Training in progress, epoch 2, 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:a32de7d5357341b365094d4b0bdeccdb3aae0d77fcd2fe97c3e7e0b29c7aab4e
|
| 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:b65ad8a4d4408d8ea2d47cfca03bb0202226b3d1773d79cdbcc48402654bb48a
|
| 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 sha256:6d0982c65fadbf6bf7482db7062c965549f8158635957f4bdc41338317eca289
|
| 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 sha256:6fe116f764987ad933a40001607fac90b24f963059ef9f0559216685859936f0
|
| 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": 2.
|
| 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,
|
|
@@ -2349,6 +2349,1180 @@
|
|
| 2349 |
"eval_samples_per_second": 0.876,
|
| 2350 |
"eval_steps_per_second": 0.438,
|
| 2351 |
"step": 7739
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2352 |
}
|
| 2353 |
],
|
| 2354 |
"logging_steps": 20,
|
|
@@ -2368,7 +3542,7 @@
|
|
| 2368 |
"attributes": {}
|
| 2369 |
}
|
| 2370 |
},
|
| 2371 |
-
"total_flos":
|
| 2372 |
"train_batch_size": 2,
|
| 2373 |
"trial_name": null,
|
| 2374 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.0795,
|
| 3 |
+
"best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-11608",
|
| 4 |
+
"epoch": 2.9998707843390617,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 11608,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2349 |
"eval_samples_per_second": 0.876,
|
| 2350 |
"eval_steps_per_second": 0.438,
|
| 2351 |
"step": 7739
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 2.000258431321876,
|
| 2355 |
+
"learning_rate": 1.8778037080254114e-07,
|
| 2356 |
+
"loss": 2.8846,
|
| 2357 |
+
"step": 7740
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 2.0054270577594004,
|
| 2361 |
+
"learning_rate": 1.8774795799299882e-07,
|
| 2362 |
+
"loss": 2.954,
|
| 2363 |
+
"step": 7760
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 2.010595684196925,
|
| 2367 |
+
"learning_rate": 1.8771554518345649e-07,
|
| 2368 |
+
"loss": 2.9074,
|
| 2369 |
+
"step": 7780
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 2.0157643106344487,
|
| 2373 |
+
"learning_rate": 1.8768313237391415e-07,
|
| 2374 |
+
"loss": 2.8973,
|
| 2375 |
+
"step": 7800
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 2.020932937071973,
|
| 2379 |
+
"learning_rate": 1.8765071956437184e-07,
|
| 2380 |
+
"loss": 2.9155,
|
| 2381 |
+
"step": 7820
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 2.0261015635094974,
|
| 2385 |
+
"learning_rate": 1.876183067548295e-07,
|
| 2386 |
+
"loss": 2.9162,
|
| 2387 |
+
"step": 7840
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 2.031270189947022,
|
| 2391 |
+
"learning_rate": 1.8758589394528716e-07,
|
| 2392 |
+
"loss": 2.8391,
|
| 2393 |
+
"step": 7860
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 2.0364388163845457,
|
| 2397 |
+
"learning_rate": 1.8755348113574485e-07,
|
| 2398 |
+
"loss": 2.8963,
|
| 2399 |
+
"step": 7880
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 2.04160744282207,
|
| 2403 |
+
"learning_rate": 1.8752106832620248e-07,
|
| 2404 |
+
"loss": 2.9148,
|
| 2405 |
+
"step": 7900
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 2.0467760692595944,
|
| 2409 |
+
"learning_rate": 1.8748865551666017e-07,
|
| 2410 |
+
"loss": 2.9378,
|
| 2411 |
+
"step": 7920
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 2.0519446956971183,
|
| 2415 |
+
"learning_rate": 1.8745624270711783e-07,
|
| 2416 |
+
"loss": 2.8385,
|
| 2417 |
+
"step": 7940
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 2.0571133221346427,
|
| 2421 |
+
"learning_rate": 1.874238298975755e-07,
|
| 2422 |
+
"loss": 2.9649,
|
| 2423 |
+
"step": 7960
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 2.062281948572167,
|
| 2427 |
+
"learning_rate": 1.8739141708803318e-07,
|
| 2428 |
+
"loss": 2.8578,
|
| 2429 |
+
"step": 7980
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 2.067450575009691,
|
| 2433 |
+
"learning_rate": 1.8735900427849085e-07,
|
| 2434 |
+
"loss": 2.8477,
|
| 2435 |
+
"step": 8000
|
| 2436 |
+
},
|
| 2437 |
+
{
|
| 2438 |
+
"epoch": 2.0726192014472153,
|
| 2439 |
+
"learning_rate": 1.873265914689485e-07,
|
| 2440 |
+
"loss": 2.8336,
|
| 2441 |
+
"step": 8020
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 2.0777878278847397,
|
| 2445 |
+
"learning_rate": 1.872941786594062e-07,
|
| 2446 |
+
"loss": 2.9005,
|
| 2447 |
+
"step": 8040
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 2.082956454322264,
|
| 2451 |
+
"learning_rate": 1.8726176584986386e-07,
|
| 2452 |
+
"loss": 2.9426,
|
| 2453 |
+
"step": 8060
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 2.088125080759788,
|
| 2457 |
+
