183908045,
|
| 2051 |
+
"eval_accuracy": 0.9428985507246377,
|
| 2052 |
+
"eval_f1": 0.8613652357494722,
|
| 2053 |
+
"eval_loss": 0.1659679114818573,
|
| 2054 |
+
"eval_precision": 0.8755364806866953,
|
| 2055 |
+
"eval_recall": 0.8476454293628809,
|
| 2056 |
+
"eval_runtime": 2.5486,
|
| 2057 |
+
"eval_samples_per_second": 135.371,
|
| 2058 |
+
"eval_steps_per_second": 17.265,
|
| 2059 |
+
"step": 2500
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 28.850574712643677,
|
| 2063 |
+
"grad_norm": 0.25289997458457947,
|
| 2064 |
+
"learning_rate": 2.1285653469561517e-06,
|
| 2065 |
+
"loss": 0.0285,
|
| 2066 |
+
"step": 2510
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 28.96551724137931,
|
| 2070 |
+
"grad_norm": 0.11768297851085663,
|
| 2071 |
+
"learning_rate": 1.9157088122605362e-06,
|
| 2072 |
+
"loss": 0.0233,
|
| 2073 |
+
"step": 2520
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 29.080459770114942,
|
| 2077 |
+
"grad_norm": 0.4232161045074463,
|
| 2078 |
+
"learning_rate": 1.7028522775649212e-06,
|
| 2079 |
+
"loss": 0.021,
|
| 2080 |
+
"step": 2530
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 29.195402298850574,
|
| 2084 |
+
"grad_norm": 0.6776716709136963,
|
| 2085 |
+
"learning_rate": 1.489995742869306e-06,
|
| 2086 |
+
"loss": 0.0282,
|
| 2087 |
+
"step": 2540
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 29.310344827586206,
|
| 2091 |
+
"grad_norm": 0.2837054431438446,
|
| 2092 |
+
"learning_rate": 1.277139208173691e-06,
|
| 2093 |
+
"loss": 0.034,
|
| 2094 |
+
"step": 2550
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 29.42528735632184,
|
| 2098 |
+
"grad_norm": 0.13374456763267517,
|
| 2099 |
+
"learning_rate": 1.0642826734780758e-06,
|
| 2100 |
+
"loss": 0.0261,
|
| 2101 |
+
"step": 2560
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 29.54022988505747,
|
| 2105 |
+
"grad_norm": 0.32526296377182007,
|
| 2106 |
+
"learning_rate": 8.514261387824606e-07,
|
| 2107 |
+
"loss": 0.0318,
|
| 2108 |
+
"step": 2570
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 29.655172413793103,
|
| 2112 |
+
"grad_norm": 0.7919621467590332,
|
| 2113 |
+
"learning_rate": 6.385696040868455e-07,
|
| 2114 |
+
"loss": 0.0305,
|
| 2115 |
+
"step": 2580
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 29.770114942528735,
|
| 2119 |
+
"grad_norm": 0.1575266569852829,
|
| 2120 |
+
"learning_rate": 4.257130693912303e-07,
|
| 2121 |
+
"loss": 0.0298,
|
| 2122 |
+
"step": 2590
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 29.885057471264368,
|
| 2126 |
+
"grad_norm": 0.35859954357147217,
|
| 2127 |
+
"learning_rate": 2.1285653469561516e-07,
|
| 2128 |
+
"loss": 0.0243,
|
| 2129 |
+
"step": 2600
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 29.885057471264368,
|
| 2133 |
+
"eval_accuracy": 0.9423188405797102,
|
| 2134 |
+
"eval_f1": 0.8589652728561304,
|
| 2135 |
+
"eval_loss": 0.16644859313964844,
|
| 2136 |
+
"eval_precision": 0.8795355587808418,
|
| 2137 |
+
"eval_recall": 0.8393351800554016,
|
| 2138 |
+
"eval_runtime": 2.5418,
|
| 2139 |
+
"eval_samples_per_second": 135.729,
|
| 2140 |
+
"eval_steps_per_second": 17.31,
|
| 2141 |
+
"step": 2600
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 30.0,
|
| 2145 |
+
"grad_norm": 0.5356839299201965,
|
| 2146 |
+
"learning_rate": 0.0,
|
| 2147 |
+
"loss": 0.0375,
|
| 2148 |
+
"step": 2610
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 30.0,
|
| 2152 |
+
"step": 2610,
|
| 2153 |
+
"total_flos": 3.2060734740537754e+18,
|
| 2154 |
+
"train_loss": 0.09901393407606074,
|
| 2155 |
+
"train_runtime": 730.7896,
|
| 2156 |
+
"train_samples_per_second": 56.61,
|
| 2157 |
+
"train_steps_per_second": 3.571
|
| 2158 |
+
}
|
| 2159 |
+
],
|
| 2160 |
+
"logging_steps": 10,
|
| 2161 |
+
"max_steps": 2610,
|
| 2162 |
+
"num_input_tokens_seen": 0,
|
| 2163 |
+
"num_train_epochs": 30,
|
| 2164 |
+
"save_steps": 500,
|
| 2165 |
+
"stateful_callbacks": {
|
| 2166 |
+
"TrainerControl": {
|
| 2167 |
+
"args": {
|
| 2168 |
+
"should_epoch_stop": false,
|
| 2169 |
+
"should_evaluate": false,
|
| 2170 |
+
"should_log": false,
|
| 2171 |
+
"should_save": true,
|
| 2172 |
+
"should_training_stop": true
|
| 2173 |
+
},
|
| 2174 |
+
"attributes": {}
|
| 2175 |
+
}
|
| 2176 |
+
},
|
| 2177 |
+
"total_flos": 3.2060734740537754e+18,
|
| 2178 |
+
"train_batch_size": 16,
|
| 2179 |
+
"trial_name": null,
|
| 2180 |
+
"trial_params": null
|
| 2181 |
+
}
|