Commit
·
5a75e6c
1
Parent(s):
309fb2a
End of training
Browse files- all_results.json +15 -0
- eval_results.json +10 -0
- runs/Aug12_10-33-28_3f0a2cf25cdb/events.out.tfevents.1660350308.3f0a2cf25cdb.443.2 +3 -0
- train_results.json +8 -0
- trainer_state.json +1639 -0
all_results.json
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 2.0,
|
3 |
+
"eval_accuracy": 0.5895981374153325,
|
4 |
+
"eval_loss": 1.842421054840088,
|
5 |
+
"eval_runtime": 6167.0931,
|
6 |
+
"eval_samples": 134942,
|
7 |
+
"eval_samples_per_second": 21.881,
|
8 |
+
"eval_steps_per_second": 10.94,
|
9 |
+
"perplexity": 6.311800993960819,
|
10 |
+
"train_loss": 2.1117283579871065,
|
11 |
+
"train_runtime": 41123.1656,
|
12 |
+
"train_samples": 134942,
|
13 |
+
"train_samples_per_second": 6.563,
|
14 |
+
"train_steps_per_second": 3.281
|
15 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 2.0,
|
3 |
+
"eval_accuracy": 0.5895981374153325,
|
4 |
+
"eval_loss": 1.842421054840088,
|
5 |
+
"eval_runtime": 6167.0931,
|
6 |
+
"eval_samples": 134942,
|
7 |
+
"eval_samples_per_second": 21.881,
|
8 |
+
"eval_steps_per_second": 10.94,
|
9 |
+
"perplexity": 6.311800993960819
|
10 |
+
}
|
runs/Aug12_10-33-28_3f0a2cf25cdb/events.out.tfevents.1660350308.3f0a2cf25cdb.443.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:82ebb65765eecb038f464ee8d462b22900261475d3e68cf7144a91c45208739f
|
3 |
+
size 369
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 2.0,
|
3 |
+
"train_loss": 2.1117283579871065,
|
4 |
+
"train_runtime": 41123.1656,
|
5 |
+
"train_samples": 134942,
|
6 |
+
"train_samples_per_second": 6.563,
|
7 |
+
"train_steps_per_second": 3.281
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,1639 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.0,
|
5 |
+
"global_step": 134942,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.01,
|
12 |
+
"learning_rate": 4.981473521957582e-05,
|
13 |
+
"loss": 2.1777,
|
14 |
+
"step": 500
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.01,
|
18 |
+
"learning_rate": 4.962947043915164e-05,
|
19 |
+
"loss": 2.1697,
|
20 |
+
"step": 1000
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.02,
|
24 |
+
"learning_rate": 4.944420565872746e-05,
|
25 |
+
"loss": 2.208,
|
26 |
+
"step": 1500
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.03,
|
30 |
+
"learning_rate": 4.9258940878303276e-05,
|
31 |
+
"loss": 2.206,
|
32 |
+
"step": 2000
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.04,
|
36 |
+
"learning_rate": 4.907367609787909e-05,
|
37 |
+
"loss": 2.2025,
|
38 |
+
"step": 2500
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.04,
|
42 |
+
"learning_rate": 4.8888411317454905e-05,
|
43 |
+
"loss": 2.2037,
|
44 |
+
"step": 3000
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.05,
|
48 |
+
"learning_rate": 4.8703146537030724e-05,
|
49 |
+
"loss": 2.1928,
|
50 |
+
"step": 3500
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.06,
|
54 |
+
"learning_rate": 4.851788175660654e-05,
|
55 |
+
"loss": 2.2034,
|
56 |
+
"step": 4000
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.07,
|
60 |
+
"learning_rate": 4.833261697618236e-05,
|
61 |
+
"loss": 2.1958,
|
62 |
+
"step": 4500
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.07,
|
66 |
+
"learning_rate": 4.814735219575818e-05,
|
67 |
+
"loss": 2.221,
|
68 |
+
"step": 5000
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.08,
|
72 |
+
"learning_rate": 4.7962087415334e-05,
|
73 |
+
"loss": 2.2158,
|
74 |
+
"step": 5500
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.09,
|
78 |
+
"learning_rate": 4.7776822634909815e-05,
|
79 |
+
"loss": 2.206,
|
80 |
+
"step": 6000
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.1,
|
84 |
+
"learning_rate": 4.7591557854485634e-05,
|
85 |
+
"loss": 2.1996,
|
86 |
+
"step": 6500
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.1,
|
90 |
+
"learning_rate": 4.740629307406145e-05,
|
91 |
+
"loss": 2.21,
|
92 |
+
"step": 7000
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.11,
|
96 |
+
"learning_rate": 4.722102829363727e-05,
|
97 |
+
"loss": 2.2156,
|
98 |
+
"step": 7500
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.12,
|
102 |
+
"learning_rate": 4.703576351321309e-05,
|
103 |
+
"loss": 2.2285,
|
104 |
+
"step": 8000
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.13,
|
108 |
+
"learning_rate": 4.685049873278891e-05,
|
109 |
+
"loss": 2.2124,
|
110 |
+
"step": 8500
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.13,
|
114 |
+
"learning_rate": 4.6665233952364725e-05,
|
115 |
+
"loss": 2.2073,
|
116 |
+
"step": 9000
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.14,
|
120 |
+
"learning_rate": 4.6479969171940543e-05,
|
121 |
+
"loss": 2.1891,
|
122 |
+
"step": 9500
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.15,
|
126 |
+
"learning_rate": 4.6294704391516355e-05,
|
127 |
+
"loss": 2.1992,
|
128 |
+
"step": 10000
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.16,
|
132 |
+
"learning_rate": 4.610943961109217e-05,
|
133 |
+
"loss": 2.1831,
|
134 |
+
"step": 10500
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 0.16,
|
138 |
+
"learning_rate": 4.592417483066799e-05,
|
139 |
+
"loss": 2.2132,
|
140 |
+
"step": 11000
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 0.17,
|
144 |
+
"learning_rate": 4.573891005024381e-05,
|
145 |
+
"loss": 2.2117,
|
146 |
+
"step": 11500
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 0.18,
|
150 |
+
"learning_rate": 4.555364526981963e-05,
|
151 |
+
"loss": 2.1857,
|
152 |
+
"step": 12000
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 0.19,
|
156 |
+
"learning_rate": 4.5368380489395447e-05,
