Commit
·
5ab3f00
1
Parent(s):
7d28d2e
Training in progress, step 200
Browse files- all_results.json +16 -16
- eval_results.json +6 -6
- pytorch_model.bin +1 -1
- runs/Feb13_22-56-31_node1/events.out.tfevents.1644764371.node1 +2 -2
- runs/Feb13_22-56-31_node1/events.out.tfevents.1644764957.node1 +3 -0
- runs/Feb13_23-17-27_node1/1644765604.990227/events.out.tfevents.1644765605.node1 +3 -0
- runs/Feb13_23-17-27_node1/events.out.tfevents.1644765604.node1 +3 -0
- runs/Feb13_23-22-36_node1/1644765898.9699917/events.out.tfevents.1644765898.node1 +3 -0
- runs/Feb13_23-22-36_node1/events.out.tfevents.1644765898.node1 +3 -0
- test_results.json +6 -6
- train_results.json +4 -4
- trainer_state.json +784 -40
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,23 +1,23 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
-
"eval_accuracy":
|
| 4 |
-
"eval_average_metrics":
|
| 5 |
"eval_combined_score": 0.7652279521674141,
|
| 6 |
-
"eval_f1":
|
| 7 |
-
"eval_loss": 0.
|
| 8 |
-
"eval_runtime":
|
| 9 |
"eval_samples": 408,
|
| 10 |
-
"eval_samples_per_second":
|
| 11 |
"eval_steps_per_second": 14.442,
|
| 12 |
-
"test_accuracy":
|
| 13 |
-
"test_average_metrics":
|
| 14 |
-
"test_f1":
|
| 15 |
-
"test_loss": 0.
|
| 16 |
-
"test_runtime":
|
| 17 |
-
"test_samples_per_second":
|
| 18 |
-
"train_loss": 0.
|
| 19 |
-
"train_runtime":
|
| 20 |
"train_samples": 3668,
|
| 21 |
-
"train_samples_per_second":
|
| 22 |
-
"train_steps_per_second":
|
| 23 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.0,
|
| 4 |
+
"eval_average_metrics": 0.0,
|
| 5 |
"eval_combined_score": 0.7652279521674141,
|
| 6 |
+
"eval_f1": 0.0,
|
| 7 |
+
"eval_loss": 0.20416393876075745,
|
| 8 |
+
"eval_runtime": 1.8778,
|
| 9 |
"eval_samples": 408,
|
| 10 |
+
"eval_samples_per_second": 108.639,
|
| 11 |
"eval_steps_per_second": 14.442,
|
| 12 |
+
"test_accuracy": 0.0,
|
| 13 |
+
"test_average_metrics": 0.0,
|
| 14 |
+
"test_f1": 0.0,
|
| 15 |
+
"test_loss": 0.1770762801170349,
|
| 16 |
+
"test_runtime": 1.5085,
|
| 17 |
+
"test_samples_per_second": 135.238,
|
| 18 |
+
"train_loss": 0.03386630766055595,
|
| 19 |
+
"train_runtime": 583.9583,
|
| 20 |
"train_samples": 3668,
|
| 21 |
+
"train_samples_per_second": 125.625,
|
| 22 |
+
"train_steps_per_second": 1.986
|
| 23 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
-
"eval_accuracy":
|
| 4 |
-
"eval_average_metrics":
|
| 5 |
-
"eval_f1":
|
| 6 |
-
"eval_loss": 0.
|
| 7 |
-
"eval_runtime":
|
| 8 |
-
"eval_samples_per_second":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.0,
|
| 4 |
+
"eval_average_metrics": 0.0,
|
| 5 |
+
"eval_f1": 0.0,
|
| 6 |
+
"eval_loss": 0.20416393876075745,
|
| 7 |
+
"eval_runtime": 1.8778,
|
| 8 |
+
"eval_samples_per_second": 108.639
|
| 9 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7551621
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b09e19d290866a921a7cc2616939f5dd9656a5f74c49c3237fddda3484eeed7f
|
| 3 |
size 7551621
|
runs/Feb13_22-56-31_node1/events.out.tfevents.1644764371.node1
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4cd309cdec289c22d17acf2860f18494f01944764852783d3cb158f00c0165c
|
| 3 |
+
size 26870
|
runs/Feb13_22-56-31_node1/events.out.tfevents.1644764957.node1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8bf1b50853f805eb0176a67e04c230c52c43f5ceaf93c00714284789985ca342
|
| 3 |
+
size 776
|
runs/Feb13_23-17-27_node1/1644765604.990227/events.out.tfevents.1644765605.node1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36f1cfd03c4861da5e4d99c214c91c07d5cbb8c3384f884a29a59df68df33498
|
| 3 |
+
size 5011
|
runs/Feb13_23-17-27_node1/events.out.tfevents.1644765604.node1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58d4c2d9ed6ec309e7ed50089f7c1741fd3aef3127c0264aacdab67b2a453a48
|
| 3 |
+
size 5839
|
runs/Feb13_23-22-36_node1/1644765898.9699917/events.out.tfevents.1644765898.node1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f92d5e16e1da9d913a0b12e83162007c5b98744ea2aca8114165390461b5eafd
|
| 3 |
+
size 5011
|
runs/Feb13_23-22-36_node1/events.out.tfevents.1644765898.node1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eade3c061dc82bd40aa8e6a09a2862a505334def00d6ccd0879556d8db16628d
|
| 3 |
+
size 7764
|
test_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
-
"test_accuracy":
|
| 4 |
-
"test_average_metrics":
|
| 5 |
-
"test_f1":
|
| 6 |
-
"test_loss": 0.