"learning_rate": 1.8722935304032152e-07,
|
| 2458 |
+
"loss": 2.8823,
|
| 2459 |
+
"step": 8080
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 2.0932937071973123,
|
| 2463 |
+
"learning_rate": 1.871969402307792e-07,
|
| 2464 |
+
"loss": 2.8766,
|
| 2465 |
+
"step": 8100
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 2.0984623336348367,
|
| 2469 |
+
"learning_rate": 1.8716452742123684e-07,
|
| 2470 |
+
"loss": 2.8903,
|
| 2471 |
+
"step": 8120
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 2.1036309600723606,
|
| 2475 |
+
"learning_rate": 1.8713211461169453e-07,
|
| 2476 |
+
"loss": 2.9383,
|
| 2477 |
+
"step": 8140
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 2.108799586509885,
|
| 2481 |
+
"learning_rate": 1.870997018021522e-07,
|
| 2482 |
+
"loss": 2.8702,
|
| 2483 |
+
"step": 8160
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 2.1139682129474093,
|
| 2487 |
+
"learning_rate": 1.8706728899260986e-07,
|
| 2488 |
+
"loss": 2.8984,
|
| 2489 |
+
"step": 8180
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 2.1191368393849332,
|
| 2493 |
+
"learning_rate": 1.8703487618306754e-07,
|
| 2494 |
+
"loss": 2.934,
|
| 2495 |
+
"step": 8200
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 2.1243054658224576,
|
| 2499 |
+
"learning_rate": 1.870024633735252e-07,
|
| 2500 |
+
"loss": 2.8964,
|
| 2501 |
+
"step": 8220
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 2.129474092259982,
|
| 2505 |
+
"learning_rate": 1.8697005056398287e-07,
|
| 2506 |
+
"loss": 2.887,
|
| 2507 |
+
"step": 8240
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 2.1346427186975063,
|
| 2511 |
+
"learning_rate": 1.8693763775444056e-07,
|
| 2512 |
+
"loss": 2.8915,
|
| 2513 |
+
"step": 8260
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 2.1398113451350302,
|
| 2517 |
+
"learning_rate": 1.8690522494489822e-07,
|
| 2518 |
+
"loss": 2.9201,
|
| 2519 |
+
"step": 8280
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 2.1449799715725546,
|
| 2523 |
+
"learning_rate": 1.8687281213535588e-07,
|
| 2524 |
+
"loss": 2.8616,
|
| 2525 |
+
"step": 8300
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 2.150148598010079,
|
| 2529 |
+
"learning_rate": 1.8684039932581354e-07,
|
| 2530 |
+
"loss": 2.8828,
|
| 2531 |
+
"step": 8320
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 2.155317224447603,
|
| 2535 |
+
"learning_rate": 1.868079865162712e-07,
|
| 2536 |
+
"loss": 2.8614,
|
| 2537 |
+
"step": 8340
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 2.1604858508851272,
|
| 2541 |
+
"learning_rate": 1.867755737067289e-07,
|
| 2542 |
+
"loss": 2.8295,
|
| 2543 |
+
"step": 8360
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 2.1656544773226516,
|
| 2547 |
+
"learning_rate": 1.8674316089718655e-07,
|
| 2548 |
+
"loss": 2.8375,
|
| 2549 |
+
"step": 8380
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 2.170823103760176,
|
| 2553 |
+
"learning_rate": 1.8671074808764422e-07,
|
| 2554 |
+
"loss": 2.844,
|
| 2555 |
+
"step": 8400
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 2.1759917301977,
|
| 2559 |
+
"learning_rate": 1.866783352781019e-07,
|
| 2560 |
+
"loss": 2.8151,
|
| 2561 |
+
"step": 8420
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 2.1811603566352242,
|
| 2565 |
+
"learning_rate": 1.8664592246855957e-07,
|
| 2566 |
+
"loss": 2.8761,
|
| 2567 |
+
"step": 8440
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 2.1863289830727486,
|
| 2571 |
+
"learning_rate": 1.8661350965901723e-07,
|
| 2572 |
+
"loss": 2.9129,
|
| 2573 |
+
"step": 8460
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 2.1914976095102725,
|
| 2577 |
+
"learning_rate": 1.8658109684947492e-07,
|
| 2578 |
+
"loss": 2.8907,
|
| 2579 |
+
"step": 8480
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 2.196666235947797,
|
| 2583 |
+
"learning_rate": 1.8654868403993255e-07,
|
| 2584 |
+
"loss": 2.9062,
|
| 2585 |
+
"step": 8500
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 2.2018348623853212,
|
| 2589 |
+
"learning_rate": 1.8651627123039024e-07,
|
| 2590 |
+
"loss": 2.9185,
|
| 2591 |
+
"step": 8520
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 2.207003488822845,
|
| 2595 |
+
"learning_rate": 1.864838584208479e-07,
|
| 2596 |
+
"loss": 2.9334,
|
| 2597 |
+
"step": 8540
|
| 2598 |
+
},
|
| 2599 |
+
{
|
| 2600 |
+
"epoch": 2.2121721152603695,
|
| 2601 |
+
"learning_rate": 1.8645144561130556e-07,
|
| 2602 |
+
"loss": 2.8598,
|
| 2603 |
+
"step": 8560
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 2.217340741697894,
|
| 2607 |
+
"learning_rate": 1.8641903280176325e-07,
|
| 2608 |
+
"loss": 2.8341,
|
| 2609 |
+
"step": 8580
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 2.222509368135418,
