Commit
•
e148429
1
Parent(s):
d6c3dd1
Training in progress, epoch 4, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1575259780
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f5cb2cfc5117f3531c8bd4177812ffc142572d96d99c17ff8e99066dae40dc25
|
| 3 |
size 1575259780
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3150397656
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e076027da10369decf3426e665cdd5f8a63464a96851c9cfa686d1bf2b95c3a
|
| 3 |
size 3150397656
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:65256b72b395fe9c8d3f39501b23485eb6c1ae7d0e4d1f4dbbccadf77e8bf228
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a39dca76f7733101110373dca301368b725f4d4854876ebc9e72e6ecc87f8875
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -3523,6 +3523,1174 @@
|
|
| 3523 |
"eval_samples_per_second": 0.887,
|
| 3524 |
"eval_steps_per_second": 0.444,
|
| 3525 |
"step": 11608
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3526 |
}
|
| 3527 |
],
|
| 3528 |
"logging_steps": 20,
|
|
@@ -3542,7 +4710,7 @@
|
|
| 3542 |
"attributes": {}
|
| 3543 |
}
|
| 3544 |
},
|
| 3545 |
-
"total_flos":
|
| 3546 |
"train_batch_size": 2,
|
| 3547 |
"trial_name": null,
|
| 3548 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.1685,
|
| 3 |
+
"best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-15478",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 15478,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3523 |
"eval_samples_per_second": 0.887,
|
| 3524 |
"eval_steps_per_second": 0.444,
|
| 3525 |
"step": 11608
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 3.0029719602015765,
|
| 3529 |
+
"learning_rate": 1.8149228575132891e-07,
|
| 3530 |
+
"loss": 2.8225,
|
| 3531 |
+
"step": 11620
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 3.0081405866391004,
|
| 3535 |
+
"learning_rate": 1.814598729417866e-07,
|
| 3536 |
+
"loss": 2.7593,
|
| 3537 |
+
"step": 11640
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 3.013309213076625,
|
| 3541 |
+
"learning_rate": 1.8142746013224426e-07,
|
| 3542 |
+
"loss": 2.7975,
|
| 3543 |
+
"step": 11660
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 3.018477839514149,
|
| 3547 |
+
"learning_rate": 1.8139504732270193e-07,
|
| 3548 |
+
"loss": 2.8483,
|
| 3549 |
+
"step": 11680
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 3.0236464659516735,
|
| 3553 |
+
"learning_rate": 1.8136263451315961e-07,
|
| 3554 |
+
"loss": 2.8307,
|
| 3555 |
+
"step": 11700
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 3.0288150923891974,
|
| 3559 |
+
"learning_rate": 1.8133022170361725e-07,
|
| 3560 |
+
"loss": 2.7256,
|
| 3561 |
+
"step": 11720
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 3.033983718826722,
|
| 3565 |
+
"learning_rate": 1.8129780889407494e-07,
|
| 3566 |
+
"loss": 2.8067,
|
| 3567 |
+
"step": 11740
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 3.039152345264246,
|
| 3571 |
+
"learning_rate": 1.812653960845326e-07,
|
| 3572 |
+
"loss": 2.7359,
|
| 3573 |
+
"step": 11760
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 3.04432097170177,
|
| 3577 |
+
"learning_rate": 1.8123298327499026e-07,
|
| 3578 |
+
"loss": 2.8118,
|
| 3579 |
+
"step": 11780
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 3.0494895981392944,
|
| 3583 |
+
"learning_rate": 1.8120057046544795e-07,
|
| 3584 |
+
"loss": 2.8233,
|
| 3585 |
+
"step": 11800
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 3.054658224576819,
|
| 3589 |
+
"learning_rate": 1.811681576559056e-07,
|
| 3590 |
+
"loss": 2.8124,
|
| 3591 |
+
"step": 11820
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 3.059826851014343,
|
| 3595 |
+
"learning_rate": 1.8113574484636327e-07,
|
| 3596 |
+
"loss": 2.7403,
|
| 3597 |
+
"step": 11840
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 3.064995477451867,
|
| 3601 |
+
"learning_rate": 1.8110333203682096e-07,
|
| 3602 |
+
"loss": 2.8573,
|
| 3603 |
+
"step": 11860
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 3.0701641038893914,
|
| 3607 |
+
"learning_rate": 1.810709192272786e-07,
|
| 3608 |
+
"loss": 2.8055,
|
| 3609 |
+
"step": 11880
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 3.075332730326916,
|
| 3613 |
+
"learning_rate": 1.8103850641773629e-07,
|
| 3614 |
+
"loss": 2.7713,
|
| 3615 |
+
"step": 11900
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 3.0805013567644397,
|
| 3619 |
+
"learning_rate": 1.8100609360819395e-07,
|
| 3620 |
+
"loss": 2.7659,
|
| 3621 |
+
"step": 11920
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 3.085669983201964,
|
| 3625 |
+
"learning_rate": 1.809736807986516e-07,
|
| 3626 |
+
"loss": 2.8172,
|
| 3627 |
+
"step": 11940
|
| 3628 |