|
| 7 |
-
"test_runtime":
|
| 8 |
-
"test_samples_per_second":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
+
"test_accuracy": 0.0,
|
| 4 |
+
"test_average_metrics": 0.0,
|
| 5 |
+
"test_f1": 0.0,
|
| 6 |
+
"test_loss": 0.1770762801170349,
|
| 7 |
+
"test_runtime": 1.5085,
|
| 8 |
+
"test_samples_per_second": 135.238
|
| 9 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 3668,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
+
"train_loss": 0.03386630766055595,
|
| 4 |
+
"train_runtime": 583.9583,
|
| 5 |
"train_samples": 3668,
|
| 6 |
+
"train_samples_per_second": 125.625,
|
| 7 |
+
"train_steps_per_second": 1.986
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,87 +1,831 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "outputs/bitfit/t5-base/mrpc/checkpoint-
|
| 4 |
"epoch": 20.0,
|
| 5 |
"global_step": 1160,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
{
|
| 11 |
"epoch": 3.45,
|
| 12 |
-
"
|
| 13 |
-
"
|
| 14 |
-
"eval_f1": 90.84507042253522,
|
| 15 |
-
"eval_loss": 0.15426486730575562,
|
| 16 |
-
"eval_runtime": 1.9622,
|
| 17 |
-
"eval_samples_per_second": 103.964,
|
| 18 |
"step": 200
|
| 19 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
{
|
| 21 |
"epoch": 6.9,
|
| 22 |
-
"eval_accuracy":
|
| 23 |
-
"eval_average_metrics":
|
| 24 |
-
"eval_f1":
|
| 25 |
-
"eval_loss": 0.
|
| 26 |
-
"eval_runtime":
|
| 27 |
-
"eval_samples_per_second":
|
| 28 |
"step": 400
|
| 29 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
{
|
| 31 |
"epoch": 8.62,
|
| 32 |
"learning_rate": 0.0001706896551724138,
|
| 33 |
-
"loss": 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
"step": 500
|
| 35 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
{
|
| 37 |
"epoch": 10.34,
|
| 38 |
-
"eval_accuracy":
|
| 39 |
-
"eval_average_metrics":
|
| 40 |
-
"eval_f1":
|
| 41 |
-
"eval_loss": 0.
|
| 42 |
-
"eval_runtime":
|
| 43 |
-
"eval_samples_per_second":
|
| 44 |
"step": 600
|
| 45 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
{
|
| 47 |
"epoch": 13.79,
|
| 48 |
-
"
|
| 49 |
-
"
|
| 50 |
-
"eval_f1": 91.54929577464787,
|
| 51 |
-
"eval_loss": 0.17383529245853424,
|
| 52 |
-
"eval_runtime": 0.718,
|
| 53 |
-
"eval_samples_per_second": 284.125,
|
| 54 |
"step": 800
|
| 55 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
{
|
| 57 |
"epoch": 17.24,
|
| 58 |
"learning_rate": 4.137931034482758e-05,
|
| 59 |
-
"loss": 0.
|
| 60 |
"step": 1000
|
| 61 |
},
|
| 62 |
{
|
| 63 |
"epoch": 17.24,
|
| 64 |
-
"eval_accuracy":
|
| 65 |
-
"eval_average_metrics":
|
| 66 |
-
"eval_f1":
|
| 67 |
-
"eval_loss": 0.
|
| 68 |
-
"eval_runtime":
|
| 69 |
-
"eval_samples_per_second":
|
| 70 |
"step": 1000
|
| 71 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 72 |
{
|
| 73 |
"epoch": 20.0,
|
| 74 |
"step": 1160,
|
| 75 |
-
"total_flos":
|
| 76 |
-
"train_loss": 0.