|
| 2613 |
+
"learning_rate": 1.8638661999222091e-07,
|
| 2614 |
+
"loss": 2.9006,
|
| 2615 |
+
"step": 8600
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 2.227677994572942,
|
| 2619 |
+
"learning_rate": 1.8635420718267858e-07,
|
| 2620 |
+
"loss": 2.8851,
|
| 2621 |
+
"step": 8620
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 2.2328466210104665,
|
| 2625 |
+
"learning_rate": 1.8632179437313626e-07,
|
| 2626 |
+
"loss": 2.8807,
|
| 2627 |
+
"step": 8640
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 2.238015247447991,
|
| 2631 |
+
"learning_rate": 1.8628938156359393e-07,
|
| 2632 |
+
"loss": 2.8064,
|
| 2633 |
+
"step": 8660
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 2.2431838738855148,
|
| 2637 |
+
"learning_rate": 1.862569687540516e-07,
|
| 2638 |
+
"loss": 2.89,
|
| 2639 |
+
"step": 8680
|
| 2640 |
+
},
|
| 2641 |
+
{
|
| 2642 |
+
"epoch": 2.248352500323039,
|
| 2643 |
+
"learning_rate": 1.8622455594450928e-07,
|
| 2644 |
+
"loss": 2.8585,
|
| 2645 |
+
"step": 8700
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 2.2535211267605635,
|
| 2649 |
+
"learning_rate": 1.861921431349669e-07,
|
| 2650 |
+
"loss": 2.8825,
|
| 2651 |
+
"step": 8720
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 2.2586897531980874,
|
| 2655 |
+
"learning_rate": 1.861597303254246e-07,
|
| 2656 |
+
"loss": 2.8311,
|
| 2657 |
+
"step": 8740
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 2.2638583796356118,
|
| 2661 |
+
"learning_rate": 1.8612731751588226e-07,
|
| 2662 |
+
"loss": 2.8857,
|
| 2663 |
+
"step": 8760
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 2.269027006073136,
|
| 2667 |
+
"learning_rate": 1.8609490470633992e-07,
|
| 2668 |
+
"loss": 2.8278,
|
| 2669 |
+
"step": 8780
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 2.2741956325106605,
|
| 2673 |
+
"learning_rate": 1.860624918967976e-07,
|
| 2674 |
+
"loss": 2.8549,
|
| 2675 |
+
"step": 8800
|
| 2676 |
+
},
|
| 2677 |
+
{
|
| 2678 |
+
"epoch": 2.2793642589481844,
|
| 2679 |
+
"learning_rate": 1.8603007908725527e-07,
|
| 2680 |
+
"loss": 2.8433,
|
| 2681 |
+
"step": 8820
|
| 2682 |
+
},
|
| 2683 |
+
{
|
| 2684 |
+
"epoch": 2.2845328853857088,
|
| 2685 |
+
"learning_rate": 1.8599766627771294e-07,
|
| 2686 |
+
"loss": 2.8928,
|
| 2687 |
+
"step": 8840
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 2.289701511823233,
|
| 2691 |
+
"learning_rate": 1.8596525346817062e-07,
|
| 2692 |
+
"loss": 2.909,
|
| 2693 |
+
"step": 8860
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 2.294870138260757,
|
| 2697 |
+
"learning_rate": 1.8593284065862829e-07,
|
| 2698 |
+
"loss": 2.8203,
|
| 2699 |
+
"step": 8880
|
| 2700 |
+
},
|
| 2701 |
+
{
|
| 2702 |
+
"epoch": 2.3000387646982814,
|
| 2703 |
+
"learning_rate": 1.8590042784908595e-07,
|
| 2704 |
+
"loss": 2.8176,
|
| 2705 |
+
"step": 8900
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 2.3052073911358058,
|
| 2709 |
+
"learning_rate": 1.858680150395436e-07,
|
| 2710 |
+
"loss": 2.8396,
|
| 2711 |
+
"step": 8920
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 2.31037601757333,
|
| 2715 |
+
"learning_rate": 1.8583560223000127e-07,
|
| 2716 |
+
"loss": 2.8307,
|
| 2717 |
+
"step": 8940
|
| 2718 |
+
},
|
| 2719 |
+
{
|
| 2720 |
+
"epoch": 2.315544644010854,
|
| 2721 |
+
"learning_rate": 1.8580318942045896e-07,
|
| 2722 |
+
"loss": 2.895,
|
| 2723 |
+
"step": 8960
|
| 2724 |
+
},
|
| 2725 |
+
{
|
| 2726 |
+
"epoch": 2.3207132704483784,
|
| 2727 |
+
"learning_rate": 1.8577077661091662e-07,
|
| 2728 |
+
"loss": 2.8683,
|
| 2729 |
+
"step": 8980
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 2.3258818968859027,
|
| 2733 |
+
"learning_rate": 1.8573836380137428e-07,
|
| 2734 |
+
"loss": 2.8272,
|
| 2735 |
+
"step": 9000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 2.3310505233234267,
|
| 2739 |
+
"learning_rate": 1.8570595099183197e-07,
|
| 2740 |
+
"loss": 2.8701,
|
| 2741 |
+
"step": 9020
|
| 2742 |
+
},
|
| 2743 |
+
{
|
| 2744 |
+
"epoch": 2.336219149760951,
|
| 2745 |
+
"learning_rate": 1.8567353818228963e-07,
|
| 2746 |
+
"loss": 2.9526,
|
| 2747 |
+
"step": 9040
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 2.3413877761984754,
|
| 2751 |
+
"learning_rate": 1.856411253727473e-07,
|
| 2752 |
+
"loss": 2.8748,
|
| 2753 |
+
"step": 9060
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 2.3465564026359993,
|
| 2757 |
+
"learning_rate": 1.8560871256320498e-07,
|
| 2758 |
+
"loss": 2.8858,
|
| 2759 |
+
"step": 9080
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 2.3517250290735237,
|
| 2763 |
+
"learning_rate": 1.8557629975366262e-07,
|
| 2764 |
+
"loss": 2.9003,
|
| 2765 |
+
"step": 9100