|
157 |
+
"loss": 2.2008,
|
158 |
+
"step": 12500
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 0.19,
|
162 |
+
"learning_rate": 4.5183115708971265e-05,
|
163 |
+
"loss": 2.1957,
|
164 |
+
"step": 13000
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 0.2,
|
168 |
+
"learning_rate": 4.499785092854708e-05,
|
169 |
+
"loss": 2.212,
|
170 |
+
"step": 13500
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.21,
|
174 |
+
"learning_rate": 4.48125861481229e-05,
|
175 |
+
"loss": 2.2022,
|
176 |
+
"step": 14000
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 0.21,
|
180 |
+
"learning_rate": 4.462732136769871e-05,
|
181 |
+
"loss": 2.1796,
|
182 |
+
"step": 14500
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 0.22,
|
186 |
+
"learning_rate": 4.444205658727453e-05,
|
187 |
+
"loss": 2.1775,
|
188 |
+
"step": 15000
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 0.23,
|
192 |
+
"learning_rate": 4.425679180685035e-05,
|
193 |
+
"loss": 2.2016,
|
194 |
+
"step": 15500
|
195 |
+
},
|
196 |
+
{
|
197 |
+
"epoch": 0.24,
|
198 |
+
"learning_rate": 4.407152702642617e-05,
|
199 |
+
"loss": 2.1926,
|
200 |
+
"step": 16000
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 0.24,
|
204 |
+
"learning_rate": 4.3886262246001986e-05,
|
205 |
+
"loss": 2.2146,
|
206 |
+
"step": 16500
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 0.25,
|
210 |
+
"learning_rate": 4.3700997465577804e-05,
|
211 |
+
"loss": 2.2015,
|
212 |
+
"step": 17000
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.26,
|
216 |
+
"learning_rate": 4.351573268515363e-05,
|
217 |
+
"loss": 2.1924,
|
218 |
+
"step": 17500
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 0.27,
|
222 |
+
"learning_rate": 4.333046790472944e-05,
|
223 |
+
"loss": 2.2027,
|
224 |
+
"step": 18000
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 0.27,
|
228 |
+
"learning_rate": 4.314520312430526e-05,
|
229 |
+
"loss": 2.1899,
|
230 |
+
"step": 18500
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 0.28,
|
234 |
+
"learning_rate": 4.295993834388108e-05,
|
235 |
+
"loss": 2.1938,
|
236 |
+
"step": 19000
|
237 |
+
},
|
238 |
+
{
|
239 |
+
"epoch": 0.29,
|
240 |
+
"learning_rate": 4.2774673563456896e-05,
|
241 |
+
"loss": 2.1898,
|
242 |
+
"step": 19500
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 0.3,
|
246 |
+
"learning_rate": 4.2589408783032714e-05,
|
247 |
+
"loss": 2.1971,
|
248 |
+
"step": 20000
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 0.3,
|
252 |
+
"learning_rate": 4.240414400260853e-05,
|
253 |
+
"loss": 2.1934,
|
254 |
+
"step": 20500
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.31,
|
258 |
+
"learning_rate": 4.221887922218435e-05,
|
259 |
+
"loss": 2.1759,
|
260 |
+
"step": 21000
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"epoch": 0.32,
|
264 |
+
"learning_rate": 4.203361444176017e-05,
|
265 |
+
"loss": 2.1806,
|
266 |
+
"step": 21500
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 0.33,
|
270 |
+
"learning_rate": 4.184834966133598e-05,
|
271 |
+
"loss": 2.1956,
|
272 |
+
"step": 22000
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 0.33,
|
276 |
+
"learning_rate": 4.16630848809118e-05,
|
277 |
+
"loss": 2.2144,
|
278 |
+
"step": 22500
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 0.34,
|
282 |
+
"learning_rate": 4.147782010048762e-05,
|
283 |
+
"loss": 2.2047,
|
284 |
+
"step": 23000
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 0.35,
|
288 |
+
"learning_rate": 4.1292555320063436e-05,
|
289 |
+
"loss": 2.1988,
|
290 |
+
"step": 23500
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 0.36,
|
294 |
+
"learning_rate": 4.1107290539639254e-05,
|
295 |
+
"loss": 2.209,
|
296 |
+
"step": 24000
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.36,
|
300 |
+
"learning_rate": 4.092202575921507e-05,
|
301 |
+
"loss": 2.1929,
|
302 |
+
"step": 24500
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"epoch": 0.37,
|
306 |
+
"learning_rate": 4.073676097879089e-05,
|
307 |
+
"loss": 2.1703,
|
308 |
+
"step": 25000
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 0.38,
|
312 |
+
"learning_rate": 4.055149619836671e-05,
|
313 |
+
"loss": 2.2071,
|
314 |
+
"step": 25500
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 0.39,
|
318 |
+
"learning_rate": 4.036623141794252e-05,
|
319 |
+
"loss": 2.1894,
|
320 |
+
"step": 26000
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 0.39,
|
324 |
+
"learning_rate": 4.018096663751834e-05,
|
325 |
+
"loss": 2.2136,
|
326 |
+
"step": 26500
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 0.4,
|
330 |
+
"learning_rate": 3.999570185709416e-05,
|
331 |
+
"loss": 2.2125,
|
332 |
+
"step": 27000
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 0.41,
|
336 |
+
"learning_rate": 3.9810437076669975e-05,
|
337 |
+
"loss": 2.1986,
|
338 |
+
"step": 27500
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.41,
|
342 |
+
"learning_rate": 3.9625172296245794e-05,
|
343 |
+
"loss": 2.203,
|
344 |
+
"step": 28000
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 0.42,
|
348 |
+
"learning_rate": 3.943990751582161e-05,
|
349 |
+
"loss": 2.1836,
|
350 |
+
"step": 28500
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 0.43,
|
354 |
+
"learning_rate": 3.925464273539744e-05,
|
355 |
+
"loss": 2.1815,
|
356 |
+
"step": 29000
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 0.44,
|
360 |
+
"learning_rate": 3.906937795497325e-05,
|
361 |
+
"loss": 2.2034,
|
362 |
+
"step": 29500
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 0.44,
|
366 |
+
"learning_rate": 3.888411317454907e-05,
|
367 |
+
"loss": 2.1769,
|