+
},
|
| 3629 |
+
{
|
| 3630 |
+
"epoch": 3.0908386096394884,
|
| 3631 |
+
"learning_rate": 1.809412679891093e-07,
|
| 3632 |
+
"loss": 2.8002,
|
| 3633 |
+
"step": 11960
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 3.0960072360770123,
|
| 3637 |
+
"learning_rate": 1.8090885517956696e-07,
|
| 3638 |
+
"loss": 2.8898,
|
| 3639 |
+
"step": 11980
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 3.1011758625145367,
|
| 3643 |
+
"learning_rate": 1.8087644237002462e-07,
|
| 3644 |
+
"loss": 2.8703,
|
| 3645 |
+
"step": 12000
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 3.106344488952061,
|
| 3649 |
+
"learning_rate": 1.808440295604823e-07,
|
| 3650 |
+
"loss": 2.7963,
|
| 3651 |
+
"step": 12020
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 3.1115131153895854,
|
| 3655 |
+
"learning_rate": 1.8081161675093997e-07,
|
| 3656 |
+
"loss": 2.7938,
|
| 3657 |
+
"step": 12040
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 3.1166817418271093,
|
| 3661 |
+
"learning_rate": 1.8077920394139763e-07,
|
| 3662 |
+
"loss": 2.7748,
|
| 3663 |
+
"step": 12060
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 3.1218503682646337,
|
| 3667 |
+
"learning_rate": 1.8074679113185532e-07,
|
| 3668 |
+
"loss": 2.7813,
|
| 3669 |
+
"step": 12080
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 3.127018994702158,
|
| 3673 |
+
"learning_rate": 1.8071437832231296e-07,
|
| 3674 |
+
"loss": 2.8148,
|
| 3675 |
+
"step": 12100
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 3.132187621139682,
|
| 3679 |
+
"learning_rate": 1.8068196551277065e-07,
|
| 3680 |
+
"loss": 2.7859,
|
| 3681 |
+
"step": 12120
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 3.1373562475772063,
|
| 3685 |
+
"learning_rate": 1.806495527032283e-07,
|
| 3686 |
+
"loss": 2.7769,
|
| 3687 |
+
"step": 12140
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 3.1425248740147307,
|
| 3691 |
+
"learning_rate": 1.8061713989368597e-07,
|
| 3692 |
+
"loss": 2.8138,
|
| 3693 |
+
"step": 12160
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 3.147693500452255,
|
| 3697 |
+
"learning_rate": 1.8058472708414366e-07,
|
| 3698 |
+
"loss": 2.8262,
|
| 3699 |
+
"step": 12180
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 3.152862126889779,
|
| 3703 |
+
"learning_rate": 1.8055231427460132e-07,
|
| 3704 |
+
"loss": 2.7442,
|
| 3705 |
+
"step": 12200
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 3.1580307533273033,
|
| 3709 |
+
"learning_rate": 1.8051990146505898e-07,
|
| 3710 |
+
"loss": 2.7795,
|
| 3711 |
+
"step": 12220
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 3.1631993797648277,
|
| 3715 |
+
"learning_rate": 1.8048748865551667e-07,
|
| 3716 |
+
"loss": 2.8031,
|
| 3717 |
+
"step": 12240
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 3.1683680062023516,
|
| 3721 |
+
"learning_rate": 1.804550758459743e-07,
|
| 3722 |
+
"loss": 2.7879,
|
| 3723 |
+
"step": 12260
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 3.173536632639876,
|
| 3727 |
+
"learning_rate": 1.80422663036432e-07,
|
| 3728 |
+
"loss": 2.8254,
|
| 3729 |
+
"step": 12280
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 3.1787052590774003,
|
| 3733 |
+
"learning_rate": 1.8039025022688966e-07,
|
| 3734 |
+
"loss": 2.7154,
|
| 3735 |
+
"step": 12300
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 3.1838738855149242,
|
| 3739 |
+
"learning_rate": 1.8035783741734732e-07,
|
| 3740 |
+
"loss": 2.8229,
|
| 3741 |
+
"step": 12320
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 3.1890425119524486,
|
| 3745 |
+
"learning_rate": 1.80325424607805e-07,
|
| 3746 |
+
"loss": 2.7942,
|
| 3747 |
+
"step": 12340
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 3.194211138389973,
|
| 3751 |
+
"learning_rate": 1.8029301179826267e-07,
|
| 3752 |
+
"loss": 2.7733,
|
| 3753 |
+
"step": 12360
|
| 3754 |
+
},
|
| 3755 |
+
{
|
| 3756 |
+
"epoch": 3.199379764827497,
|
| 3757 |
+
"learning_rate": 1.8026059898872033e-07,
|
| 3758 |
+
"loss": 2.775,
|
| 3759 |
+
"step": 12380
|
| 3760 |
+
},
|
| 3761 |
+
{
|
| 3762 |
+
"epoch": 3.2045483912650212,
|
| 3763 |
+
"learning_rate": 1.8022818617917802e-07,
|
| 3764 |
+
"loss": 2.7004,
|
| 3765 |
+
"step": 12400
|
| 3766 |
+
},
|
| 3767 |
+
{
|
| 3768 |
+
"epoch": 3.2097170177025456,
|
| 3769 |
+
"learning_rate": 1.8019577336963565e-07,
|
| 3770 |
+
"loss": 2.8086,
|
| 3771 |
+
"step": 12420
|
| 3772 |
+
},
|
| 3773 |
+
{
|
| 3774 |
+
"epoch": 3.21488564414007,
|
| 3775 |
+
"learning_rate": 1.8016336056009334e-07,
|
| 3776 |
+
"loss": 2.7877,
|
| 3777 |
+
"step": 12440
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 3.220054270577594,
|
| 3781 |
+
"learning_rate": 1.80130947750551e-07,