|
| 2766 |
+
},
|
| 2767 |
+
{
|
| 2768 |
+
"epoch": 2.356893655511048,
|
| 2769 |
+
"learning_rate": 1.855438869441203e-07,
|
| 2770 |
+
"loss": 2.8281,
|
| 2771 |
+
"step": 9120
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 2.362062281948572,
|
| 2775 |
+
"learning_rate": 1.8551147413457797e-07,
|
| 2776 |
+
"loss": 2.81,
|
| 2777 |
+
"step": 9140
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 2.3672309083860963,
|
| 2781 |
+
"learning_rate": 1.8547906132503563e-07,
|
| 2782 |
+
"loss": 2.8814,
|
| 2783 |
+
"step": 9160
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 2.3723995348236206,
|
| 2787 |
+
"learning_rate": 1.8544664851549332e-07,
|
| 2788 |
+
"loss": 2.9104,
|
| 2789 |
+
"step": 9180
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 2.377568161261145,
|
| 2793 |
+
"learning_rate": 1.8541423570595098e-07,
|
| 2794 |
+
"loss": 2.8546,
|
| 2795 |
+
"step": 9200
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 2.382736787698669,
|
| 2799 |
+
"learning_rate": 1.8538182289640864e-07,
|
| 2800 |
+
"loss": 2.8443,
|
| 2801 |
+
"step": 9220
|
| 2802 |
+
},
|
| 2803 |
+
{
|
| 2804 |
+
"epoch": 2.3879054141361933,
|
| 2805 |
+
"learning_rate": 1.8534941008686633e-07,
|
| 2806 |
+
"loss": 2.8323,
|
| 2807 |
+
"step": 9240
|
| 2808 |
+
},
|
| 2809 |
+
{
|
| 2810 |
+
"epoch": 2.3930740405737176,
|
| 2811 |
+
"learning_rate": 1.85316997277324e-07,
|
| 2812 |
+
"loss": 2.8195,
|
| 2813 |
+
"step": 9260
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 2.3982426670112416,
|
| 2817 |
+
"learning_rate": 1.8528458446778166e-07,
|
| 2818 |
+
"loss": 2.7998,
|
| 2819 |
+
"step": 9280
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 2.403411293448766,
|
| 2823 |
+
"learning_rate": 1.8525217165823934e-07,
|
| 2824 |
+
"loss": 2.85,
|
| 2825 |
+
"step": 9300
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 2.4085799198862903,
|
| 2829 |
+
"learning_rate": 1.8521975884869698e-07,
|
| 2830 |
+
"loss": 2.8015,
|
| 2831 |
+
"step": 9320
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 2.4137485463238146,
|
| 2835 |
+
"learning_rate": 1.8518734603915467e-07,
|
| 2836 |
+
"loss": 2.8752,
|
| 2837 |
+
"step": 9340
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 2.4189171727613386,
|
| 2841 |
+
"learning_rate": 1.8515493322961233e-07,
|
| 2842 |
+
"loss": 2.7702,
|
| 2843 |
+
"step": 9360
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 2.424085799198863,
|
| 2847 |
+
"learning_rate": 1.8512252042007e-07,
|
| 2848 |
+
"loss": 2.9099,
|
| 2849 |
+
"step": 9380
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 2.4292544256363873,
|
| 2853 |
+
"learning_rate": 1.8509010761052768e-07,
|
| 2854 |
+
"loss": 2.8563,
|
| 2855 |
+
"step": 9400
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 2.434423052073911,
|
| 2859 |
+
"learning_rate": 1.8505769480098534e-07,
|
| 2860 |
+
"loss": 2.7581,
|
| 2861 |
+
"step": 9420
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 2.4395916785114355,
|
| 2865 |
+
"learning_rate": 1.85025281991443e-07,
|
| 2866 |
+
"loss": 2.9199,
|
| 2867 |
+
"step": 9440
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 2.44476030494896,
|
| 2871 |
+
"learning_rate": 1.849928691819007e-07,
|
| 2872 |
+
"loss": 2.8404,
|
| 2873 |
+
"step": 9460
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 2.4499289313864843,
|
| 2877 |
+
"learning_rate": 1.8496045637235835e-07,
|
| 2878 |
+
"loss": 2.8321,
|
| 2879 |
+
"step": 9480
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 2.455097557824008,
|
| 2883 |
+
"learning_rate": 1.8492804356281602e-07,
|
| 2884 |
+
"loss": 2.781,
|
| 2885 |
+
"step": 9500
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 2.4602661842615325,
|
| 2889 |
+
"learning_rate": 1.8489563075327368e-07,
|
| 2890 |
+
"loss": 2.9004,
|
| 2891 |
+
"step": 9520
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 2.465434810699057,
|
| 2895 |
+
"learning_rate": 1.8486321794373134e-07,
|
| 2896 |
+
"loss": 2.8439,
|
| 2897 |
+
"step": 9540
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 2.470603437136581,
|
| 2901 |
+
"learning_rate": 1.8483080513418903e-07,
|
| 2902 |
+
"loss": 2.9077,
|
| 2903 |
+
"step": 9560
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 2.475772063574105,
|
| 2907 |
+
"learning_rate": 1.847983923246467e-07,
|
| 2908 |
+
"loss": 2.8844,
|
| 2909 |
+
"step": 9580
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 2.4809406900116295,
|
| 2913 |
+
"learning_rate": 1.8476597951510435e-07,
|
| 2914 |
+
"loss": 2.8135,
|
| 2915 |
+
"step": 9600
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 2.4861093164491535,
|
| 2919 |
+
"learning_rate": 1.8473356670556204e-07,
|
| 2920 |