368 |
+
"step": 30000
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 0.45,
|
372 |
+
"learning_rate": 3.8698848394124885e-05,
|
373 |
+
"loss": 2.1994,
|
374 |
+
"step": 30500
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 0.46,
|
378 |
+
"learning_rate": 3.8513583613700704e-05,
|
379 |
+
"loss": 2.1856,
|
380 |
+
"step": 31000
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.47,
|
384 |
+
"learning_rate": 3.832831883327652e-05,
|
385 |
+
"loss": 2.1853,
|
386 |
+
"step": 31500
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 0.47,
|
390 |
+
"learning_rate": 3.814305405285234e-05,
|
391 |
+
"loss": 2.1941,
|
392 |
+
"step": 32000
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 0.48,
|
396 |
+
"learning_rate": 3.795778927242816e-05,
|
397 |
+
"loss": 2.2018,
|
398 |
+
"step": 32500
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 0.49,
|
402 |
+
"learning_rate": 3.777252449200398e-05,
|
403 |
+
"loss": 2.2126,
|
404 |
+
"step": 33000
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 0.5,
|
408 |
+
"learning_rate": 3.7587259711579795e-05,
|
409 |
+
"loss": 2.2037,
|
410 |
+
"step": 33500
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 0.5,
|
414 |
+
"learning_rate": 3.7401994931155607e-05,
|
415 |
+
"loss": 2.1956,
|
416 |
+
"step": 34000
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 0.51,
|
420 |
+
"learning_rate": 3.7216730150731425e-05,
|
421 |
+
"loss": 2.1852,
|
422 |
+
"step": 34500
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.52,
|
426 |
+
"learning_rate": 3.703146537030724e-05,
|
427 |
+
"loss": 2.1795,
|
428 |
+
"step": 35000
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 0.53,
|
432 |
+
"learning_rate": 3.684620058988306e-05,
|
433 |
+
"loss": 2.2005,
|
434 |
+
"step": 35500
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 0.53,
|
438 |
+
"learning_rate": 3.666093580945888e-05,
|
439 |
+
"loss": 2.1901,
|
440 |
+
"step": 36000
|
441 |
+
},
|
442 |
+
{
|
443 |
+
"epoch": 0.54,
|
444 |
+
"learning_rate": 3.64756710290347e-05,
|
445 |
+
"loss": 2.1844,
|
446 |
+
"step": 36500
|
447 |
+
},
|
448 |
+
{
|
449 |
+
"epoch": 0.55,
|
450 |
+
"learning_rate": 3.6290406248610516e-05,
|
451 |
+
"loss": 2.1723,
|
452 |
+
"step": 37000
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 0.56,
|
456 |
+
"learning_rate": 3.6105141468186335e-05,
|
457 |
+
"loss": 2.1801,
|
458 |
+
"step": 37500
|
459 |
+
},
|
460 |
+
{
|
461 |
+
"epoch": 0.56,
|
462 |
+
"learning_rate": 3.5919876687762146e-05,
|
463 |
+
"loss": 2.1857,
|
464 |
+
"step": 38000
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.57,
|
468 |
+
"learning_rate": 3.5734611907337965e-05,
|
469 |
+
"loss": 2.1781,
|
470 |
+
"step": 38500
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 0.58,
|
474 |
+
"learning_rate": 3.554934712691378e-05,
|
475 |
+
"loss": 2.1711,
|
476 |
+
"step": 39000
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 0.59,
|
480 |
+
"learning_rate": 3.53640823464896e-05,
|
481 |
+
"loss": 2.1726,
|
482 |
+
"step": 39500
|
483 |
+
},
|
484 |
+
{
|
485 |
+
"epoch": 0.59,
|
486 |
+
"learning_rate": 3.5178817566065426e-05,
|
487 |
+
"loss": 2.2039,
|
488 |
+
"step": 40000
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 0.6,
|
492 |
+
"learning_rate": 3.4993552785641245e-05,
|
493 |
+
"loss": 2.1792,
|
494 |
+
"step": 40500
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 0.61,
|
498 |
+
"learning_rate": 3.480828800521706e-05,
|
499 |
+
"loss": 2.1813,
|
500 |
+
"step": 41000
|
501 |
+
},
|
502 |
+
{
|
503 |
+
"epoch": 0.62,
|
504 |
+
"learning_rate": 3.4623023224792874e-05,
|
505 |
+
"loss": 2.1731,
|
506 |
+
"step": 41500
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.62,
|
510 |
+
"learning_rate": 3.443775844436869e-05,
|
511 |
+
"loss": 2.1821,
|
512 |
+
"step": 42000
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 0.63,
|
516 |
+
"learning_rate": 3.425249366394451e-05,
|
517 |
+
"loss": 2.1767,
|
518 |
+
"step": 42500
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"epoch": 0.64,
|
522 |
+
"learning_rate": 3.406722888352033e-05,
|
523 |
+
"loss": 2.1693,
|
524 |
+
"step": 43000
|
525 |
+
},
|
526 |
+
{
|
527 |
+
"epoch": 0.64,
|
528 |
+
"learning_rate": 3.388196410309615e-05,
|
529 |
+
"loss": 2.1945,
|
530 |
+
"step": 43500
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"epoch": 0.65,
|
534 |
+
"learning_rate": 3.3696699322671966e-05,
|
535 |
+
"loss": 2.1872,
|
536 |
+
"step": 44000
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 0.66,
|
540 |
+
"learning_rate": 3.3511434542247784e-05,
|
541 |
+
"loss": 2.168,
|
542 |
+
"step": 44500
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 0.67,
|
546 |
+
"learning_rate": 3.33261697618236e-05,
|
547 |
+
"loss": 2.1672,
|
548 |
+
"step": 45000
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.67,
|
552 |
+
"learning_rate": 3.314090498139942e-05,
|
553 |
+
"loss": 2.1779,
|
554 |
+
"step": 45500
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 0.68,
|
558 |
+
"learning_rate": 3.295564020097523e-05,
|
559 |
+
"loss": 2.1873,
|
560 |
+
"step": 46000
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 0.69,
|
564 |
+
"learning_rate": 3.277037542055105e-05,
|
565 |
+
"loss": 2.1972,
|
566 |
+
"step": 46500
|
567 |
+
},
|
568 |
+
{
|
569 |
+
"epoch": 0.7,
|
570 |
+
"learning_rate": 3.258511064012687e-05,
|
571 |
+
"loss": 2.1923,
|
572 |
+
"step": 47000
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"epoch": 0.7,
|
576 |
+
"learning_rate": 3.239984585970269e-05,
|
577 |
+
"loss": 2.1494,
|
578 |
+
"step": 47500
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 0.71,