|
| 77 |
-
"train_runtime":
|
| 78 |
-
"train_samples_per_second":
|
| 79 |
-
"train_steps_per_second":
|
| 80 |
}
|
| 81 |
],
|
| 82 |
"max_steps": 1160,
|
| 83 |
"num_train_epochs": 20,
|
| 84 |
-
"total_flos":
|
| 85 |
"trial_name": null,
|
| 86 |
"trial_params": null
|
| 87 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.0,
|
| 3 |
+
"best_model_checkpoint": "outputs/bitfit/t5-base/mrpc/checkpoint-200",
|
| 4 |
"epoch": 20.0,
|
| 5 |
"global_step": 1160,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.17,
|
| 12 |
+
"learning_rate": 0.00029741379310344827,
|
| 13 |
+
"loss": 2.2211,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.34,
|
| 18 |
+
"learning_rate": 0.0002948275862068965,
|
| 19 |
+
"loss": 0.194,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.52,
|
| 24 |
+
"learning_rate": 0.0002922413793103448,
|
| 25 |
+
"loss": 0.1031,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.69,
|
| 30 |
+
"learning_rate": 0.0002896551724137931,
|
| 31 |
+
"loss": 0.121,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.86,
|
| 36 |
+
"learning_rate": 0.0002870689655172413,
|
| 37 |
+
"loss": 0.1258,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.03,
|
| 42 |
+
"learning_rate": 0.0002844827586206896,
|
| 43 |
+
"loss": 0.0988,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.21,
|
| 48 |
+
"learning_rate": 0.0002818965517241379,
|
| 49 |
+
"loss": 0.0688,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.38,
|
| 54 |
+
"learning_rate": 0.0002793103448275862,
|
| 55 |
+
"loss": 0.081,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.55,
|
| 60 |
+
"learning_rate": 0.0002767241379310345,
|
| 61 |
+
"loss": 0.0719,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.72,
|
| 66 |
+
"learning_rate": 0.0002741379310344827,
|
| 67 |
+
"loss": 0.0606,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.72,
|
| 72 |
+
"eval_accuracy": 0.0,
|
| 73 |
+
"eval_average_metrics": 0.0,
|
| 74 |
+
"eval_f1": 0.0,
|
| 75 |
+
"eval_loss": 0.157722607254982,
|
| 76 |
+
"eval_runtime": 1.6787,
|
| 77 |
+
"eval_samples_per_second": 121.522,
|
| 78 |
+
"step": 100
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.9,
|
| 82 |
+
"learning_rate": 0.000271551724137931,
|
| 83 |
+
"loss": 0.0726,
|
| 84 |
+
"step": 110
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 2.07,
|
| 88 |
+
"learning_rate": 0.0002689655172413793,
|
| 89 |
+
"loss": 0.0686,
|
| 90 |
+
"step": 120
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 2.24,
|
| 94 |
+
"learning_rate": 0.0002663793103448276,
|
| 95 |
+
"loss": 0.0445,
|
| 96 |
+
"step": 130
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 2.41,
|
| 100 |
+
"learning_rate": 0.00026379310344827584,
|
| 101 |
+
"loss": 0.0414,
|
| 102 |
+
"step": 140
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 2.59,
|
| 106 |
+
"learning_rate": 0.00026120689655172413,
|
| 107 |
+
"loss": 0.0431,
|
| 108 |
+
"step": 150
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 2.76,
|
| 112 |
+
"learning_rate": 0.00025862068965517237,
|
| 113 |
+
"loss": 0.0274,
|
| 114 |
+
"step": 160
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 2.93,
|
| 118 |
+
"learning_rate": 0.00025603448275862066,
|
| 119 |
+
"loss": 0.0399,
|
| 120 |
+
"step": 170
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 3.1,
|
| 124 |
+
"learning_rate": 0.00025344827586206895,
|
| 125 |
+
"loss": 0.032,
|
| 126 |
+
"step": 180
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 3.28,
|
| 130 |
+
"learning_rate": 0.0002508620689655172,
|
| 131 |
+
"loss": 0.0163,
|
| 132 |
+
"step": 190
|
| 133 |
+
},
|
| 134 |
{
|
| 135 |
"epoch": 3.45,
|
| 136 |
+
"learning_rate": 0.0002482758620689655,
|
| 137 |
+
"loss": 0.0196,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
"step": 200
|
| 139 |
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 3.45,
|
| 142 |
+
"eval_accuracy": 0.0,
|
| 143 |
+