+
"loss": 2.8818,
|
| 2921 |
+
"step": 9620
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 2.491277942886678,
|
| 2925 |
+
"learning_rate": 1.847011538960197e-07,
|
| 2926 |
+
"loss": 2.8716,
|
| 2927 |
+
"step": 9640
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 2.496446569324202,
|
| 2931 |
+
"learning_rate": 1.8466874108647736e-07,
|
| 2932 |
+
"loss": 2.8514,
|
| 2933 |
+
"step": 9660
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 2.501615195761726,
|
| 2937 |
+
"learning_rate": 1.8463632827693505e-07,
|
| 2938 |
+
"loss": 2.8514,
|
| 2939 |
+
"step": 9680
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 2.5067838221992504,
|
| 2943 |
+
"learning_rate": 1.846039154673927e-07,
|
| 2944 |
+
"loss": 2.8991,
|
| 2945 |
+
"step": 9700
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 2.511952448636775,
|
| 2949 |
+
"learning_rate": 1.8457150265785038e-07,
|
| 2950 |
+
"loss": 2.8639,
|
| 2951 |
+
"step": 9720
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 2.517121075074299,
|
| 2955 |
+
"learning_rate": 1.8453908984830804e-07,
|
| 2956 |
+
"loss": 2.839,
|
| 2957 |
+
"step": 9740
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 2.522289701511823,
|
| 2961 |
+
"learning_rate": 1.845066770387657e-07,
|
| 2962 |
+
"loss": 2.8364,
|
| 2963 |
+
"step": 9760
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 2.5274583279493474,
|
| 2967 |
+
"learning_rate": 1.844742642292234e-07,
|
| 2968 |
+
"loss": 2.8632,
|
| 2969 |
+
"step": 9780
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 2.532626954386872,
|
| 2973 |
+
"learning_rate": 1.8444185141968105e-07,
|
| 2974 |
+
"loss": 2.8573,
|
| 2975 |
+
"step": 9800
|
| 2976 |
+
},
|
| 2977 |
+
{
|
| 2978 |
+
"epoch": 2.5377955808243957,
|
| 2979 |
+
"learning_rate": 1.844094386101387e-07,
|
| 2980 |
+
"loss": 2.869,
|
| 2981 |
+
"step": 9820
|
| 2982 |
+
},
|
| 2983 |
+
{
|
| 2984 |
+
"epoch": 2.54296420726192,
|
| 2985 |
+
"learning_rate": 1.843770258005964e-07,
|
| 2986 |
+
"loss": 2.8879,
|
| 2987 |
+
"step": 9840
|
| 2988 |
+
},
|
| 2989 |
+
{
|
| 2990 |
+
"epoch": 2.5481328336994444,
|
| 2991 |
+
"learning_rate": 1.8434461299105406e-07,
|
| 2992 |
+
"loss": 2.8887,
|
| 2993 |
+
"step": 9860
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 2.553301460136969,
|
| 2997 |
+
"learning_rate": 1.8431220018151172e-07,
|
| 2998 |
+
"loss": 2.817,
|
| 2999 |
+
"step": 9880
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 2.5584700865744927,
|
| 3003 |
+
"learning_rate": 1.842797873719694e-07,
|
| 3004 |
+
"loss": 2.8758,
|
| 3005 |
+
"step": 9900
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 2.563638713012017,
|
| 3009 |
+
"learning_rate": 1.8424737456242705e-07,
|
| 3010 |
+
"loss": 2.8466,
|
| 3011 |
+
"step": 9920
|
| 3012 |
+
},
|
| 3013 |
+
{
|
| 3014 |
+
"epoch": 2.5688073394495414,
|
| 3015 |
+
"learning_rate": 1.8421496175288474e-07,
|
| 3016 |
+
"loss": 2.8203,
|
| 3017 |
+
"step": 9940
|
| 3018 |
+
},
|
| 3019 |
+
{
|
| 3020 |
+
"epoch": 2.5739759658870653,
|
| 3021 |
+
"learning_rate": 1.841825489433424e-07,
|
| 3022 |
+
"loss": 2.8138,
|
| 3023 |
+
"step": 9960
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 2.5791445923245897,
|
| 3027 |
+
"learning_rate": 1.8415013613380006e-07,
|
| 3028 |
+
"loss": 2.8538,
|
| 3029 |
+
"step": 9980
|
| 3030 |
+
},
|
| 3031 |
+
{
|
| 3032 |
+
"epoch": 2.584313218762114,
|
| 3033 |
+
"learning_rate": 1.8411772332425775e-07,
|
| 3034 |
+
"loss": 2.779,
|
| 3035 |
+
"step": 10000
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 2.5894818451996384,
|
| 3039 |
+
"learning_rate": 1.840853105147154e-07,
|
| 3040 |
+
"loss": 2.8533,
|
| 3041 |
+
"step": 10020
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 2.5946504716371623,
|
| 3045 |
+
"learning_rate": 1.8405289770517307e-07,
|
| 3046 |
+
"loss": 2.8389,
|
| 3047 |
+
"step": 10040
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 2.5998190980746867,
|
| 3051 |
+
"learning_rate": 1.8402048489563076e-07,
|
| 3052 |
+
"loss": 2.9132,
|
| 3053 |
+
"step": 10060
|
| 3054 |
+
},
|
| 3055 |
+
{
|
| 3056 |
+
"epoch": 2.6049877245122106,
|
| 3057 |
+
"learning_rate": 1.8398807208608842e-07,
|
| 3058 |
+
"loss": 2.8213,
|
| 3059 |
+
"step": 10080
|
| 3060 |
+
},
|
| 3061 |
+
{
|
| 3062 |
+
"epoch": 2.610156350949735,
|
| 3063 |
+
"learning_rate": 1.8395565927654608e-07,
|
| 3064 |
+
"loss": 2.7963,
|
| 3065 |
+
"step": 10100
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 2.6153249773872593,
|
| 3069 |
+
"learning_rate": 1.8392324646700374e-07,
|
| 3070 |
+
"loss": 2.7789,
|
| 3071 |
+
"step": 10120
|
| 3072 |
+
},
|
| 3073 |
+
{
|
| 3074 |
+
"epoch": 2.6204936038247837,
|
| 3075 |
+