|
| 3782 |
+
"loss": 2.8041,
|
| 3783 |
+
"step": 12460
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 3.225222897015118,
|
| 3787 |
+
"learning_rate": 1.8009853494100867e-07,
|
| 3788 |
+
"loss": 2.8454,
|
| 3789 |
+
"step": 12480
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 3.2303915234526426,
|
| 3793 |
+
"learning_rate": 1.8006612213146635e-07,
|
| 3794 |
+
"loss": 2.8097,
|
| 3795 |
+
"step": 12500
|
| 3796 |
+
},
|
| 3797 |
+
{
|
| 3798 |
+
"epoch": 3.2355601498901665,
|
| 3799 |
+
"learning_rate": 1.8003370932192402e-07,
|
| 3800 |
+
"loss": 2.8309,
|
| 3801 |
+
"step": 12520
|
| 3802 |
+
},
|
| 3803 |
+
{
|
| 3804 |
+
"epoch": 3.240728776327691,
|
| 3805 |
+
"learning_rate": 1.8000129651238168e-07,
|
| 3806 |
+
"loss": 2.7646,
|
| 3807 |
+
"step": 12540
|
| 3808 |
+
},
|
| 3809 |
+
{
|
| 3810 |
+
"epoch": 3.245897402765215,
|
| 3811 |
+
"learning_rate": 1.7996888370283937e-07,
|
| 3812 |
+
"loss": 2.7901,
|
| 3813 |
+
"step": 12560
|
| 3814 |
+
},
|
| 3815 |
+
{
|
| 3816 |
+
"epoch": 3.2510660292027396,
|
| 3817 |
+
"learning_rate": 1.79936470893297e-07,
|
| 3818 |
+
"loss": 2.7912,
|
| 3819 |
+
"step": 12580
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 3.2562346556402635,
|
| 3823 |
+
"learning_rate": 1.799040580837547e-07,
|
| 3824 |
+
"loss": 2.8004,
|
| 3825 |
+
"step": 12600
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 3.261403282077788,
|
| 3829 |
+
"learning_rate": 1.7987164527421235e-07,
|
| 3830 |
+
"loss": 2.7432,
|
| 3831 |
+
"step": 12620
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 3.266571908515312,
|
| 3835 |
+
"learning_rate": 1.7983923246467001e-07,
|
| 3836 |
+
"loss": 2.7917,
|
| 3837 |
+
"step": 12640
|
| 3838 |
+
},
|
| 3839 |
+
{
|
| 3840 |
+
"epoch": 3.271740534952836,
|
| 3841 |
+
"learning_rate": 1.798068196551277e-07,
|
| 3842 |
+
"loss": 2.8395,
|
| 3843 |
+
"step": 12660
|
| 3844 |
+
},
|
| 3845 |
+
{
|
| 3846 |
+
"epoch": 3.2769091613903605,
|
| 3847 |
+
"learning_rate": 1.7977440684558536e-07,
|
| 3848 |
+
"loss": 2.7566,
|
| 3849 |
+
"step": 12680
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 3.282077787827885,
|
| 3853 |
+
"learning_rate": 1.7974199403604303e-07,
|
| 3854 |
+
"loss": 2.7603,
|
| 3855 |
+
"step": 12700
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 3.287246414265409,
|
| 3859 |
+
"learning_rate": 1.7970958122650071e-07,
|
| 3860 |
+
"loss": 2.7443,
|
| 3861 |
+
"step": 12720
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 3.292415040702933,
|
| 3865 |
+
"learning_rate": 1.7967716841695835e-07,
|
| 3866 |
+
"loss": 2.7244,
|
| 3867 |
+
"step": 12740
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 3.2975836671404575,
|
| 3871 |
+
"learning_rate": 1.7964475560741604e-07,
|
| 3872 |
+
"loss": 2.7736,
|
| 3873 |
+
"step": 12760
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 3.302752293577982,
|
| 3877 |
+
"learning_rate": 1.7961234279787373e-07,
|
| 3878 |
+
"loss": 2.7509,
|
| 3879 |
+
"step": 12780
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 3.3079209200155058,
|
| 3883 |
+
"learning_rate": 1.7957992998833136e-07,
|
| 3884 |
+
"loss": 2.8058,
|
| 3885 |
+
"step": 12800
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 3.31308954645303,
|
| 3889 |
+
"learning_rate": 1.7954751717878905e-07,
|
| 3890 |
+
"loss": 2.7647,
|
| 3891 |
+
"step": 12820
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 3.3182581728905545,
|
| 3895 |
+
"learning_rate": 1.795151043692467e-07,
|
| 3896 |
+
"loss": 2.7883,
|
| 3897 |
+
"step": 12840
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 3.3234267993280784,
|
| 3901 |
+
"learning_rate": 1.7948269155970437e-07,
|
| 3902 |
+
"loss": 2.8089,
|
| 3903 |
+
"step": 12860
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 3.3285954257656027,
|
| 3907 |
+
"learning_rate": 1.7945027875016206e-07,
|
| 3908 |
+
"loss": 2.7858,
|
| 3909 |
+
"step": 12880
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 3.333764052203127,
|
| 3913 |
+
"learning_rate": 1.7941786594061972e-07,
|
| 3914 |
+
"loss": 2.7449,
|
| 3915 |
+
"step": 12900
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 3.338932678640651,
|
| 3919 |
+
"learning_rate": 1.7938545313107739e-07,
|
| 3920 |
+
"loss": 2.7406,
|
| 3921 |
+
"step": 12920
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 3.3441013050781754,
|
| 3925 |
+
"learning_rate": 1.7935304032153507e-07,
|
| 3926 |
+
"loss": 2.8112,
|
| 3927 |
+
"step": 12940
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 3.3492699315156997,
|
| 3931 |
+
"learning_rate": 1.793206275119927e-07,
|
| 3932 |
+
"loss": 2.7738,
|
| 3933 |
+
"step": 12960
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 3.354438557953224,