|
582 |
+
"learning_rate": 3.2214581079278506e-05,
|
583 |
+
"loss": 2.1825,
|
584 |
+
"step": 48000
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 0.72,
|
588 |
+
"learning_rate": 3.2029316298854324e-05,
|
589 |
+
"loss": 2.1538,
|
590 |
+
"step": 48500
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.73,
|
594 |
+
"learning_rate": 3.184405151843014e-05,
|
595 |
+
"loss": 2.1737,
|
596 |
+
"step": 49000
|
597 |
+
},
|
598 |
+
{
|
599 |
+
"epoch": 0.73,
|
600 |
+
"learning_rate": 3.165878673800596e-05,
|
601 |
+
"loss": 2.194,
|
602 |
+
"step": 49500
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 0.74,
|
606 |
+
"learning_rate": 3.147352195758177e-05,
|
607 |
+
"loss": 2.1813,
|
608 |
+
"step": 50000
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 0.75,
|
612 |
+
"learning_rate": 3.128825717715759e-05,
|
613 |
+
"loss": 2.1702,
|
614 |
+
"step": 50500
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 0.76,
|
618 |
+
"learning_rate": 3.110299239673341e-05,
|
619 |
+
"loss": 2.1708,
|
620 |
+
"step": 51000
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 0.76,
|
624 |
+
"learning_rate": 3.0917727616309234e-05,
|
625 |
+
"loss": 2.1997,
|
626 |
+
"step": 51500
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 0.77,
|
630 |
+
"learning_rate": 3.073246283588505e-05,
|
631 |
+
"loss": 2.1953,
|
632 |
+
"step": 52000
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.78,
|
636 |
+
"learning_rate": 3.054719805546087e-05,
|
637 |
+
"loss": 2.1743,
|
638 |
+
"step": 52500
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 0.79,
|
642 |
+
"learning_rate": 3.0361933275036685e-05,
|
643 |
+
"loss": 2.1933,
|
644 |
+
"step": 53000
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 0.79,
|
648 |
+
"learning_rate": 3.0176668494612504e-05,
|
649 |
+
"loss": 2.1624,
|
650 |
+
"step": 53500
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 0.8,
|
654 |
+
"learning_rate": 2.9991403714188322e-05,
|
655 |
+
"loss": 2.1986,
|
656 |
+
"step": 54000
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 0.81,
|
660 |
+
"learning_rate": 2.9806138933764137e-05,
|
661 |
+
"loss": 2.159,
|
662 |
+
"step": 54500
|
663 |
+
},
|
664 |
+
{
|
665 |
+
"epoch": 0.82,
|
666 |
+
"learning_rate": 2.9620874153339955e-05,
|
667 |
+
"loss": 2.187,
|
668 |
+
"step": 55000
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 0.82,
|
672 |
+
"learning_rate": 2.9435609372915773e-05,
|
673 |
+
"loss": 2.1734,
|
674 |
+
"step": 55500
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 0.83,
|
678 |
+
"learning_rate": 2.9250344592491592e-05,
|
679 |
+
"loss": 2.1653,
|
680 |
+
"step": 56000
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 0.84,
|
684 |
+
"learning_rate": 2.9065079812067407e-05,
|
685 |
+
"loss": 2.176,
|
686 |
+
"step": 56500
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 0.84,
|
690 |
+
"learning_rate": 2.8879815031643225e-05,
|
691 |
+
"loss": 2.1778,
|
692 |
+
"step": 57000
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 0.85,
|
696 |
+
"learning_rate": 2.8694550251219043e-05,
|
697 |
+
"loss": 2.1855,
|
698 |
+
"step": 57500
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 0.86,
|
702 |
+
"learning_rate": 2.850928547079486e-05,
|
703 |
+
"loss": 2.1771,
|
704 |
+
"step": 58000
|
705 |
+
},
|
706 |
+
{
|
707 |
+
"epoch": 0.87,
|
708 |
+
"learning_rate": 2.832402069037068e-05,
|
709 |
+
"loss": 2.1719,
|
710 |
+
"step": 58500
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 0.87,
|
714 |
+
"learning_rate": 2.8138755909946495e-05,
|
715 |
+
"loss": 2.1632,
|
716 |
+
"step": 59000
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 0.88,
|
720 |
+
"learning_rate": 2.7953491129522313e-05,
|
721 |
+
"loss": 2.1678,
|
722 |
+
"step": 59500
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 0.89,
|
726 |
+
"learning_rate": 2.776822634909813e-05,
|
727 |
+
"loss": 2.1671,
|
728 |
+
"step": 60000
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 0.9,
|
732 |
+
"learning_rate": 2.758296156867395e-05,
|
733 |
+
"loss": 2.1724,
|
734 |
+
"step": 60500
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 0.9,
|
738 |
+
"learning_rate": 2.7397696788249765e-05,
|
739 |
+
"loss": 2.1716,
|
740 |
+
"step": 61000
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 0.91,
|
744 |
+
"learning_rate": 2.7212432007825583e-05,
|
745 |
+
"loss": 2.1606,
|
746 |
+
"step": 61500
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 0.92,
|
750 |
+
"learning_rate": 2.70271672274014e-05,
|
751 |
+
"loss": 2.1842,
|
752 |
+
"step": 62000
|
753 |
+
},
|
754 |
+
{
|
755 |
+
"epoch": 0.93,
|
756 |
+
"learning_rate": 2.684190244697722e-05,
|
757 |
+
"loss": 2.1652,
|
758 |
+
"step": 62500
|
759 |
+
},
|
760 |
+
{
|
761 |
+
"epoch": 0.93,
|
762 |
+
"learning_rate": 2.665663766655304e-05,
|
763 |
+
"loss": 2.1792,
|
764 |
+
"step": 63000
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"epoch": 0.94,
|
768 |
+
"learning_rate": 2.647137288612886e-05,
|
769 |
+
"loss": 2.1641,
|
770 |
+
"step": 63500
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"epoch": 0.95,
|
774 |
+
"learning_rate": 2.6286108105704678e-05,
|
775 |
+
"loss": 2.1609,
|
776 |
+
"step": 64000
|
777 |
+
},
|
778 |
+
{
|
779 |
+
"epoch": 0.96,
|
780 |
+
"learning_rate": 2.6100843325280493e-05,
|
781 |
+
"loss": 2.1632,
|
782 |
+
"step": 64500
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 0.96,
|
786 |
+
"learning_rate": 2.591557854485631e-05,
|
787 |
+
"loss": 2.1831,
|
788 |
+
"step": 65000
|
789 |
+
},
|
790 |
+
{
|
791 |
+
"epoch": 0.97,
|
792 |
+
"learning_rate": 2.573031376443213e-05,
|
793 |
+