"learning_rate": 1.838908336574614e-07,
|
| 3076 |
+
"loss": 2.8891,
|
| 3077 |
+
"step": 10140
|
| 3078 |
+
},
|
| 3079 |
+
{
|
| 3080 |
+
"epoch": 2.625662230262308,
|
| 3081 |
+
"learning_rate": 1.838584208479191e-07,
|
| 3082 |
+
"loss": 2.8305,
|
| 3083 |
+
"step": 10160
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 2.630830856699832,
|
| 3087 |
+
"learning_rate": 1.8382600803837676e-07,
|
| 3088 |
+
"loss": 2.8333,
|
| 3089 |
+
"step": 10180
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 2.6359994831373563,
|
| 3093 |
+
"learning_rate": 1.8379359522883442e-07,
|
| 3094 |
+
"loss": 2.8771,
|
| 3095 |
+
"step": 10200
|
| 3096 |
+
},
|
| 3097 |
+
{
|
| 3098 |
+
"epoch": 2.6411681095748802,
|
| 3099 |
+
"learning_rate": 1.837611824192921e-07,
|
| 3100 |
+
"loss": 2.8126,
|
| 3101 |
+
"step": 10220
|
| 3102 |
+
},
|
| 3103 |
+
{
|
| 3104 |
+
"epoch": 2.6463367360124046,
|
| 3105 |
+
"learning_rate": 1.8372876960974977e-07,
|
| 3106 |
+
"loss": 2.8727,
|
| 3107 |
+
"step": 10240
|
| 3108 |
+
},
|
| 3109 |
+
{
|
| 3110 |
+
"epoch": 2.651505362449929,
|
| 3111 |
+
"learning_rate": 1.8369635680020743e-07,
|
| 3112 |
+
"loss": 2.8323,
|
| 3113 |
+
"step": 10260
|
| 3114 |
+
},
|
| 3115 |
+
{
|
| 3116 |
+
"epoch": 2.6566739888874533,
|
| 3117 |
+
"learning_rate": 1.8366394399066512e-07,
|
| 3118 |
+
"loss": 2.7787,
|
| 3119 |
+
"step": 10280
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 2.6618426153249772,
|
| 3123 |
+
"learning_rate": 1.8363153118112275e-07,
|
| 3124 |
+
"loss": 2.8147,
|
| 3125 |
+
"step": 10300
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 2.6670112417625016,
|
| 3129 |
+
"learning_rate": 1.8359911837158044e-07,
|
| 3130 |
+
"loss": 2.7373,
|
| 3131 |
+
"step": 10320
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 2.672179868200026,
|
| 3135 |
+
"learning_rate": 1.835667055620381e-07,
|
| 3136 |
+
"loss": 2.8365,
|
| 3137 |
+
"step": 10340
|
| 3138 |
+
},
|
| 3139 |
+
{
|
| 3140 |
+
"epoch": 2.67734849463755,
|
| 3141 |
+
"learning_rate": 1.8353429275249577e-07,
|
| 3142 |
+
"loss": 2.7914,
|
| 3143 |
+
"step": 10360
|
| 3144 |
+
},
|
| 3145 |
+
{
|
| 3146 |
+
"epoch": 2.6825171210750742,
|
| 3147 |
+
"learning_rate": 1.8350187994295346e-07,
|
| 3148 |
+
"loss": 2.8569,
|
| 3149 |
+
"step": 10380
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 2.6876857475125986,
|
| 3153 |
+
"learning_rate": 1.8346946713341112e-07,
|
| 3154 |
+
"loss": 2.8851,
|
| 3155 |
+
"step": 10400
|
| 3156 |
+
},
|
| 3157 |
+
{
|
| 3158 |
+
"epoch": 2.692854373950123,
|
| 3159 |
+
"learning_rate": 1.8343705432386878e-07,
|
| 3160 |
+
"loss": 2.8061,
|
| 3161 |
+
"step": 10420
|
| 3162 |
+
},
|
| 3163 |
+
{
|
| 3164 |
+
"epoch": 2.698023000387647,
|
| 3165 |
+
"learning_rate": 1.8340464151432647e-07,
|
| 3166 |
+
"loss": 2.8493,
|
| 3167 |
+
"step": 10440
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 2.7031916268251712,
|
| 3171 |
+
"learning_rate": 1.8337222870478413e-07,
|
| 3172 |
+
"loss": 2.7935,
|
| 3173 |
+
"step": 10460
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 2.7083602532626956,
|
| 3177 |
+
"learning_rate": 1.833398158952418e-07,
|
| 3178 |
+
"loss": 2.8418,
|
| 3179 |
+
"step": 10480
|
| 3180 |
+
},
|
| 3181 |
+
{
|
| 3182 |
+
"epoch": 2.7135288797002195,
|
| 3183 |
+
"learning_rate": 1.8330740308569948e-07,
|
| 3184 |
+
"loss": 2.7704,
|
| 3185 |
+
"step": 10500
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 2.718697506137744,
|
| 3189 |
+
"learning_rate": 1.8327499027615711e-07,
|
| 3190 |
+
"loss": 2.87,
|
| 3191 |
+
"step": 10520
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 2.723866132575268,
|
| 3195 |
+
"learning_rate": 1.832425774666148e-07,
|
| 3196 |
+
"loss": 2.8444,
|
| 3197 |
+
"step": 10540
|
| 3198 |
+
},
|
| 3199 |
+
{
|
| 3200 |
+
"epoch": 2.7290347590127926,
|
| 3201 |
+
"learning_rate": 1.8321016465707246e-07,
|
| 3202 |
+
"loss": 2.7719,
|
| 3203 |
+
"step": 10560
|
| 3204 |
+
},
|
| 3205 |
+
{
|
| 3206 |
+
"epoch": 2.7342033854503165,
|
| 3207 |
+
"learning_rate": 1.8317775184753013e-07,
|
| 3208 |
+
"loss": 2.823,
|
| 3209 |
+
"step": 10580
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 2.739372011887841,
|
| 3213 |
+
"learning_rate": 1.8314533903798782e-07,
|
| 3214 |
+
"loss": 2.8562,
|
| 3215 |
+
"step": 10600
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 2.7445406383253648,
|
| 3219 |
+
"learning_rate": 1.8311292622844548e-07,
|
| 3220 |
+
"loss": 2.8903,
|
| 3221 |
+
"step": 10620
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 2.749709264762889,
|
| 3225 |
+
"learning_rate": 1.8308051341890314e-07,
|
| 3226 |
+
"loss": 2.8069,
|
| 3227 |
+
"step": 10640