"eval_average_metrics": 0.0,
|
| 144 |
+
"eval_f1": 0.0,
|
| 145 |
+
"eval_loss": 0.20416393876075745,
|
| 146 |
+
"eval_runtime": 1.3232,
|
| 147 |
+
"eval_samples_per_second": 154.176,
|
| 148 |
+
"step": 200
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 3.62,
|
| 152 |
+
"learning_rate": 0.00024568965517241377,
|
| 153 |
+
"loss": 0.0236,
|
| 154 |
+
"step": 210
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 3.79,
|
| 158 |
+
"learning_rate": 0.00024310344827586203,
|
| 159 |
+
"loss": 0.0354,
|
| 160 |
+
"step": 220
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 3.97,
|
| 164 |
+
"learning_rate": 0.00024051724137931033,
|
| 165 |
+
"loss": 0.0277,
|
| 166 |
+
"step": 230
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 4.14,
|
| 170 |
+
"learning_rate": 0.00023793103448275862,
|
| 171 |
+
"loss": 0.0194,
|
| 172 |
+
"step": 240
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 4.31,
|
| 176 |
+
"learning_rate": 0.00023534482758620685,
|
| 177 |
+
"loss": 0.0153,
|
| 178 |
+
"step": 250
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 4.48,
|
| 182 |
+
"learning_rate": 0.00023275862068965515,
|
| 183 |
+
"loss": 0.0129,
|
| 184 |
+
"step": 260
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 4.66,
|
| 188 |
+
"learning_rate": 0.00023017241379310344,
|
| 189 |
+
"loss": 0.0112,
|
| 190 |
+
"step": 270
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 4.83,
|
| 194 |
+
"learning_rate": 0.0002275862068965517,
|
| 195 |
+
"loss": 0.0083,
|
| 196 |
+
"step": 280
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 5.0,
|
| 200 |
+
"learning_rate": 0.000225,
|
| 201 |
+
"loss": 0.0167,
|
| 202 |
+
"step": 290
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 5.17,
|
| 206 |
+
"learning_rate": 0.00022241379310344826,
|
| 207 |
+
"loss": 0.0184,
|
| 208 |
+
"step": 300
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 5.17,
|
| 212 |
+
"eval_accuracy": 0.0,
|
| 213 |
+
"eval_average_metrics": 0.0,
|
| 214 |
+
"eval_f1": 0.0,
|
| 215 |
+
"eval_loss": 0.2221006453037262,
|
| 216 |
+
"eval_runtime": 1.4478,
|
| 217 |
+
"eval_samples_per_second": 140.908,
|
| 218 |
+
"step": 300
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 5.34,
|
| 222 |
+
"learning_rate": 0.00021982758620689652,
|
| 223 |
+
"loss": 0.0053,
|
| 224 |
+
"step": 310
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 5.52,
|
| 228 |
+
"learning_rate": 0.00021724137931034481,
|
| 229 |
+
"loss": 0.0049,
|
| 230 |
+
"step": 320
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 5.69,
|
| 234 |
+
"learning_rate": 0.0002146551724137931,
|
| 235 |
+
"loss": 0.0096,
|
| 236 |
+
"step": 330
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 5.86,
|
| 240 |
+
"learning_rate": 0.00021206896551724134,
|
| 241 |
+
"loss": 0.008,
|
| 242 |
+
"step": 340
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 6.03,
|
| 246 |
+
"learning_rate": 0.00020948275862068963,
|
| 247 |
+
"loss": 0.0099,
|
| 248 |
+
"step": 350
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 6.21,
|
| 252 |
+
"learning_rate": 0.00020689655172413793,
|
| 253 |
+
"loss": 0.0128,
|
| 254 |
+
"step": 360
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 6.38,
|
| 258 |
+
"learning_rate": 0.0002043103448275862,
|
| 259 |
+
"loss": 0.0026,
|
| 260 |
+
"step": 370
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 6.55,
|
| 264 |
+
"learning_rate": 0.00020172413793103448,
|
| 265 |
+
"loss": 0.002,
|
| 266 |
+
"step": 380
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 6.72,
|
| 270 |
+
"learning_rate": 0.00019913793103448275,
|
| 271 |
+
"loss": 0.0062,
|
| 272 |
+
"step": 390
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 6.9,
|
| 276 |
+
"learning_rate": 0.000196551724137931,
|
| 277 |
+
"loss": 0.0099,
|
| 278 |
+
"step": 400
|
| 279 |
+
},
|
| 280 |
{
|
| 281 |
"epoch": 6.9,
|
| 282 |
+
"eval_accuracy": 0.0,
|
| 283 |
+