"loss": 2.1679,
|
794 |
+
"step": 65500
|
795 |
+
},
|
796 |
+
{
|
797 |
+
"epoch": 0.98,
|
798 |
+
"learning_rate": 2.5545048984007948e-05,
|
799 |
+
"loss": 2.1602,
|
800 |
+
"step": 66000
|
801 |
+
},
|
802 |
+
{
|
803 |
+
"epoch": 0.99,
|
804 |
+
"learning_rate": 2.5359784203583763e-05,
|
805 |
+
"loss": 2.1663,
|
806 |
+
"step": 66500
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 0.99,
|
810 |
+
"learning_rate": 2.517451942315958e-05,
|
811 |
+
"loss": 2.1854,
|
812 |
+
"step": 67000
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 1.0,
|
816 |
+
"learning_rate": 2.49892546427354e-05,
|
817 |
+
"loss": 2.165,
|
818 |
+
"step": 67500
|
819 |
+
},
|
820 |
+
{
|
821 |
+
"epoch": 1.01,
|
822 |
+
"learning_rate": 2.4803989862311218e-05,
|
823 |
+
"loss": 2.007,
|
824 |
+
"step": 68000
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 1.02,
|
828 |
+
"learning_rate": 2.4618725081887032e-05,
|
829 |
+
"loss": 2.021,
|
830 |
+
"step": 68500
|
831 |
+
},
|
832 |
+
{
|
833 |
+
"epoch": 1.02,
|
834 |
+
"learning_rate": 2.443346030146285e-05,
|
835 |
+
"loss": 2.0256,
|
836 |
+
"step": 69000
|
837 |
+
},
|
838 |
+
{
|
839 |
+
"epoch": 1.03,
|
840 |
+
"learning_rate": 2.424819552103867e-05,
|
841 |
+
"loss": 2.0075,
|
842 |
+
"step": 69500
|
843 |
+
},
|
844 |
+
{
|
845 |
+
"epoch": 1.04,
|
846 |
+
"learning_rate": 2.4062930740614487e-05,
|
847 |
+
"loss": 2.0186,
|
848 |
+
"step": 70000
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 1.04,
|
852 |
+
"learning_rate": 2.3877665960190306e-05,
|
853 |
+
"loss": 2.0247,
|
854 |
+
"step": 70500
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 1.05,
|
858 |
+
"learning_rate": 2.369240117976612e-05,
|
859 |
+
"loss": 2.0036,
|
860 |
+
"step": 71000
|
861 |
+
},
|
862 |
+
{
|
863 |
+
"epoch": 1.06,
|
864 |
+
"learning_rate": 2.3507136399341942e-05,
|
865 |
+
"loss": 2.0254,
|
866 |
+
"step": 71500
|
867 |
+
},
|
868 |
+
{
|
869 |
+
"epoch": 1.07,
|
870 |
+
"learning_rate": 2.332187161891776e-05,
|
871 |
+
"loss": 2.0216,
|
872 |
+
"step": 72000
|
873 |
+
},
|
874 |
+
{
|
875 |
+
"epoch": 1.07,
|
876 |
+
"learning_rate": 2.3136606838493576e-05,
|
877 |
+
"loss": 2.0236,
|
878 |
+
"step": 72500
|
879 |
+
},
|
880 |
+
{
|
881 |
+
"epoch": 1.08,
|
882 |
+
"learning_rate": 2.2951342058069394e-05,
|
883 |
+
"loss": 2.0131,
|
884 |
+
"step": 73000
|
885 |
+
},
|
886 |
+
{
|
887 |
+
"epoch": 1.09,
|
888 |
+
"learning_rate": 2.2766077277645212e-05,
|
889 |
+
"loss": 2.0136,
|
890 |
+
"step": 73500
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 1.1,
|
894 |
+
"learning_rate": 2.258081249722103e-05,
|
895 |
+
"loss": 2.0133,
|
896 |
+
"step": 74000
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 1.1,
|
900 |
+
"learning_rate": 2.2395547716796845e-05,
|
901 |
+
"loss": 2.017,
|
902 |
+
"step": 74500
|
903 |
+
},
|
904 |
+
{
|
905 |
+
"epoch": 1.11,
|
906 |
+
"learning_rate": 2.2210282936372664e-05,
|
907 |
+
"loss": 2.0366,
|
908 |
+
"step": 75000
|
909 |
+
},
|
910 |
+
{
|
911 |
+
"epoch": 1.12,
|
912 |
+
"learning_rate": 2.2025018155948482e-05,
|
913 |
+
"loss": 2.0203,
|
914 |
+
"step": 75500
|
915 |
+
},
|
916 |
+
{
|
917 |
+
"epoch": 1.13,
|
918 |
+
"learning_rate": 2.18397533755243e-05,
|
919 |
+
"loss": 2.0226,
|
920 |
+
"step": 76000
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"epoch": 1.13,
|
924 |
+
"learning_rate": 2.165448859510012e-05,
|
925 |
+
"loss": 2.0357,
|
926 |
+
"step": 76500
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 1.14,
|
930 |
+
"learning_rate": 2.1469223814675937e-05,
|
931 |
+
"loss": 2.0249,
|
932 |
+
"step": 77000
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 1.15,
|
936 |
+
"learning_rate": 2.1283959034251755e-05,
|
937 |
+
"loss": 2.0487,
|
938 |
+
"step": 77500
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 1.16,
|
942 |
+
"learning_rate": 2.1098694253827574e-05,
|
943 |
+
"loss": 2.0277,
|
944 |
+
"step": 78000
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 1.16,
|
948 |
+
"learning_rate": 2.091342947340339e-05,
|
949 |
+
"loss": 2.0468,
|
950 |
+
"step": 78500
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 1.17,
|
954 |
+
"learning_rate": 2.0728164692979207e-05,
|
955 |
+
"loss": 2.035,
|
956 |
+
"step": 79000
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 1.18,
|
960 |
+
"learning_rate": 2.0542899912555025e-05,
|
961 |
+
"loss": 2.0286,
|
962 |
+
"step": 79500
|
963 |
+
},
|
964 |
+
{
|
965 |
+
"epoch": 1.19,
|
966 |
+
"learning_rate": 2.0357635132130843e-05,
|
967 |
+
"loss": 2.0409,
|
968 |
+
"step": 80000
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 1.19,
|
972 |
+
"learning_rate": 2.0172370351706658e-05,
|
973 |
+
"loss": 2.024,
|
974 |
+
"step": 80500
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 1.2,
|
978 |
+
"learning_rate": 1.9987105571282477e-05,
|
979 |
+
"loss": 2.0153,
|
980 |
+
"step": 81000
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 1.21,
|
984 |
+
"learning_rate": 1.9801840790858295e-05,
|
985 |
+
"loss": 2.0234,
|
986 |
+
"step": 81500
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 1.22,
|
990 |
+
"learning_rate": 1.9616576010434113e-05,
|
991 |
+
"loss": 2.0282,
|
992 |
+
"step": 82000
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 1.22,
|
996 |
+
"learning_rate": 1.9431311230009928e-05,
|
997 |
+
"loss": 2.025,
|
998 |
+
"step": 82500
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 1.23,
|
1002 |
+
"learning_rate": 1.924604644958575e-05,
|
1003 |
+
"loss": 2.0451,
|
1004 |
+
"step": 83000
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 1.24,
|
1008 |
+
"learning_rate": 1.9060781669161568e-05,