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 2.7548778912004135,
|
| 3231 |
+
"learning_rate": 1.8304810060936083e-07,
|
| 3232 |
+
"loss": 2.7696,
|
| 3233 |
+
"step": 10660
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 2.760046517637938,
|
| 3237 |
+
"learning_rate": 1.830156877998185e-07,
|
| 3238 |
+
"loss": 2.7616,
|
| 3239 |
+
"step": 10680
|
| 3240 |
+
},
|
| 3241 |
+
{
|
| 3242 |
+
"epoch": 2.765215144075462,
|
| 3243 |
+
"learning_rate": 1.8298327499027615e-07,
|
| 3244 |
+
"loss": 2.7976,
|
| 3245 |
+
"step": 10700
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 2.770383770512986,
|
| 3249 |
+
"learning_rate": 1.829508621807338e-07,
|
| 3250 |
+
"loss": 2.8024,
|
| 3251 |
+
"step": 10720
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 2.7755523969505105,
|
| 3255 |
+
"learning_rate": 1.8291844937119147e-07,
|
| 3256 |
+
"loss": 2.8195,
|
| 3257 |
+
"step": 10740
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 2.7807210233880344,
|
| 3261 |
+
"learning_rate": 1.8288603656164916e-07,
|
| 3262 |
+
"loss": 2.8792,
|
| 3263 |
+
"step": 10760
|
| 3264 |
+
},
|
| 3265 |
+
{
|
| 3266 |
+
"epoch": 2.7858896498255588,
|
| 3267 |
+
"learning_rate": 1.8285362375210682e-07,
|
| 3268 |
+
"loss": 2.8563,
|
| 3269 |
+
"step": 10780
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 2.791058276263083,
|
| 3273 |
+
"learning_rate": 1.8282121094256449e-07,
|
| 3274 |
+
"loss": 2.7792,
|
| 3275 |
+
"step": 10800
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 2.7962269027006075,
|
| 3279 |
+
"learning_rate": 1.8278879813302218e-07,
|
| 3280 |
+
"loss": 2.8382,
|
| 3281 |
+
"step": 10820
|
| 3282 |
+
},
|
| 3283 |
+
{
|
| 3284 |
+
"epoch": 2.8013955291381314,
|
| 3285 |
+
"learning_rate": 1.8275638532347984e-07,
|
| 3286 |
+
"loss": 2.7697,
|
| 3287 |
+
"step": 10840
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 2.8065641555756558,
|
| 3291 |
+
"learning_rate": 1.827239725139375e-07,
|
| 3292 |
+
"loss": 2.8915,
|
| 3293 |
+
"step": 10860
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 2.81173278201318,
|
| 3297 |
+
"learning_rate": 1.826915597043952e-07,
|
| 3298 |
+
"loss": 2.8202,
|
| 3299 |
+
"step": 10880
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 2.816901408450704,
|
| 3303 |
+
"learning_rate": 1.8265914689485282e-07,
|
| 3304 |
+
"loss": 2.7699,
|
| 3305 |
+
"step": 10900
|
| 3306 |
+
},
|
| 3307 |
+
{
|
| 3308 |
+
"epoch": 2.8220700348882284,
|
| 3309 |
+
"learning_rate": 1.826267340853105e-07,
|
| 3310 |
+
"loss": 2.8795,
|
| 3311 |
+
"step": 10920
|
| 3312 |
+
},
|
| 3313 |
+
{
|
| 3314 |
+
"epoch": 2.8272386613257527,
|
| 3315 |
+
"learning_rate": 1.8259432127576817e-07,
|
| 3316 |
+
"loss": 2.7777,
|
| 3317 |
+
"step": 10940
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 2.832407287763277,
|
| 3321 |
+
"learning_rate": 1.8256190846622583e-07,
|
| 3322 |
+
"loss": 2.8015,
|
| 3323 |
+
"step": 10960
|
| 3324 |
+
},
|
| 3325 |
+
{
|
| 3326 |
+
"epoch": 2.837575914200801,
|
| 3327 |
+
"learning_rate": 1.8252949565668352e-07,
|
| 3328 |
+
"loss": 2.8306,
|
| 3329 |
+
"step": 10980
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 2.8427445406383254,
|
| 3333 |
+
"learning_rate": 1.8249708284714118e-07,
|
| 3334 |
+
"loss": 2.986,
|
| 3335 |
+
"step": 11000
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 2.8479131670758497,
|
| 3339 |
+
"learning_rate": 1.8246467003759885e-07,
|
| 3340 |
+
"loss": 2.785,
|
| 3341 |
+
"step": 11020
|
| 3342 |
+
},
|
| 3343 |
+
{
|
| 3344 |
+
"epoch": 2.8530817935133737,
|
| 3345 |
+
"learning_rate": 1.8243225722805653e-07,
|
| 3346 |
+
"loss": 2.7796,
|
| 3347 |
+
"step": 11040
|
| 3348 |
+
},
|
| 3349 |
+
{
|
| 3350 |
+
"epoch": 2.858250419950898,
|
| 3351 |
+
"learning_rate": 1.823998444185142e-07,
|
| 3352 |
+
"loss": 2.7851,
|
| 3353 |
+
"step": 11060
|
| 3354 |
+
},
|
| 3355 |
+
{
|
| 3356 |
+
"epoch": 2.8634190463884224,
|
| 3357 |
+
"learning_rate": 1.8236743160897186e-07,
|
| 3358 |
+
"loss": 2.8133,
|
| 3359 |
+
"step": 11080
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 2.8685876728259467,
|
| 3363 |
+
"learning_rate": 1.8233501879942955e-07,
|
| 3364 |
+
"loss": 2.7767,
|
| 3365 |
+
"step": 11100
|
| 3366 |
+
},
|
| 3367 |
+
{
|
| 3368 |
+
"epoch": 2.8737562992634706,
|
| 3369 |
+
"learning_rate": 1.8230260598988718e-07,
|
| 3370 |
+
"loss": 2.8456,
|
| 3371 |
+
"step": 11120
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 2.878924925700995,
|
| 3375 |
+
"learning_rate": 1.8227019318034487e-07,
|
| 3376 |
+
"loss": 2.7697,
|
| 3377 |
+
"step": 11140
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 2.884093552138519,
|
| 3381 |
+
"learning_rate": 1.8223778037080253e-07,
|