|
| 3937 |
+
"learning_rate": 1.792882147024504e-07,
|
| 3938 |
+
"loss": 2.7914,
|
| 3939 |
+
"step": 12980
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 3.359607184390748,
|
| 3943 |
+
"learning_rate": 1.7925580189290806e-07,
|
| 3944 |
+
"loss": 2.8128,
|
| 3945 |
+
"step": 13000
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 3.3647758108282724,
|
| 3949 |
+
"learning_rate": 1.7922338908336572e-07,
|
| 3950 |
+
"loss": 2.8135,
|
| 3951 |
+
"step": 13020
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 3.3699444372657967,
|
| 3955 |
+
"learning_rate": 1.791909762738234e-07,
|
| 3956 |
+
"loss": 2.7928,
|
| 3957 |
+
"step": 13040
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 3.3751130637033206,
|
| 3961 |
+
"learning_rate": 1.7915856346428107e-07,
|
| 3962 |
+
"loss": 2.7973,
|
| 3963 |
+
"step": 13060
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 3.380281690140845,
|
| 3967 |
+
"learning_rate": 1.7912615065473873e-07,
|
| 3968 |
+
"loss": 2.8088,
|
| 3969 |
+
"step": 13080
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 3.3854503165783694,
|
| 3973 |
+
"learning_rate": 1.7909373784519642e-07,
|
| 3974 |
+
"loss": 2.7603,
|
| 3975 |
+
"step": 13100
|
| 3976 |
+
},
|
| 3977 |
+
{
|
| 3978 |
+
"epoch": 3.3906189430158937,
|
| 3979 |
+
"learning_rate": 1.7906132503565406e-07,
|
| 3980 |
+
"loss": 2.7204,
|
| 3981 |
+
"step": 13120
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 3.3957875694534176,
|
| 3985 |
+
"learning_rate": 1.7902891222611175e-07,
|
| 3986 |
+
"loss": 2.7494,
|
| 3987 |
+
"step": 13140
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 3.400956195890942,
|
| 3991 |
+
"learning_rate": 1.7899649941656943e-07,
|
| 3992 |
+
"loss": 2.7304,
|
| 3993 |
+
"step": 13160
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 3.4061248223284664,
|
| 3997 |
+
"learning_rate": 1.7896408660702707e-07,
|
| 3998 |
+
"loss": 2.7921,
|
| 3999 |
+
"step": 13180
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 3.4112934487659903,
|
| 4003 |
+
"learning_rate": 1.7893167379748476e-07,
|
| 4004 |
+
"loss": 2.7574,
|
| 4005 |
+
"step": 13200
|
| 4006 |
+
},
|
| 4007 |
+
{
|
| 4008 |
+
"epoch": 3.4164620752035146,
|
| 4009 |
+
"learning_rate": 1.7889926098794242e-07,
|
| 4010 |
+
"loss": 2.7649,
|
| 4011 |
+
"step": 13220
|
| 4012 |
+
},
|
| 4013 |
+
{
|
| 4014 |
+
"epoch": 3.421630701641039,
|
| 4015 |
+
"learning_rate": 1.7886684817840008e-07,
|
| 4016 |
+
"loss": 2.7618,
|
| 4017 |
+
"step": 13240
|
| 4018 |
+
},
|
| 4019 |
+
{
|
| 4020 |
+
"epoch": 3.4267993280785634,
|
| 4021 |
+
"learning_rate": 1.7883443536885777e-07,
|
| 4022 |
+
"loss": 2.7628,
|
| 4023 |
+
"step": 13260
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 3.4319679545160873,
|
| 4027 |
+
"learning_rate": 1.7880202255931543e-07,
|
| 4028 |
+
"loss": 2.7988,
|
| 4029 |
+
"step": 13280
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 3.4371365809536116,
|
| 4033 |
+
"learning_rate": 1.787696097497731e-07,
|
| 4034 |
+
"loss": 2.7147,
|
| 4035 |
+
"step": 13300
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 3.442305207391136,
|
| 4039 |
+
"learning_rate": 1.7873719694023078e-07,
|
| 4040 |
+
"loss": 2.8114,
|
| 4041 |
+
"step": 13320
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 3.44747383382866,
|
| 4045 |
+
"learning_rate": 1.7870478413068842e-07,
|
| 4046 |
+
"loss": 2.7474,
|
| 4047 |
+
"step": 13340
|
| 4048 |
+
},
|
| 4049 |
+
{
|
| 4050 |
+
"epoch": 3.4526424602661843,
|
| 4051 |
+
"learning_rate": 1.786723713211461e-07,
|
| 4052 |
+
"loss": 2.7718,
|
| 4053 |
+
"step": 13360
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 3.4578110867037086,
|
| 4057 |
+
"learning_rate": 1.786399585116038e-07,
|
| 4058 |
+
"loss": 2.821,
|
| 4059 |
+
"step": 13380
|
| 4060 |
+
},
|
| 4061 |
+
{
|
| 4062 |
+
"epoch": 3.4629797131412325,
|
| 4063 |
+
"learning_rate": 1.7860754570206143e-07,
|
| 4064 |
+
"loss": 2.706,
|
| 4065 |
+
"step": 13400
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 3.468148339578757,
|
| 4069 |
+
"learning_rate": 1.7857513289251912e-07,
|
| 4070 |
+
"loss": 2.7982,
|
| 4071 |
+
"step": 13420
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 3.4733169660162813,
|
| 4075 |
+
"learning_rate": 1.7854272008297678e-07,
|
| 4076 |
+
"loss": 2.7446,
|
| 4077 |
+
"step": 13440
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 3.478485592453805,
|
| 4081 |
+
"learning_rate": 1.7851030727343444e-07,
|
| 4082 |
+
"loss": 2.749,
|
| 4083 |
+
"step": 13460
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 3.4836542188913295,
|
| 4087 |
+
"learning_rate": 1.7847789446389213e-07,
|
| 4088 |
+