|
1009 |
+
"loss": 2.0347,
|
1010 |
+
"step": 83500
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 1.24,
|
1014 |
+
"learning_rate": 1.8875516888737386e-05,
|
1015 |
+
"loss": 2.0412,
|
1016 |
+
"step": 84000
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 1.25,
|
1020 |
+
"learning_rate": 1.86902521083132e-05,
|
1021 |
+
"loss": 2.0236,
|
1022 |
+
"step": 84500
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 1.26,
|
1026 |
+
"learning_rate": 1.850498732788902e-05,
|
1027 |
+
"loss": 2.0277,
|
1028 |
+
"step": 85000
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 1.27,
|
1032 |
+
"learning_rate": 1.8319722547464838e-05,
|
1033 |
+
"loss": 2.0431,
|
1034 |
+
"step": 85500
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 1.27,
|
1038 |
+
"learning_rate": 1.8134457767040656e-05,
|
1039 |
+
"loss": 2.0354,
|
1040 |
+
"step": 86000
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 1.28,
|
1044 |
+
"learning_rate": 1.794919298661647e-05,
|
1045 |
+
"loss": 2.0388,
|
1046 |
+
"step": 86500
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 1.29,
|
1050 |
+
"learning_rate": 1.776392820619229e-05,
|
1051 |
+
"loss": 2.0355,
|
1052 |
+
"step": 87000
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 1.3,
|
1056 |
+
"learning_rate": 1.7578663425768108e-05,
|
1057 |
+
"loss": 2.0239,
|
1058 |
+
"step": 87500
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 1.3,
|
1062 |
+
"learning_rate": 1.7393398645343926e-05,
|
1063 |
+
"loss": 2.0209,
|
1064 |
+
"step": 88000
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 1.31,
|
1068 |
+
"learning_rate": 1.7208133864919744e-05,
|
1069 |
+
"loss": 2.0399,
|
1070 |
+
"step": 88500
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 1.32,
|
1074 |
+
"learning_rate": 1.7022869084495563e-05,
|
1075 |
+
"loss": 2.0427,
|
1076 |
+
"step": 89000
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 1.33,
|
1080 |
+
"learning_rate": 1.683760430407138e-05,
|
1081 |
+
"loss": 2.0448,
|
1082 |
+
"step": 89500
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 1.33,
|
1086 |
+
"learning_rate": 1.66523395236472e-05,
|
1087 |
+
"loss": 2.0358,
|
1088 |
+
"step": 90000
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 1.34,
|
1092 |
+
"learning_rate": 1.6467074743223014e-05,
|
1093 |
+
"loss": 2.0449,
|
1094 |
+
"step": 90500
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 1.35,
|
1098 |
+
"learning_rate": 1.6281809962798833e-05,
|
1099 |
+
"loss": 2.0267,
|
1100 |
+
"step": 91000
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 1.36,
|
1104 |
+
"learning_rate": 1.609654518237465e-05,
|
1105 |
+
"loss": 2.0233,
|
1106 |
+
"step": 91500
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 1.36,
|
1110 |
+
"learning_rate": 1.591128040195047e-05,
|
1111 |
+
"loss": 2.0241,
|
1112 |
+
"step": 92000
|
1113 |
+
},
|
1114 |
+
{
|
1115 |
+
"epoch": 1.37,
|
1116 |
+
"learning_rate": 1.5726015621526284e-05,
|
1117 |
+
"loss": 2.0344,
|
1118 |
+
"step": 92500
|
1119 |
+
},
|
1120 |
+
{
|
1121 |
+
"epoch": 1.38,
|
1122 |
+
"learning_rate": 1.5540750841102102e-05,
|
1123 |
+
"loss": 2.0414,
|
1124 |
+
"step": 93000
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 1.39,
|
1128 |
+
"learning_rate": 1.535548606067792e-05,
|
1129 |
+
"loss": 2.0423,
|
1130 |
+
"step": 93500
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 1.39,
|
1134 |
+
"learning_rate": 1.5170221280253737e-05,
|
1135 |
+
"loss": 2.0231,
|
1136 |
+
"step": 94000
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 1.4,
|
1140 |
+
"learning_rate": 1.4984956499829559e-05,
|
1141 |
+
"loss": 2.0444,
|
1142 |
+
"step": 94500
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 1.41,
|
1146 |
+
"learning_rate": 1.4799691719405376e-05,
|
1147 |
+
"loss": 2.0327,
|
1148 |
+
"step": 95000
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 1.42,
|
1152 |
+
"learning_rate": 1.4614426938981194e-05,
|
1153 |
+
"loss": 2.0347,
|
1154 |
+
"step": 95500
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 1.42,
|
1158 |
+
"learning_rate": 1.442916215855701e-05,
|
1159 |
+
"loss": 2.0287,
|
1160 |
+
"step": 96000
|
1161 |
+
},
|
1162 |
+
{
|
1163 |
+
"epoch": 1.43,
|
1164 |
+
"learning_rate": 1.4243897378132829e-05,
|
1165 |
+
"loss": 2.0359,
|
1166 |
+
"step": 96500
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 1.44,
|
1170 |
+
"learning_rate": 1.4058632597708645e-05,
|
1171 |
+
"loss": 2.0329,
|
1172 |
+
"step": 97000
|
1173 |
+
},
|
1174 |
+
{
|
1175 |
+
"epoch": 1.45,
|
1176 |
+
"learning_rate": 1.3873367817284464e-05,
|
1177 |
+
"loss": 2.0221,
|
1178 |
+
"step": 97500
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 1.45,
|
1182 |
+
"learning_rate": 1.368810303686028e-05,
|
1183 |
+
"loss": 2.0234,
|
1184 |
+
"step": 98000
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 1.46,
|
1188 |
+
"learning_rate": 1.3502838256436099e-05,
|
1189 |
+
"loss": 2.0478,
|
1190 |
+
"step": 98500
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 1.47,
|
1194 |
+
"learning_rate": 1.3317573476011915e-05,
|
1195 |
+
"loss": 2.0401,
|
1196 |
+
"step": 99000
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 1.47,
|
1200 |
+
"learning_rate": 1.3132308695587734e-05,
|
1201 |
+
"loss": 2.0402,
|
1202 |
+
"step": 99500
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 1.48,
|
1206 |
+
"learning_rate": 1.2947043915163554e-05,
|
1207 |
+
"loss": 2.0335,
|
1208 |
+
"step": 100000
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 1.49,
|
1212 |
+
"learning_rate": 1.2761779134739372e-05,
|
1213 |
+
"loss": 2.0453,
|
1214 |
+
"step": 100500
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 1.5,