| 3382 |
+
"loss": 2.7501,
|
| 3383 |
+
"step": 11160
|
| 3384 |
+
},
|
| 3385 |
+
{
|
| 3386 |
+
"epoch": 2.8892621785760433,
|
| 3387 |
+
"learning_rate": 1.822053675612602e-07,
|
| 3388 |
+
"loss": 2.8102,
|
| 3389 |
+
"step": 11180
|
| 3390 |
+
},
|
| 3391 |
+
{
|
| 3392 |
+
"epoch": 2.8944308050135676,
|
| 3393 |
+
"learning_rate": 1.8217295475171788e-07,
|
| 3394 |
+
"loss": 2.8303,
|
| 3395 |
+
"step": 11200
|
| 3396 |
+
},
|
| 3397 |
+
{
|
| 3398 |
+
"epoch": 2.899599431451092,
|
| 3399 |
+
"learning_rate": 1.8214054194217554e-07,
|
| 3400 |
+
"loss": 2.8061,
|
| 3401 |
+
"step": 11220
|
| 3402 |
+
},
|
| 3403 |
+
{
|
| 3404 |
+
"epoch": 2.9047680578886164,
|
| 3405 |
+
"learning_rate": 1.821081291326332e-07,
|
| 3406 |
+
"loss": 2.853,
|
| 3407 |
+
"step": 11240
|
| 3408 |
+
},
|
| 3409 |
+
{
|
| 3410 |
+
"epoch": 2.9099366843261403,
|
| 3411 |
+
"learning_rate": 1.820757163230909e-07,
|
| 3412 |
+
"loss": 2.8419,
|
| 3413 |
+
"step": 11260
|
| 3414 |
+
},
|
| 3415 |
+
{
|
| 3416 |
+
"epoch": 2.9151053107636646,
|
| 3417 |
+
"learning_rate": 1.8204330351354856e-07,
|
| 3418 |
+
"loss": 2.8818,
|
| 3419 |
+
"step": 11280
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 2.9202739372011886,
|
| 3423 |
+
"learning_rate": 1.8201089070400622e-07,
|
| 3424 |
+
"loss": 2.7992,
|
| 3425 |
+
"step": 11300
|
| 3426 |
+
},
|
| 3427 |
+
{
|
| 3428 |
+
"epoch": 2.925442563638713,
|
| 3429 |
+
"learning_rate": 1.8197847789446388e-07,
|
| 3430 |
+
"loss": 2.8553,
|
| 3431 |
+
"step": 11320
|
| 3432 |
+
},
|
| 3433 |
+
{
|
| 3434 |
+
"epoch": 2.9306111900762373,
|
| 3435 |
+
"learning_rate": 1.8194606508492154e-07,
|
| 3436 |
+
"loss": 2.7901,
|
| 3437 |
+
"step": 11340
|
| 3438 |
+
},
|
| 3439 |
+
{
|
| 3440 |
+
"epoch": 2.9357798165137616,
|
| 3441 |
+
"learning_rate": 1.8191365227537923e-07,
|
| 3442 |
+
"loss": 2.7795,
|
| 3443 |
+
"step": 11360
|
| 3444 |
+
},
|
| 3445 |
+
{
|
| 3446 |
+
"epoch": 2.9409484429512855,
|
| 3447 |
+
"learning_rate": 1.818812394658369e-07,
|
| 3448 |
+
"loss": 2.7977,
|
| 3449 |
+
"step": 11380
|
| 3450 |
+
},
|
| 3451 |
+
{
|
| 3452 |
+
"epoch": 2.94611706938881,
|
| 3453 |
+
"learning_rate": 1.8184882665629455e-07,
|
| 3454 |
+
"loss": 2.7824,
|
| 3455 |
+
"step": 11400
|
| 3456 |
+
},
|
| 3457 |
+
{
|
| 3458 |
+
"epoch": 2.9512856958263343,
|
| 3459 |
+
"learning_rate": 1.8181641384675224e-07,
|
| 3460 |
+
"loss": 2.7671,
|
| 3461 |
+
"step": 11420
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 2.956454322263858,
|
| 3465 |
+
"learning_rate": 1.817840010372099e-07,
|
| 3466 |
+
"loss": 2.8553,
|
| 3467 |
+
"step": 11440
|
| 3468 |
+
},
|
| 3469 |
+
{
|
| 3470 |
+
"epoch": 2.9616229487013825,
|
| 3471 |
+
"learning_rate": 1.8175158822766757e-07,
|
| 3472 |
+
"loss": 2.7253,
|
| 3473 |
+
"step": 11460
|
| 3474 |
+
},
|
| 3475 |
+
{
|
| 3476 |
+
"epoch": 2.966791575138907,
|
| 3477 |
+
"learning_rate": 1.8171917541812525e-07,
|
| 3478 |
+
"loss": 2.8338,
|
| 3479 |
+
"step": 11480
|
| 3480 |
+
},
|
| 3481 |
+
{
|
| 3482 |
+
"epoch": 2.9719602015764313,
|
| 3483 |
+
"learning_rate": 1.816867626085829e-07,
|
| 3484 |
+
"loss": 2.7843,
|
| 3485 |
+
"step": 11500
|
| 3486 |
+
},
|
| 3487 |
+
{
|
| 3488 |
+
"epoch": 2.977128828013955,
|
| 3489 |
+
"learning_rate": 1.8165434979904058e-07,
|
| 3490 |
+
"loss": 2.8181,
|
| 3491 |
+
"step": 11520
|
| 3492 |
+
},
|
| 3493 |
+
{
|
| 3494 |
+
"epoch": 2.9822974544514795,
|
| 3495 |
+
"learning_rate": 1.8162193698949824e-07,
|
| 3496 |
+
"loss": 2.7918,
|
| 3497 |
+
"step": 11540
|
| 3498 |
+
},
|
| 3499 |
+
{
|
| 3500 |
+
"epoch": 2.987466080889004,
|
| 3501 |
+
"learning_rate": 1.815895241799559e-07,
|
| 3502 |
+
"loss": 2.8051,
|
| 3503 |
+
"step": 11560
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 2.992634707326528,
|
| 3507 |
+
"learning_rate": 1.815571113704136e-07,
|
| 3508 |
+
"loss": 2.8276,
|
| 3509 |
+
"step": 11580
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 2.997803333764052,
|
| 3513 |
+
"learning_rate": 1.8152469856087125e-07,
|
| 3514 |
+
"loss": 2.8109,
|
| 3515 |
+
"step": 11600
|
| 3516 |
+
},
|
| 3517 |
+
{
|
| 3518 |
+
"epoch": 2.9998707843390617,
|
| 3519 |
+
"eval_bleu": 0.0795,
|
| 3520 |
+
"eval_gen_len": 111.475,
|
| 3521 |
+
"eval_loss": 2.76475191116333,
|
| 3522 |
+
"eval_runtime": 1941.3663,
|
| 3523 |
+
"eval_samples_per_second": 0.887,
|
| 3524 |
+
"eval_steps_per_second": 0.444,
|
| 3525 |
+
"step": 11608
|
| 3526 |
}
|
| 3527 |
],
|
| 3528 |
"logging_steps": 20,
|
|
|
|
| 3542 |
"attributes": {}
|
| 3543 |
}
|
| 3544 |
},
|
| 3545 |
+
"total_flos": 7.619114628911923e+16,
|
| 3546 |
"train_batch_size": 2,
|
| 3547 |
"trial_name": null,
|
| 3548 |
"trial_params": null
|