"loss": 2.7231,
|
| 4089 |
+
"step": 13480
|
| 4090 |
+
},
|
| 4091 |
+
{
|
| 4092 |
+
"epoch": 3.488822845328854,
|
| 4093 |
+
"learning_rate": 1.784454816543498e-07,
|
| 4094 |
+
"loss": 2.7026,
|
| 4095 |
+
"step": 13500
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 3.4939914717663783,
|
| 4099 |
+
"learning_rate": 1.7841306884480745e-07,
|
| 4100 |
+
"loss": 2.7534,
|
| 4101 |
+
"step": 13520
|
| 4102 |
+
},
|
| 4103 |
+
{
|
| 4104 |
+
"epoch": 3.499160098203902,
|
| 4105 |
+
"learning_rate": 1.7838065603526514e-07,
|
| 4106 |
+
"loss": 2.8366,
|
| 4107 |
+
"step": 13540
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 3.5043287246414265,
|
| 4111 |
+
"learning_rate": 1.7834824322572278e-07,
|
| 4112 |
+
"loss": 2.7885,
|
| 4113 |
+
"step": 13560
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 3.509497351078951,
|
| 4117 |
+
"learning_rate": 1.7831583041618047e-07,
|
| 4118 |
+
"loss": 2.7273,
|
| 4119 |
+
"step": 13580
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 3.514665977516475,
|
| 4123 |
+
"learning_rate": 1.7828341760663813e-07,
|
| 4124 |
+
"loss": 2.7864,
|
| 4125 |
+
"step": 13600
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 3.519834603953999,
|
| 4129 |
+
"learning_rate": 1.782510047970958e-07,
|
| 4130 |
+
"loss": 2.7448,
|
| 4131 |
+
"step": 13620
|
| 4132 |
+
},
|
| 4133 |
+
{
|
| 4134 |
+
"epoch": 3.5250032303915235,
|
| 4135 |
+
"learning_rate": 1.7821859198755348e-07,
|
| 4136 |
+
"loss": 2.7384,
|
| 4137 |
+
"step": 13640
|
| 4138 |
+
},
|
| 4139 |
+
{
|
| 4140 |
+
"epoch": 3.530171856829048,
|
| 4141 |
+
"learning_rate": 1.7818617917801114e-07,
|
| 4142 |
+
"loss": 2.7694,
|
| 4143 |
+
"step": 13660
|
| 4144 |
+
},
|
| 4145 |
+
{
|
| 4146 |
+
"epoch": 3.535340483266572,
|
| 4147 |
+
"learning_rate": 1.781537663684688e-07,
|
| 4148 |
+
"loss": 2.7501,
|
| 4149 |
+
"step": 13680
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 3.540509109704096,
|
| 4153 |
+
"learning_rate": 1.781213535589265e-07,
|
| 4154 |
+
"loss": 2.8046,
|
| 4155 |
+
"step": 13700
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 3.5456777361416205,
|
| 4159 |
+
"learning_rate": 1.7808894074938413e-07,
|
| 4160 |
+
"loss": 2.752,
|
| 4161 |
+
"step": 13720
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 3.5508463625791444,
|
| 4165 |
+
"learning_rate": 1.7805652793984181e-07,
|
| 4166 |
+
"loss": 2.771,
|
| 4167 |
+
"step": 13740
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 3.556014989016669,
|
| 4171 |
+
"learning_rate": 1.780241151302995e-07,
|
| 4172 |
+
"loss": 2.7431,
|
| 4173 |
+
"step": 13760
|
| 4174 |
+
},
|
| 4175 |
+
{
|
| 4176 |
+
"epoch": 3.561183615454193,
|
| 4177 |
+
"learning_rate": 1.7799170232075714e-07,
|
| 4178 |
+
"loss": 2.7601,
|
| 4179 |
+
"step": 13780
|
| 4180 |
+
},
|
| 4181 |
+
{
|
| 4182 |
+
"epoch": 3.5663522418917175,
|
| 4183 |
+
"learning_rate": 1.7795928951121483e-07,
|
| 4184 |
+
"loss": 2.8301,
|
| 4185 |
+
"step": 13800
|
| 4186 |
+
},
|
| 4187 |
+
{
|
| 4188 |
+
"epoch": 3.5715208683292414,
|
| 4189 |
+
"learning_rate": 1.779268767016725e-07,
|
| 4190 |
+
"loss": 2.7157,
|
| 4191 |
+
"step": 13820
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 3.576689494766766,
|
| 4195 |
+
"learning_rate": 1.7789446389213015e-07,
|
| 4196 |
+
"loss": 2.7602,
|
| 4197 |
+
"step": 13840
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 3.5818581212042897,
|
| 4201 |
+
"learning_rate": 1.7786205108258784e-07,
|
| 4202 |
+
"loss": 2.7926,
|
| 4203 |
+
"step": 13860
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 3.587026747641814,
|
| 4207 |
+
"learning_rate": 1.778296382730455e-07,
|
| 4208 |
+
"loss": 2.76,
|
| 4209 |
+
"step": 13880
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 3.5921953740793384,
|
| 4213 |
+
"learning_rate": 1.7779722546350316e-07,
|
| 4214 |
+
"loss": 2.735,
|
| 4215 |
+
"step": 13900
|
| 4216 |
+
},
|
| 4217 |
+
{
|
| 4218 |
+
"epoch": 3.597364000516863,
|
| 4219 |
+
"learning_rate": 1.7776481265396085e-07,
|
| 4220 |
+
"loss": 2.7026,
|
| 4221 |
+
"step": 13920
|
| 4222 |
+
},
|
| 4223 |
+
{
|
| 4224 |
+
"epoch": 3.602532626954387,
|
| 4225 |
+
"learning_rate": 1.7773239984441849e-07,
|
| 4226 |
+
"loss": 2.7219,
|
| 4227 |
+
"step": 13940
|
| 4228 |
+
},
|
| 4229 |
+
{
|
| 4230 |
+
"epoch": 3.607701253391911,
|
| 4231 |
+
"learning_rate": 1.7769998703487617e-07,
|
| 4232 |
+
"loss": 2.821,
|
| 4233 |
+
"step": 13960
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 3.6128698798294354,
|
| 4237 |
+
"learning_rate": 1.7766757422533386e-07,
|
| 4238 |
+
"loss": 2.6779,
|
| 4239 |
+
"step": 13980
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 3.6180385062669593,
|