"eval_average_metrics": 0.0,
|
| 284 |
+
"eval_f1": 0.0,
|
| 285 |
+
"eval_loss": 0.24928051233291626,
|
| 286 |
+
"eval_runtime": 1.0803,
|
| 287 |
+
"eval_samples_per_second": 188.836,
|
| 288 |
"step": 400
|
| 289 |
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 7.07,
|
| 292 |
+
"learning_rate": 0.0001939655172413793,
|
| 293 |
+
"loss": 0.0138,
|
| 294 |
+
"step": 410
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 7.24,
|
| 298 |
+
"learning_rate": 0.0001913793103448276,
|
| 299 |
+
"loss": 0.0016,
|
| 300 |
+
"step": 420
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 7.41,
|
| 304 |
+
"learning_rate": 0.00018879310344827583,
|
| 305 |
+
"loss": 0.0044,
|
| 306 |
+
"step": 430
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 7.59,
|
| 310 |
+
"learning_rate": 0.00018620689655172412,
|
| 311 |
+
"loss": 0.0039,
|
| 312 |
+
"step": 440
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 7.76,
|
| 316 |
+
"learning_rate": 0.0001836206896551724,
|
| 317 |
+
"loss": 0.0063,
|
| 318 |
+
"step": 450
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 7.93,
|
| 322 |
+
"learning_rate": 0.00018103448275862068,
|
| 323 |
+
"loss": 0.0076,
|
| 324 |
+
"step": 460
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 8.1,
|
| 328 |
+
"learning_rate": 0.00017844827586206897,
|
| 329 |
+
"loss": 0.005,
|
| 330 |
+
"step": 470
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 8.28,
|
| 334 |
+
"learning_rate": 0.0001758620689655172,
|
| 335 |
+
"loss": 0.0018,
|
| 336 |
+
"step": 480
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 8.45,
|
| 340 |
+
"learning_rate": 0.0001732758620689655,
|
| 341 |
+
"loss": 0.0033,
|
| 342 |
+
"step": 490
|
| 343 |
+
},
|
| 344 |
{
|
| 345 |
"epoch": 8.62,
|
| 346 |
"learning_rate": 0.0001706896551724138,
|
| 347 |
+
"loss": 0.0049,
|
| 348 |
+
"step": 500
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 8.62,
|
| 352 |
+
"eval_accuracy": 0.0,
|
| 353 |
+
"eval_average_metrics": 0.0,
|
| 354 |
+
"eval_f1": 0.0,
|
| 355 |
+
"eval_loss": 0.26071295142173767,
|
| 356 |
+
"eval_runtime": 1.6002,
|
| 357 |
+
"eval_samples_per_second": 127.481,
|
| 358 |
"step": 500
|
| 359 |
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 8.79,
|
| 362 |
+
"learning_rate": 0.00016810344827586203,
|
| 363 |
+
"loss": 0.0027,
|
| 364 |
+
"step": 510
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 8.97,
|
| 368 |
+
"learning_rate": 0.00016551724137931032,
|
| 369 |
+
"loss": 0.0052,
|
| 370 |
+
"step": 520
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 9.14,
|
| 374 |
+
"learning_rate": 0.0001629310344827586,
|
| 375 |
+
"loss": 0.0003,
|
| 376 |
+
"step": 530
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 9.31,
|
| 380 |
+
"learning_rate": 0.00016034482758620688,
|
| 381 |
+
"loss": 0.0017,
|
| 382 |
+
"step": 540
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 9.48,
|
| 386 |
+
"learning_rate": 0.00015775862068965517,
|
| 387 |
+
"loss": 0.0016,
|
| 388 |
+
"step": 550
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 9.66,
|
| 392 |
+
"learning_rate": 0.00015517241379310346,
|
| 393 |
+
"loss": 0.0037,
|
| 394 |
+
"step": 560
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 9.83,
|
| 398 |
+
"learning_rate": 0.0001525862068965517,
|
| 399 |
+
"loss": 0.0008,
|
| 400 |
+
"step": 570
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 10.0,
|
| 404 |
+
"learning_rate": 0.00015,
|
| 405 |
+
"loss": 0.0065,
|
| 406 |
+
"step": 580
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 10.17,
|
| 410 |
+
"learning_rate": 0.00014741379310344825,
|
| 411 |
+
"loss": 0.0005,
|
| 412 |
+
"step": 590
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 10.34,
|
| 416 |
+
"learning_rate": 0.00014482758620689654,
|
| 417 |
+
"loss": 0.0012,
|
| 418 |
+
"step": 600
|
| 419 |
+
},
|
| 420 |
{
|
| 421 |
"epoch": 10.34,
|
| 422 |
+
"eval_accuracy": 0.0,
|
| 423 |
+