|
1218 |
+
"learning_rate": 1.2576514354315189e-05,
|
1219 |
+
"loss": 2.0324,
|
1220 |
+
"step": 101000
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 1.5,
|
1224 |
+
"learning_rate": 1.2391249573891007e-05,
|
1225 |
+
"loss": 2.0247,
|
1226 |
+
"step": 101500
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"epoch": 1.51,
|
1230 |
+
"learning_rate": 1.2205984793466823e-05,
|
1231 |
+
"loss": 2.0301,
|
1232 |
+
"step": 102000
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 1.52,
|
1236 |
+
"learning_rate": 1.2020720013042642e-05,
|
1237 |
+
"loss": 2.0293,
|
1238 |
+
"step": 102500
|
1239 |
+
},
|
1240 |
+
{
|
1241 |
+
"epoch": 1.53,
|
1242 |
+
"learning_rate": 1.1835455232618458e-05,
|
1243 |
+
"loss": 2.0333,
|
1244 |
+
"step": 103000
|
1245 |
+
},
|
1246 |
+
{
|
1247 |
+
"epoch": 1.53,
|
1248 |
+
"learning_rate": 1.1650190452194277e-05,
|
1249 |
+
"loss": 2.0599,
|
1250 |
+
"step": 103500
|
1251 |
+
},
|
1252 |
+
{
|
1253 |
+
"epoch": 1.54,
|
1254 |
+
"learning_rate": 1.1464925671770093e-05,
|
1255 |
+
"loss": 2.0278,
|
1256 |
+
"step": 104000
|
1257 |
+
},
|
1258 |
+
{
|
1259 |
+
"epoch": 1.55,
|
1260 |
+
"learning_rate": 1.1279660891345913e-05,
|
1261 |
+
"loss": 2.039,
|
1262 |
+
"step": 104500
|
1263 |
+
},
|
1264 |
+
{
|
1265 |
+
"epoch": 1.56,
|
1266 |
+
"learning_rate": 1.109439611092173e-05,
|
1267 |
+
"loss": 2.042,
|
1268 |
+
"step": 105000
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 1.56,
|
1272 |
+
"learning_rate": 1.0909131330497548e-05,
|
1273 |
+
"loss": 2.0592,
|
1274 |
+
"step": 105500
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 1.57,
|
1278 |
+
"learning_rate": 1.0723866550073365e-05,
|
1279 |
+
"loss": 2.0512,
|
1280 |
+
"step": 106000
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 1.58,
|
1284 |
+
"learning_rate": 1.0538601769649183e-05,
|
1285 |
+
"loss": 2.0496,
|
1286 |
+
"step": 106500
|
1287 |
+
},
|
1288 |
+
{
|
1289 |
+
"epoch": 1.59,
|
1290 |
+
"learning_rate": 1.0353336989225001e-05,
|
1291 |
+
"loss": 2.0359,
|
1292 |
+
"step": 107000
|
1293 |
+
},
|
1294 |
+
{
|
1295 |
+
"epoch": 1.59,
|
1296 |
+
"learning_rate": 1.016807220880082e-05,
|
1297 |
+
"loss": 2.0427,
|
1298 |
+
"step": 107500
|
1299 |
+
},
|
1300 |
+
{
|
1301 |
+
"epoch": 1.6,
|
1302 |
+
"learning_rate": 9.982807428376636e-06,
|
1303 |
+
"loss": 2.0655,
|
1304 |
+
"step": 108000
|
1305 |
+
},
|
1306 |
+
{
|
1307 |
+
"epoch": 1.61,
|
1308 |
+
"learning_rate": 9.797542647952455e-06,
|
1309 |
+
"loss": 2.0554,
|
1310 |
+
"step": 108500
|
1311 |
+
},
|
1312 |
+
{
|
1313 |
+
"epoch": 1.62,
|
1314 |
+
"learning_rate": 9.612277867528271e-06,
|
1315 |
+
"loss": 2.0359,
|
1316 |
+
"step": 109000
|
1317 |
+
},
|
1318 |
+
{
|
1319 |
+
"epoch": 1.62,
|
1320 |
+
"learning_rate": 9.42701308710409e-06,
|
1321 |
+
"loss": 2.056,
|
1322 |
+
"step": 109500
|
1323 |
+
},
|
1324 |
+
{
|
1325 |
+
"epoch": 1.63,
|
1326 |
+
"learning_rate": 9.241748306679908e-06,
|
1327 |
+
"loss": 2.0273,
|
1328 |
+
"step": 110000
|
1329 |
+
},
|
1330 |
+
{
|
1331 |
+
"epoch": 1.64,
|
1332 |
+
"learning_rate": 9.056483526255726e-06,
|
1333 |
+
"loss": 2.0323,
|
1334 |
+
"step": 110500
|
1335 |
+
},
|
1336 |
+
{
|
1337 |
+
"epoch": 1.65,
|
1338 |
+
"learning_rate": 8.871218745831543e-06,
|
1339 |
+
"loss": 2.0288,
|
1340 |
+
"step": 111000
|
1341 |
+
},
|
1342 |
+
{
|
1343 |
+
"epoch": 1.65,
|
1344 |
+
"learning_rate": 8.685953965407361e-06,
|
1345 |
+
"loss": 2.0188,
|
1346 |
+
"step": 111500
|
1347 |
+
},
|
1348 |
+
{
|
1349 |
+
"epoch": 1.66,
|
1350 |
+
"learning_rate": 8.500689184983178e-06,
|
1351 |
+
"loss": 2.0162,
|
1352 |
+
"step": 112000
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 1.67,
|
1356 |
+
"learning_rate": 8.315424404558996e-06,
|
1357 |
+
"loss": 2.0407,
|
1358 |
+
"step": 112500
|
1359 |
+
},
|
1360 |
+
{
|
1361 |
+
"epoch": 1.67,
|
1362 |
+
"learning_rate": 8.130159624134814e-06,
|
1363 |
+
"loss": 2.059,
|
1364 |
+
"step": 113000
|
1365 |
+
},
|
1366 |
+
{
|
1367 |
+
"epoch": 1.68,
|
1368 |
+
"learning_rate": 7.944894843710633e-06,
|
1369 |
+
"loss": 2.0614,
|
1370 |
+
"step": 113500
|
1371 |
+
},
|
1372 |
+
{
|
1373 |
+
"epoch": 1.69,
|
1374 |
+
"learning_rate": 7.75963006328645e-06,
|
1375 |
+
"loss": 2.0158,
|
1376 |
+
"step": 114000
|
1377 |
+
},
|
1378 |
+
{
|
1379 |
+
"epoch": 1.7,
|
1380 |
+
"learning_rate": 7.574365282862267e-06,
|
1381 |
+
"loss": 2.0197,
|
1382 |
+
"step": 114500
|
1383 |
+
},
|
1384 |
+
{
|
1385 |
+
"epoch": 1.7,
|
1386 |
+
"learning_rate": 7.389100502438084e-06,
|
1387 |
+
"loss": 2.0359,
|
1388 |
+
"step": 115000
|
1389 |
+
},
|
1390 |
+
{
|
1391 |
+
"epoch": 1.71,
|
1392 |
+
"learning_rate": 7.203835722013903e-06,
|
1393 |
+
"loss": 2.0233,
|
1394 |
+
"step": 115500
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 1.72,
|
1398 |
+
"learning_rate": 7.018570941589721e-06,
|
1399 |
+
"loss": 2.0306,
|
1400 |
+
"step": 116000
|
1401 |
+
},
|
1402 |
+
{
|
1403 |
+
"epoch": 1.73,
|
1404 |
+
"learning_rate": 6.833306161165538e-06,
|
1405 |
+
"loss": 2.0297,
|
1406 |
+
"step": 116500
|
1407 |
+
},
|
1408 |
+
{
|
1409 |
+
"epoch": 1.73,
|
1410 |
+
"learning_rate": 6.648041380741356e-06,
|
1411 |
+
"loss": 2.0465,
|
1412 |
+
"step": 117000
|
1413 |
+
},
|
1414 |
+
{
|
1415 |
+
"epoch": 1.74,
|
1416 |
+
"learning_rate": 6.462776600317173e-06,
|
1417 |
+
"loss": 2.0386,
|
1418 |
+
"step": 117500
|
1419 |
+
},
|
1420 |
+
{
|
1421 |
+
"epoch": 1.75,
|
1422 |
+
"learning_rate": 6.277511819892991e-06,
|
1423 |
+
"loss": 2.0534,
|
1424 |
+
"step": 118000
|
1425 |
+
},
|
1426 |
+
{
|
1427 |
+
"epoch": 1.76,
|
1428 |
+
"learning_rate": 6.092247039468809e-06,