| 4243 |
+
"learning_rate": 1.776351614157915e-07,
|
| 4244 |
+
"loss": 2.7756,
|
| 4245 |
+
"step": 14000
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 3.6232071327044837,
|
| 4249 |
+
"learning_rate": 1.7760274860624919e-07,
|
| 4250 |
+
"loss": 2.7093,
|
| 4251 |
+
"step": 14020
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 3.628375759142008,
|
| 4255 |
+
"learning_rate": 1.7757033579670685e-07,
|
| 4256 |
+
"loss": 2.8079,
|
| 4257 |
+
"step": 14040
|
| 4258 |
+
},
|
| 4259 |
+
{
|
| 4260 |
+
"epoch": 3.6335443855795324,
|
| 4261 |
+
"learning_rate": 1.775379229871645e-07,
|
| 4262 |
+
"loss": 2.7358,
|
| 4263 |
+
"step": 14060
|
| 4264 |
+
},
|
| 4265 |
+
{
|
| 4266 |
+
"epoch": 3.6387130120170563,
|
| 4267 |
+
"learning_rate": 1.775055101776222e-07,
|
| 4268 |
+
"loss": 2.7401,
|
| 4269 |
+
"step": 14080
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 3.6438816384545807,
|
| 4273 |
+
"learning_rate": 1.7747309736807986e-07,
|
| 4274 |
+
"loss": 2.7088,
|
| 4275 |
+
"step": 14100
|
| 4276 |
+
},
|
| 4277 |
+
{
|
| 4278 |
+
"epoch": 3.649050264892105,
|
| 4279 |
+
"learning_rate": 1.7744068455853752e-07,
|
| 4280 |
+
"loss": 2.6936,
|
| 4281 |
+
"step": 14120
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 3.654218891329629,
|
| 4285 |
+
"learning_rate": 1.774082717489952e-07,
|
| 4286 |
+
"loss": 2.7538,
|
| 4287 |
+
"step": 14140
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 3.6593875177671533,
|
| 4291 |
+
"learning_rate": 1.7737585893945285e-07,
|
| 4292 |
+
"loss": 2.7446,
|
| 4293 |
+
"step": 14160
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 3.6645561442046777,
|
| 4297 |
+
"learning_rate": 1.7734344612991053e-07,
|
| 4298 |
+
"loss": 2.7566,
|
| 4299 |
+
"step": 14180
|
| 4300 |
+
},
|
| 4301 |
+
{
|
| 4302 |
+
"epoch": 3.669724770642202,
|
| 4303 |
+
"learning_rate": 1.773110333203682e-07,
|
| 4304 |
+
"loss": 2.7146,
|
| 4305 |
+
"step": 14200
|
| 4306 |
+
},
|
| 4307 |
+
{
|
| 4308 |
+
"epoch": 3.674893397079726,
|
| 4309 |
+
"learning_rate": 1.7727862051082586e-07,
|
| 4310 |
+
"loss": 2.7953,
|
| 4311 |
+
"step": 14220
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 3.6800620235172503,
|
| 4315 |
+
"learning_rate": 1.7724620770128355e-07,
|
| 4316 |
+
"loss": 2.7185,
|
| 4317 |
+
"step": 14240
|
| 4318 |
+
},
|
| 4319 |
+
{
|
| 4320 |
+
"epoch": 3.6852306499547747,
|
| 4321 |
+
"learning_rate": 1.772137948917412e-07,
|
| 4322 |
+
"loss": 2.7957,
|
| 4323 |
+
"step": 14260
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 3.6903992763922986,
|
| 4327 |
+
"learning_rate": 1.7718138208219887e-07,
|
| 4328 |
+
"loss": 2.7017,
|
| 4329 |
+
"step": 14280
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 3.695567902829823,
|
| 4333 |
+
"learning_rate": 1.7714896927265656e-07,
|
| 4334 |
+
"loss": 2.7374,
|
| 4335 |
+
"step": 14300
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 3.7007365292673473,
|
| 4339 |
+
"learning_rate": 1.771165564631142e-07,
|
| 4340 |
+
"loss": 2.7442,
|
| 4341 |
+
"step": 14320
|
| 4342 |
+
},
|
| 4343 |
+
{
|
| 4344 |
+
"epoch": 3.7059051557048717,
|
| 4345 |
+
"learning_rate": 1.7708414365357188e-07,
|
| 4346 |
+
"loss": 2.7792,
|
| 4347 |
+
"step": 14340
|
| 4348 |
+
},
|
| 4349 |
+
{
|
| 4350 |
+
"epoch": 3.7110737821423956,
|
| 4351 |
+
"learning_rate": 1.7705173084402957e-07,
|
| 4352 |
+
"loss": 2.7355,
|
| 4353 |
+
"step": 14360
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 3.71624240857992,
|
| 4357 |
+
"learning_rate": 1.770193180344872e-07,
|
| 4358 |
+
"loss": 2.7571,
|
| 4359 |
+
"step": 14380
|
| 4360 |
+
},
|
| 4361 |
+
{
|
| 4362 |
+
"epoch": 3.721411035017444,
|
| 4363 |
+
"learning_rate": 1.769869052249449e-07,
|
| 4364 |
+
"loss": 2.6742,
|
| 4365 |
+
"step": 14400
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 3.726579661454968,
|
| 4369 |
+
"learning_rate": 1.7695449241540256e-07,
|
| 4370 |
+
"loss": 2.7332,
|
| 4371 |
+
"step": 14420
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 3.7317482878924926,
|
| 4375 |
+
"learning_rate": 1.7692207960586022e-07,
|
| 4376 |
+
"loss": 2.7142,
|
| 4377 |
+
"step": 14440
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 3.736916914330017,
|
| 4381 |
+
"learning_rate": 1.768896667963179e-07,
|
| 4382 |
+
"loss": 2.7682,
|
| 4383 |
+
"step": 14460
|
| 4384 |
+
},
|
| 4385 |
+
{
|
| 4386 |
+
"epoch": 3.7420855407675413,
|
| 4387 |
+
"learning_rate": 1.7685725398677557e-07,
|
| 4388 |
+
"loss": 2.8006,
|
| 4389 |
+
"step": 14480
|
| 4390 |
+
},
|
| 4391 |
+
{
|
| 4392 |
+
"epoch": 3.747254167205065,
|
| 4393 |
+
"learning_rate": 1.7682484117723323e-07,
|
| 4394 |
+
"loss": 2.7649,
|
| 4395 |
+
"step": 14500
|