"eval_average_metrics": 0.0,
|
| 424 |
+
"eval_f1": 0.0,
|
| 425 |
+
"eval_loss": 0.3229135274887085,
|
| 426 |
+
"eval_runtime": 1.6354,
|
| 427 |
+
"eval_samples_per_second": 124.74,
|
| 428 |
"step": 600
|
| 429 |
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 10.52,
|
| 432 |
+
"learning_rate": 0.0001422413793103448,
|
| 433 |
+
"loss": 0.0015,
|
| 434 |
+
"step": 610
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 10.69,
|
| 438 |
+
"learning_rate": 0.0001396551724137931,
|
| 439 |
+
"loss": 0.0031,
|
| 440 |
+
"step": 620
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 10.86,
|
| 444 |
+
"learning_rate": 0.00013706896551724136,
|
| 445 |
+
"loss": 0.009,
|
| 446 |
+
"step": 630
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 11.03,
|
| 450 |
+
"learning_rate": 0.00013448275862068965,
|
| 451 |
+
"loss": 0.0008,
|
| 452 |
+
"step": 640
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 11.21,
|
| 456 |
+
"learning_rate": 0.00013189655172413792,
|
| 457 |
+
"loss": 0.0002,
|
| 458 |
+
"step": 650
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 11.38,
|
| 462 |
+
"learning_rate": 0.00012931034482758618,
|
| 463 |
+
"loss": 0.0002,
|
| 464 |
+
"step": 660
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 11.55,
|
| 468 |
+
"learning_rate": 0.00012672413793103447,
|
| 469 |
+
"loss": 0.004,
|
| 470 |
+
"step": 670
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 11.72,
|
| 474 |
+
"learning_rate": 0.00012413793103448274,
|
| 475 |
+
"loss": 0.004,
|
| 476 |
+
"step": 680
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 11.9,
|
| 480 |
+
"learning_rate": 0.00012155172413793102,
|
| 481 |
+
"loss": 0.0005,
|
| 482 |
+
"step": 690
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 12.07,
|
| 486 |
+
"learning_rate": 0.00011896551724137931,
|
| 487 |
+
"loss": 0.003,
|
| 488 |
+
"step": 700
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 12.07,
|
| 492 |
+
"eval_accuracy": 0.0,
|
| 493 |
+
"eval_average_metrics": 0.0,
|
| 494 |
+
"eval_f1": 0.0,
|
| 495 |
+
"eval_loss": 0.3359452784061432,
|
| 496 |
+
"eval_runtime": 1.9034,
|
| 497 |
+
"eval_samples_per_second": 107.177,
|
| 498 |
+
"step": 700
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 12.24,
|
| 502 |
+
"learning_rate": 0.00011637931034482757,
|
| 503 |
+
"loss": 0.0005,
|
| 504 |
+
"step": 710
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 12.41,
|
| 508 |
+
"learning_rate": 0.00011379310344827585,
|
| 509 |
+
"loss": 0.0002,
|
| 510 |
+
"step": 720
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 12.59,
|
| 514 |
+
"learning_rate": 0.00011120689655172413,
|
| 515 |
+
"loss": 0.0001,
|
| 516 |
+
"step": 730
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 12.76,
|
| 520 |
+
"learning_rate": 0.00010862068965517241,
|
| 521 |
+
"loss": 0.0024,
|
| 522 |
+
"step": 740
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 12.93,
|
| 526 |
+
"learning_rate": 0.00010603448275862067,
|
| 527 |
+
"loss": 0.0002,
|
| 528 |
+
"step": 750
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 13.1,
|
| 532 |
+
"learning_rate": 0.00010344827586206896,
|
| 533 |
+
"loss": 0.0003,
|
| 534 |
+
"step": 760
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 13.28,
|
| 538 |
+
"learning_rate": 0.00010086206896551724,
|
| 539 |
+
"loss": 0.0007,
|
| 540 |
+
"step": 770
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 13.45,
|
| 544 |
+
"learning_rate": 9.82758620689655e-05,
|
| 545 |
+
"loss": 0.0,
|
| 546 |
+
"step": 780
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 13.62,
|
| 550 |
+
"learning_rate": 9.56896551724138e-05,
|
| 551 |
+
"loss": 0.0001,
|
| 552 |
+
"step": 790
|
| 553 |
+
},
|
| 554 |
{
|
| 555 |
"epoch": 13.79,
|
| 556 |
+
"learning_rate": 9.310344827586206e-05,
|
| 557 |
+
"loss": 0.0003,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 558 |
"step": 800
|
| 559 |
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 13.79,