|
1429 |
+
"loss": 2.0351,
|
1430 |
+
"step": 118500
|
1431 |
+
},
|
1432 |
+
{
|
1433 |
+
"epoch": 1.76,
|
1434 |
+
"learning_rate": 5.906982259044627e-06,
|
1435 |
+
"loss": 2.0361,
|
1436 |
+
"step": 119000
|
1437 |
+
},
|
1438 |
+
{
|
1439 |
+
"epoch": 1.77,
|
1440 |
+
"learning_rate": 5.721717478620445e-06,
|
1441 |
+
"loss": 2.0462,
|
1442 |
+
"step": 119500
|
1443 |
+
},
|
1444 |
+
{
|
1445 |
+
"epoch": 1.78,
|
1446 |
+
"learning_rate": 5.536452698196262e-06,
|
1447 |
+
"loss": 2.0551,
|
1448 |
+
"step": 120000
|
1449 |
+
},
|
1450 |
+
{
|
1451 |
+
"epoch": 1.79,
|
1452 |
+
"learning_rate": 5.3511879177720805e-06,
|
1453 |
+
"loss": 2.0461,
|
1454 |
+
"step": 120500
|
1455 |
+
},
|
1456 |
+
{
|
1457 |
+
"epoch": 1.79,
|
1458 |
+
"learning_rate": 5.165923137347898e-06,
|
1459 |
+
"loss": 2.0384,
|
1460 |
+
"step": 121000
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 1.8,
|
1464 |
+
"learning_rate": 4.980658356923715e-06,
|
1465 |
+
"loss": 2.0189,
|
1466 |
+
"step": 121500
|
1467 |
+
},
|
1468 |
+
{
|
1469 |
+
"epoch": 1.81,
|
1470 |
+
"learning_rate": 4.795393576499534e-06,
|
1471 |
+
"loss": 2.0275,
|
1472 |
+
"step": 122000
|
1473 |
+
},
|
1474 |
+
{
|
1475 |
+
"epoch": 1.82,
|
1476 |
+
"learning_rate": 4.610128796075351e-06,
|
1477 |
+
"loss": 2.0477,
|
1478 |
+
"step": 122500
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 1.82,
|
1482 |
+
"learning_rate": 4.424864015651169e-06,
|
1483 |
+
"loss": 2.0411,
|
1484 |
+
"step": 123000
|
1485 |
+
},
|
1486 |
+
{
|
1487 |
+
"epoch": 1.83,
|
1488 |
+
"learning_rate": 4.239599235226987e-06,
|
1489 |
+
"loss": 2.0358,
|
1490 |
+
"step": 123500
|
1491 |
+
},
|
1492 |
+
{
|
1493 |
+
"epoch": 1.84,
|
1494 |
+
"learning_rate": 4.054334454802804e-06,
|
1495 |
+
"loss": 2.0485,
|
1496 |
+
"step": 124000
|
1497 |
+
},
|
1498 |
+
{
|
1499 |
+
"epoch": 1.85,
|
1500 |
+
"learning_rate": 3.869069674378622e-06,
|
1501 |
+
"loss": 2.0406,
|
1502 |
+
"step": 124500
|
1503 |
+
},
|
1504 |
+
{
|
1505 |
+
"epoch": 1.85,
|
1506 |
+
"learning_rate": 3.68380489395444e-06,
|
1507 |
+
"loss": 2.038,
|
1508 |
+
"step": 125000
|
1509 |
+
},
|
1510 |
+
{
|
1511 |
+
"epoch": 1.86,
|
1512 |
+
"learning_rate": 3.4985401135302576e-06,
|
1513 |
+
"loss": 2.0437,
|
1514 |
+
"step": 125500
|
1515 |
+
},
|
1516 |
+
{
|
1517 |
+
"epoch": 1.87,
|
1518 |
+
"learning_rate": 3.313275333106075e-06,
|
1519 |
+
"loss": 2.026,
|
1520 |
+
"step": 126000
|
1521 |
+
},
|
1522 |
+
{
|
1523 |
+
"epoch": 1.87,
|
1524 |
+
"learning_rate": 3.1280105526818934e-06,
|
1525 |
+
"loss": 2.054,
|
1526 |
+
"step": 126500
|
1527 |
+
},
|
1528 |
+
{
|
1529 |
+
"epoch": 1.88,
|
1530 |
+
"learning_rate": 2.942745772257711e-06,
|
1531 |
+
"loss": 2.0416,
|
1532 |
+
"step": 127000
|
1533 |
+
},
|
1534 |
+
{
|
1535 |
+
"epoch": 1.89,
|
1536 |
+
"learning_rate": 2.7574809918335287e-06,
|
1537 |
+
"loss": 2.0198,
|
1538 |
+
"step": 127500
|
1539 |
+
},
|
1540 |
+
{
|
1541 |
+
"epoch": 1.9,
|
1542 |
+
"learning_rate": 2.5722162114093466e-06,
|
1543 |
+
"loss": 2.0431,
|
1544 |
+
"step": 128000
|
1545 |
+
},
|
1546 |
+
{
|
1547 |
+
"epoch": 1.9,
|
1548 |
+
"learning_rate": 2.386951430985164e-06,
|
1549 |
+
"loss": 2.0247,
|
1550 |
+
"step": 128500
|
1551 |
+
},
|
1552 |
+
{
|
1553 |
+
"epoch": 1.91,
|
1554 |
+
"learning_rate": 2.201686650560982e-06,
|
1555 |
+
"loss": 2.0406,
|
1556 |
+
"step": 129000
|
1557 |
+
},
|
1558 |
+
{
|
1559 |
+
"epoch": 1.92,
|
1560 |
+
"learning_rate": 2.0164218701368e-06,
|
1561 |
+
"loss": 2.0438,
|
1562 |
+
"step": 129500
|
1563 |
+
},
|
1564 |
+
{
|
1565 |
+
"epoch": 1.93,
|
1566 |
+
"learning_rate": 1.8311570897126173e-06,
|
1567 |
+
"loss": 2.073,
|
1568 |
+
"step": 130000
|
1569 |
+
},
|
1570 |
+
{
|
1571 |
+
"epoch": 1.93,
|
1572 |
+
"learning_rate": 1.645892309288435e-06,
|
1573 |
+
"loss": 2.0388,
|
1574 |
+
"step": 130500
|
1575 |
+
},
|
1576 |
+
{
|
1577 |
+
"epoch": 1.94,
|
1578 |
+
"learning_rate": 1.4606275288642529e-06,
|
1579 |
+
"loss": 2.0336,
|
1580 |
+
"step": 131000
|
1581 |
+
},
|
1582 |
+
{
|
1583 |
+
"epoch": 1.95,
|
1584 |
+
"learning_rate": 1.2753627484400705e-06,
|
1585 |
+
"loss": 2.0644,
|
1586 |
+
"step": 131500
|
1587 |
+
},
|
1588 |
+
{
|
1589 |
+
"epoch": 1.96,
|
1590 |
+
"learning_rate": 1.0900979680158882e-06,
|
1591 |
+
"loss": 2.0585,
|
1592 |
+
"step": 132000
|
1593 |
+
},
|
1594 |
+
{
|
1595 |
+
"epoch": 1.96,
|
1596 |
+
"learning_rate": 9.048331875917061e-07,
|
1597 |
+
"loss": 2.0549,
|
1598 |
+
"step": 132500
|
1599 |
+
},
|
1600 |
+
{
|
1601 |
+
"epoch": 1.97,
|
1602 |
+
"learning_rate": 7.195684071675239e-07,
|
1603 |
+
"loss": 2.0455,
|
1604 |
+
"step": 133000
|
1605 |
+
},
|
1606 |
+
{
|
1607 |
+
"epoch": 1.98,
|
1608 |
+
"learning_rate": 5.343036267433415e-07,
|
1609 |
+
"loss": 2.0656,
|
1610 |
+
"step": 133500
|
1611 |
+
},
|
1612 |
+
{
|
1613 |
+
"epoch": 1.99,
|
1614 |
+
"learning_rate": 3.490388463191593e-07,
|
1615 |
+
"loss": 2.0592,
|
1616 |
+
"step": 134000
|
1617 |
+
},
|
1618 |
+
{
|
1619 |
+
"epoch": 1.99,
|
1620 |
+
"learning_rate": 1.637740658949771e-07,
|
1621 |
+
"loss": 2.0737,
|
1622 |
+
"step": 134500
|
1623 |
+
},
|
1624 |
+
{
|
1625 |
+
"epoch": 2.0,
|
1626 |
+
"step": 134942,
|
1627 |
+
"total_flos": 2.5064145824199475e+17,
|
1628 |
+
"train_loss": 2.1117283579871065,
|
1629 |
+
"train_runtime": 41123.1656,
|
1630 |
+
"train_samples_per_second": 6.563,
|
1631 |
+
"train_steps_per_second": 3.281
|
1632 |
+
}
|
1633 |
+
],
|
1634 |
+
"max_steps": 134942,
|
1635 |
+
"num_train_epochs": 2,
|
1636 |
+
"total_flos": 2.5064145824199475e+17,
|
1637 |
+
"trial_name": null,
|
1638 |
+
"trial_params": null
|
1639 |
+
}
|