| 4396 |
+
},
|
| 4397 |
+
{
|
| 4398 |
+
"epoch": 3.7524227936425896,
|
| 4399 |
+
"learning_rate": 1.7679242836769092e-07,
|
| 4400 |
+
"loss": 2.7049,
|
| 4401 |
+
"step": 14520
|
| 4402 |
+
},
|
| 4403 |
+
{
|
| 4404 |
+
"epoch": 3.7575914200801135,
|
| 4405 |
+
"learning_rate": 1.7676001555814855e-07,
|
| 4406 |
+
"loss": 2.7923,
|
| 4407 |
+
"step": 14540
|
| 4408 |
+
},
|
| 4409 |
+
{
|
| 4410 |
+
"epoch": 3.762760046517638,
|
| 4411 |
+
"learning_rate": 1.7672760274860624e-07,
|
| 4412 |
+
"loss": 2.6909,
|
| 4413 |
+
"step": 14560
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 3.767928672955162,
|
| 4417 |
+
"learning_rate": 1.7669518993906393e-07,
|
| 4418 |
+
"loss": 2.7651,
|
| 4419 |
+
"step": 14580
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 3.7730972993926866,
|
| 4423 |
+
"learning_rate": 1.7666277712952157e-07,
|
| 4424 |
+
"loss": 2.6994,
|
| 4425 |
+
"step": 14600
|
| 4426 |
+
},
|
| 4427 |
+
{
|
| 4428 |
+
"epoch": 3.7782659258302105,
|
| 4429 |
+
"learning_rate": 1.7663036431997925e-07,
|
| 4430 |
+
"loss": 2.7794,
|
| 4431 |
+
"step": 14620
|
| 4432 |
+
},
|
| 4433 |
+
{
|
| 4434 |
+
"epoch": 3.783434552267735,
|
| 4435 |
+
"learning_rate": 1.7659795151043692e-07,
|
| 4436 |
+
"loss": 2.712,
|
| 4437 |
+
"step": 14640
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 3.788603178705259,
|
| 4441 |
+
"learning_rate": 1.7656553870089458e-07,
|
| 4442 |
+
"loss": 2.698,
|
| 4443 |
+
"step": 14660
|
| 4444 |
+
},
|
| 4445 |
+
{
|
| 4446 |
+
"epoch": 3.793771805142783,
|
| 4447 |
+
"learning_rate": 1.7653312589135227e-07,
|
| 4448 |
+
"loss": 2.6473,
|
| 4449 |
+
"step": 14680
|
| 4450 |
+
},
|
| 4451 |
+
{
|
| 4452 |
+
"epoch": 3.7989404315803075,
|
| 4453 |
+
"learning_rate": 1.7650071308180993e-07,
|
| 4454 |
+
"loss": 2.7231,
|
| 4455 |
+
"step": 14700
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 3.804109058017832,
|
| 4459 |
+
"learning_rate": 1.764683002722676e-07,
|
| 4460 |
+
"loss": 2.7367,
|
| 4461 |
+
"step": 14720
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 3.809277684455356,
|
| 4465 |
+
"learning_rate": 1.7643588746272528e-07,
|
| 4466 |
+
"loss": 2.7298,
|
| 4467 |
+
"step": 14740
|
| 4468 |
+
},
|
| 4469 |
+
{
|
| 4470 |
+
"epoch": 3.81444631089288,
|
| 4471 |
+
"learning_rate": 1.764034746531829e-07,
|
| 4472 |
+
"loss": 2.7584,
|
| 4473 |
+
"step": 14760
|
| 4474 |
+
},
|
| 4475 |
+
{
|
| 4476 |
+
"epoch": 3.8196149373304045,
|
| 4477 |
+
"learning_rate": 1.763710618436406e-07,
|
| 4478 |
+
"loss": 2.7566,
|
| 4479 |
+
"step": 14780
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 3.824783563767929,
|
| 4483 |
+
"learning_rate": 1.7633864903409826e-07,
|
| 4484 |
+
"loss": 2.753,
|
| 4485 |
+
"step": 14800
|
| 4486 |
+
},
|
| 4487 |
+
{
|
| 4488 |
+
"epoch": 3.8299521902054527,
|
| 4489 |
+
"learning_rate": 1.7630623622455592e-07,
|
| 4490 |
+
"loss": 2.7702,
|
| 4491 |
+
"step": 14820
|
| 4492 |
+
},
|
| 4493 |
+
{
|
| 4494 |
+
"epoch": 3.835120816642977,
|
| 4495 |
+
"learning_rate": 1.7627382341501361e-07,
|
| 4496 |
+
"loss": 2.7236,
|
| 4497 |
+
"step": 14840
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 3.8402894430805015,
|
| 4501 |
+
"learning_rate": 1.7624141060547128e-07,
|
| 4502 |
+
"loss": 2.8067,
|
| 4503 |
+
"step": 14860
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 3.845458069518026,
|
| 4507 |
+
"learning_rate": 1.7620899779592894e-07,
|
| 4508 |
+
"loss": 2.723,
|
| 4509 |
+
"step": 14880
|
| 4510 |
+
},
|
| 4511 |
+
{
|
| 4512 |
+
"epoch": 3.8506266959555497,
|
| 4513 |
+
"learning_rate": 1.7617658498638663e-07,
|
| 4514 |
+
"loss": 2.7566,
|
| 4515 |
+
"step": 14900
|
| 4516 |
+
},
|
| 4517 |
+
{
|
| 4518 |
+
"epoch": 3.855795322393074,
|
| 4519 |
+
"learning_rate": 1.7614417217684426e-07,
|
| 4520 |
+
"loss": 2.6923,
|
| 4521 |
+
"step": 14920
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 3.860963948830598,
|
| 4525 |
+
"learning_rate": 1.7611175936730195e-07,
|
| 4526 |
+
"loss": 2.7525,
|
| 4527 |
+
"step": 14940
|
| 4528 |
+
},
|
| 4529 |
+
{
|
| 4530 |
+
"epoch": 3.8661325752681224,
|
| 4531 |
+
"learning_rate": 1.7607934655775964e-07,
|
| 4532 |
+
"loss": 2.7261,
|
| 4533 |
+
"step": 14960
|
| 4534 |
+
},
|
| 4535 |
+
{
|
| 4536 |
+
"epoch": 3.8713012017056467,
|
| 4537 |
+
"learning_rate": 1.7604693374821727e-07,
|
| 4538 |
+
"loss": 2.7879,
|
| 4539 |
+
"step": 14980
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 3.876469828143171,
|
| 4543 |
+
"learning_rate": 1.7601452093867496e-07,
|
| 4544 |
+
"loss": 2.7061,
|
| 4545 |
+
"step": 15000
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 3.8816384545806955,
|
| 4549 |
+
"learning_rate": 1.7598210812913262e-07,