|
| 562 |
+
"eval_accuracy": 0.0,
|
| 563 |
+
"eval_average_metrics": 0.0,
|
| 564 |
+
"eval_f1": 0.0,
|
| 565 |
+
"eval_loss": 0.40144264698028564,
|
| 566 |
+
"eval_runtime": 1.2399,
|
| 567 |
+
"eval_samples_per_second": 164.525,
|
| 568 |
+
"step": 800
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 13.97,
|
| 572 |
+
"learning_rate": 9.051724137931034e-05,
|
| 573 |
+
"loss": 0.003,
|
| 574 |
+
"step": 810
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 14.14,
|
| 578 |
+
"learning_rate": 8.79310344827586e-05,
|
| 579 |
+
"loss": 0.0,
|
| 580 |
+
"step": 820
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 14.31,
|
| 584 |
+
"learning_rate": 8.53448275862069e-05,
|
| 585 |
+
"loss": 0.0004,
|
| 586 |
+
"step": 830
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 14.48,
|
| 590 |
+
"learning_rate": 8.275862068965516e-05,
|
| 591 |
+
"loss": 0.0001,
|
| 592 |
+
"step": 840
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 14.66,
|
| 596 |
+
"learning_rate": 8.017241379310344e-05,
|
| 597 |
+
"loss": 0.0001,
|
| 598 |
+
"step": 850
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 14.83,
|
| 602 |
+
"learning_rate": 7.758620689655173e-05,
|
| 603 |
+
"loss": 0.0008,
|
| 604 |
+
"step": 860
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 15.0,
|
| 608 |
+
"learning_rate": 7.5e-05,
|
| 609 |
+
"loss": 0.0001,
|
| 610 |
+
"step": 870
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 15.17,
|
| 614 |
+
"learning_rate": 7.241379310344827e-05,
|
| 615 |
+
"loss": 0.0001,
|
| 616 |
+
"step": 880
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 15.34,
|
| 620 |
+
"learning_rate": 6.982758620689655e-05,
|
| 621 |
+
"loss": 0.0018,
|
| 622 |
+
"step": 890
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 15.52,
|
| 626 |
+
"learning_rate": 6.724137931034483e-05,
|
| 627 |
+
"loss": 0.0001,
|
| 628 |
+
"step": 900
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 15.52,
|
| 632 |
+
"eval_accuracy": 0.0,
|
| 633 |
+
"eval_average_metrics": 0.0,
|
| 634 |
+
"eval_f1": 0.0,
|
| 635 |
+
"eval_loss": 0.37248921394348145,
|
| 636 |
+
"eval_runtime": 1.2617,
|
| 637 |
+
"eval_samples_per_second": 161.691,
|
| 638 |
+
"step": 900
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 15.69,
|
| 642 |
+
"learning_rate": 6.465517241379309e-05,
|
| 643 |
+
"loss": 0.0,
|
| 644 |
+
"step": 910
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 15.86,
|
| 648 |
+
"learning_rate": 6.206896551724137e-05,
|
| 649 |
+
"loss": 0.0,
|
| 650 |
+
"step": 920
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 16.03,
|
| 654 |
+
"learning_rate": 5.9482758620689654e-05,
|
| 655 |
+
"loss": 0.0001,
|
| 656 |
+
"step": 930
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 16.21,
|
| 660 |
+
"learning_rate": 5.6896551724137926e-05,
|
| 661 |
+
"loss": 0.0007,
|
| 662 |
+
"step": 940
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 16.38,
|
| 666 |
+
"learning_rate": 5.4310344827586204e-05,
|
| 667 |
+
"loss": 0.0,
|
| 668 |
+
"step": 950
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 16.55,
|
| 672 |
+
"learning_rate": 5.172413793103448e-05,
|
| 673 |
+
"loss": 0.0,
|
| 674 |
+
"step": 960
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 16.72,
|
| 678 |
+
"learning_rate": 4.913793103448275e-05,
|
| 679 |
+
"loss": 0.0,
|
| 680 |
+
"step": 970
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 16.9,
|
| 684 |
+
"learning_rate": 4.655172413793103e-05,
|
| 685 |
+
"loss": 0.0001,
|
| 686 |
+
"step": 980
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 17.07,
|
| 690 |
+
"learning_rate": 4.39655172413793e-05,
|
| 691 |
+
"loss": 0.0,
|
| 692 |
+
"step": 990
|
| 693 |
+
},
|
| 694 |
{
|
| 695 |
"epoch": 17.24,
|
| 696 |
"learning_rate": 4.137931034482758e-05,
|
| 697 |
+
"loss": 0.0001,
|
| 698 |
"step": 1000
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 17.24,