|
| 4550 |
+
"loss": 2.6878,
|
| 4551 |
+
"step": 15020
|
| 4552 |
+
},
|
| 4553 |
+
{
|
| 4554 |
+
"epoch": 3.8868070810182194,
|
| 4555 |
+
"learning_rate": 1.7594969531959028e-07,
|
| 4556 |
+
"loss": 2.7831,
|
| 4557 |
+
"step": 15040
|
| 4558 |
+
},
|
| 4559 |
+
{
|
| 4560 |
+
"epoch": 3.8919757074557437,
|
| 4561 |
+
"learning_rate": 1.7591728251004797e-07,
|
| 4562 |
+
"loss": 2.751,
|
| 4563 |
+
"step": 15060
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 3.8971443338932676,
|
| 4567 |
+
"learning_rate": 1.7588486970050564e-07,
|
| 4568 |
+
"loss": 2.7191,
|
| 4569 |
+
"step": 15080
|
| 4570 |
+
},
|
| 4571 |
+
{
|
| 4572 |
+
"epoch": 3.902312960330792,
|
| 4573 |
+
"learning_rate": 1.758524568909633e-07,
|
| 4574 |
+
"loss": 2.7581,
|
| 4575 |
+
"step": 15100
|
| 4576 |
+
},
|
| 4577 |
+
{
|
| 4578 |
+
"epoch": 3.9074815867683164,
|
| 4579 |
+
"learning_rate": 1.7582004408142099e-07,
|
| 4580 |
+
"loss": 2.7016,
|
| 4581 |
+
"step": 15120
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 3.9126502132058407,
|
| 4585 |
+
"learning_rate": 1.7578763127187862e-07,
|
| 4586 |
+
"loss": 2.7306,
|
| 4587 |
+
"step": 15140
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 3.9178188396433646,
|
| 4591 |
+
"learning_rate": 1.757552184623363e-07,
|
| 4592 |
+
"loss": 2.7835,
|
| 4593 |
+
"step": 15160
|
| 4594 |
+
},
|
| 4595 |
+
{
|
| 4596 |
+
"epoch": 3.922987466080889,
|
| 4597 |
+
"learning_rate": 1.75722805652794e-07,
|
| 4598 |
+
"loss": 2.7854,
|
| 4599 |
+
"step": 15180
|
| 4600 |
+
},
|
| 4601 |
+
{
|
| 4602 |
+
"epoch": 3.9281560925184134,
|
| 4603 |
+
"learning_rate": 1.7569039284325163e-07,
|
| 4604 |
+
"loss": 2.7345,
|
| 4605 |
+
"step": 15200
|
| 4606 |
+
},
|
| 4607 |
+
{
|
| 4608 |
+
"epoch": 3.9333247189559373,
|
| 4609 |
+
"learning_rate": 1.7565798003370932e-07,
|
| 4610 |
+
"loss": 2.758,
|
| 4611 |
+
"step": 15220
|
| 4612 |
+
},
|
| 4613 |
+
{
|
| 4614 |
+
"epoch": 3.9384933453934616,
|
| 4615 |
+
"learning_rate": 1.7562556722416698e-07,
|
| 4616 |
+
"loss": 2.7777,
|
| 4617 |
+
"step": 15240
|
| 4618 |
+
},
|
| 4619 |
+
{
|
| 4620 |
+
"epoch": 3.943661971830986,
|
| 4621 |
+
"learning_rate": 1.7559315441462464e-07,
|
| 4622 |
+
"loss": 2.8035,
|
| 4623 |
+
"step": 15260
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 3.9488305982685104,
|
| 4627 |
+
"learning_rate": 1.7556074160508233e-07,
|
| 4628 |
+
"loss": 2.7233,
|
| 4629 |
+
"step": 15280
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 3.9539992247060343,
|
| 4633 |
+
"learning_rate": 1.7552832879554e-07,
|
| 4634 |
+
"loss": 2.7346,
|
| 4635 |
+
"step": 15300
|
| 4636 |
+
},
|
| 4637 |
+
{
|
| 4638 |
+
"epoch": 3.9591678511435586,
|
| 4639 |
+
"learning_rate": 1.7549591598599766e-07,
|
| 4640 |
+
"loss": 2.7189,
|
| 4641 |
+
"step": 15320
|
| 4642 |
+
},
|
| 4643 |
+
{
|
| 4644 |
+
"epoch": 3.964336477581083,
|
| 4645 |
+
"learning_rate": 1.7546350317645535e-07,
|
| 4646 |
+
"loss": 2.69,
|
| 4647 |
+
"step": 15340
|
| 4648 |
+
},
|
| 4649 |
+
{
|
| 4650 |
+
"epoch": 3.969505104018607,
|
| 4651 |
+
"learning_rate": 1.7543109036691298e-07,
|
| 4652 |
+
"loss": 2.7707,
|
| 4653 |
+
"step": 15360
|
| 4654 |
+
},
|
| 4655 |
+
{
|
| 4656 |
+
"epoch": 3.9746737304561313,
|
| 4657 |
+
"learning_rate": 1.7539867755737067e-07,
|
| 4658 |
+
"loss": 2.7585,
|
| 4659 |
+
"step": 15380
|
| 4660 |
+
},
|
| 4661 |
+
{
|
| 4662 |
+
"epoch": 3.9798423568936556,
|
| 4663 |
+
"learning_rate": 1.7536626474782833e-07,
|
| 4664 |
+
"loss": 2.6792,
|
| 4665 |
+
"step": 15400
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 3.98501098333118,
|
| 4669 |
+
"learning_rate": 1.75333851938286e-07,
|
| 4670 |
+
"loss": 2.7379,
|
| 4671 |
+
"step": 15420
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 3.990179609768704,
|
| 4675 |
+
"learning_rate": 1.7530143912874368e-07,
|
| 4676 |
+
"loss": 2.6914,
|
| 4677 |
+
"step": 15440
|
| 4678 |
+
},
|
| 4679 |
+
{
|
| 4680 |
+
"epoch": 3.9953482362062283,
|
| 4681 |
+
"learning_rate": 1.7526902631920134e-07,
|
| 4682 |
+
"loss": 2.7623,
|
| 4683 |
+
"step": 15460
|
| 4684 |
+
},
|
| 4685 |
+
{
|
| 4686 |
+
"epoch": 4.0,
|
| 4687 |
+
"eval_bleu": 0.1685,
|
| 4688 |
+
"eval_gen_len": 105.7747,
|
| 4689 |
+
"eval_loss": 2.695643186569214,
|
| 4690 |
+
"eval_runtime": 1862.1701,
|
| 4691 |
+
"eval_samples_per_second": 0.925,
|
| 4692 |
+
"eval_steps_per_second": 0.462,
|
| 4693 |
+
"step": 15478
|
| 4694 |
}
|
| 4695 |
],
|
| 4696 |
"logging_steps": 20,
|
|
|
|
| 4710 |
"attributes": {}
|
| 4711 |
}
|
| 4712 |
},
|
| 4713 |
+
"total_flos": 1.0160427286939238e+17,
|
| 4714 |
"train_batch_size": 2,
|
| 4715 |
"trial_name": null,
|
| 4716 |
"trial_params": null
|