|
| 702 |
+
"eval_accuracy": 0.0,
|
| 703 |
+
"eval_average_metrics": 0.0,
|
| 704 |
+
"eval_f1": 0.0,
|
| 705 |
+
"eval_loss": 0.38377830386161804,
|
| 706 |
+
"eval_runtime": 1.3412,
|
| 707 |
+
"eval_samples_per_second": 152.099,
|
| 708 |
"step": 1000
|
| 709 |
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 17.41,
|
| 712 |
+
"learning_rate": 3.8793103448275865e-05,
|
| 713 |
+
"loss": 0.0,
|
| 714 |
+
"step": 1010
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 17.59,
|
| 718 |
+
"learning_rate": 3.6206896551724136e-05,
|
| 719 |
+
"loss": 0.0,
|
| 720 |
+
"step": 1020
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 17.76,
|
| 724 |
+
"learning_rate": 3.3620689655172414e-05,
|
| 725 |
+
"loss": 0.0001,
|
| 726 |
+
"step": 1030
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 17.93,
|
| 730 |
+
"learning_rate": 3.1034482758620685e-05,
|
| 731 |
+
"loss": 0.0004,
|
| 732 |
+
"step": 1040
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 18.1,
|
| 736 |
+
"learning_rate": 2.8448275862068963e-05,
|
| 737 |
+
"loss": 0.0,
|
| 738 |
+
"step": 1050
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 18.28,
|
| 742 |
+
"learning_rate": 2.586206896551724e-05,
|
| 743 |
+
"loss": 0.0001,
|
| 744 |
+
"step": 1060
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 18.45,
|
| 748 |
+
"learning_rate": 2.3275862068965515e-05,
|
| 749 |
+
"loss": 0.0002,
|
| 750 |
+
"step": 1070
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 18.62,
|
| 754 |
+
"learning_rate": 2.068965517241379e-05,
|
| 755 |
+
"loss": 0.0,
|
| 756 |
+
"step": 1080
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 18.79,
|
| 760 |
+
"learning_rate": 1.8103448275862068e-05,
|
| 761 |
+
"loss": 0.0,
|
| 762 |
+
"step": 1090
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 18.97,
|
| 766 |
+
"learning_rate": 1.5517241379310342e-05,
|
| 767 |
+
"loss": 0.0,
|
| 768 |
+
"step": 1100
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 18.97,
|
| 772 |
+
"eval_accuracy": 0.0,
|
| 773 |
+
"eval_average_metrics": 0.0,
|
| 774 |
+
"eval_f1": 0.0,
|
| 775 |
+
"eval_loss": 0.4036322236061096,
|
| 776 |
+
"eval_runtime": 1.3625,
|
| 777 |
+
"eval_samples_per_second": 149.728,
|
| 778 |
+
"step": 1100
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 19.14,
|
| 782 |
+
"learning_rate": 1.293103448275862e-05,
|
| 783 |
+
"loss": 0.0,
|
| 784 |
+
"step": 1110
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 19.31,
|
| 788 |
+
"learning_rate": 1.0344827586206895e-05,
|
| 789 |
+
"loss": 0.0,
|
| 790 |
+
"step": 1120
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 19.48,
|
| 794 |
+
"learning_rate": 7.758620689655171e-06,
|
| 795 |
+
"loss": 0.0,
|
| 796 |
+
"step": 1130
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 19.66,
|
| 800 |
+
"learning_rate": 5.1724137931034475e-06,
|
| 801 |
+
"loss": 0.0,
|
| 802 |
+
"step": 1140
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 19.83,
|
| 806 |
+
"learning_rate": 2.5862068965517237e-06,
|
| 807 |
+
"loss": 0.0001,
|
| 808 |
+
"step": 1150
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 20.0,
|
| 812 |
+
"learning_rate": 0.0,
|
| 813 |
+
"loss": 0.0001,
|
| 814 |
+
"step": 1160
|
| 815 |
+
},
|
| 816 |
{
|
| 817 |
"epoch": 20.0,
|
| 818 |
"step": 1160,
|
| 819 |
+
"total_flos": 9652606012532736.0,
|
| 820 |
+
"train_loss": 0.03386630766055595,
|
| 821 |
+
"train_runtime": 583.9583,
|
| 822 |
+
"train_samples_per_second": 125.625,
|
| 823 |
+
"train_steps_per_second": 1.986
|
| 824 |
}
|
| 825 |
],
|
| 826 |
"max_steps": 1160,
|
| 827 |
"num_train_epochs": 20,
|
| 828 |
+
"total_flos": 9652606012532736.0,
|
| 829 |
"trial_name": null,
|
| 830 |
"trial_params": null
|
| 831 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3183
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9fbe4afb7644cd203f43cc1a7b0a7c5af3dfc2635d1ea70a03c3deaa68c6a6e2
|
| 3 |
size 3183
|