Commit
•
c9df23c
1
Parent(s):
09fd33b
End of training
Browse files- all_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +1245 -33
all_results.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_accuracy": 0.5895981374153325,
|
| 4 |
"eval_loss": 1.842421054840088,
|
| 5 |
"eval_runtime": 6167.0931,
|
|
@@ -7,9 +7,9 @@
|
|
| 7 |
"eval_samples_per_second": 21.881,
|
| 8 |
"eval_steps_per_second": 10.94,
|
| 9 |
"perplexity": 6.311800993960819,
|
| 10 |
-
"train_loss":
|
| 11 |
-
"train_runtime":
|
| 12 |
"train_samples": 134942,
|
| 13 |
-
"train_samples_per_second":
|
| 14 |
-
"train_steps_per_second":
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
"eval_accuracy": 0.5895981374153325,
|
| 4 |
"eval_loss": 1.842421054840088,
|
| 5 |
"eval_runtime": 6167.0931,
|
|
|
|
| 7 |
"eval_samples_per_second": 21.881,
|
| 8 |
"eval_steps_per_second": 10.94,
|
| 9 |
"perplexity": 6.311800993960819,
|
| 10 |
+
"train_loss": 0.612412237187797,
|
| 11 |
+
"train_runtime": 12901.2681,
|
| 12 |
"train_samples": 134942,
|
| 13 |
+
"train_samples_per_second": 20.919,
|
| 14 |
+
"train_steps_per_second": 20.919
|
| 15 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 134942,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.612412237187797,
|
| 4 |
+
"train_runtime": 12901.2681,
|
| 5 |
"train_samples": 134942,
|
| 6 |
+
"train_samples_per_second": 20.919,
|
| 7 |
+
"train_steps_per_second": 20.919
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -1988,60 +1988,1272 @@
|
|
| 1988 |
"step": 165000
|
| 1989 |
},
|
| 1990 |
{
|
| 1991 |
-
"epoch":
|
| 1992 |
-
"learning_rate":
|
| 1993 |
-
"loss": 1.
|
| 1994 |
"step": 165500
|
| 1995 |
},
|
| 1996 |
{
|
| 1997 |
-
"epoch":
|
| 1998 |
-
"learning_rate":
|
| 1999 |
-
"loss": 1.
|
| 2000 |
"step": 166000
|
| 2001 |
},
|
| 2002 |
{
|
| 2003 |
-
"epoch":
|
| 2004 |
-
"learning_rate":
|
| 2005 |
-
"loss": 1.
|
| 2006 |
"step": 166500
|
| 2007 |
},
|
| 2008 |
{
|
| 2009 |
-
"epoch":
|
| 2010 |
-
"learning_rate":
|
| 2011 |
-
"loss": 1.
|
| 2012 |
"step": 167000
|
| 2013 |
},
|
| 2014 |
{
|
| 2015 |
-
"epoch":
|
| 2016 |
-
"learning_rate": 3.
|
| 2017 |
-
"loss": 1.
|
| 2018 |
"step": 167500
|
| 2019 |
},
|
| 2020 |
{
|
| 2021 |
-
"epoch":
|
| 2022 |
-
"learning_rate":
|
| 2023 |
-
"loss": 1.
|
| 2024 |
"step": 168000
|
| 2025 |
},
|
| 2026 |
{
|
| 2027 |
-
"epoch":
|
| 2028 |
-
"learning_rate":
|
| 2029 |
-
"loss": 1.
|
| 2030 |
"step": 168500
|
| 2031 |
},
|
| 2032 |
{
|
| 2033 |
-
"epoch":
|
| 2034 |
-
"
|
| 2035 |
-
"
|
| 2036 |
-
"
|
| 2037 |
-
|
| 2038 |
-
|
| 2039 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2040 |
}
|
| 2041 |
],
|
| 2042 |
-
"max_steps":
|
| 2043 |
-
"num_train_epochs":
|
| 2044 |
-
"total_flos":
|
| 2045 |
"trial_name": null,
|
| 2046 |
"trial_params": null
|
| 2047 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"global_step": 269884,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 1988 |
"step": 165000
|
| 1989 |
},
|
| 1990 |
{
|
| 1991 |
+
"epoch": 1.23,
|
| 1992 |
+
"learning_rate": 3.7735471535919134e-05,
|
| 1993 |
+
"loss": 1.7432,
|
| 1994 |
"step": 165500
|
| 1995 |
},
|
| 1996 |
{
|
| 1997 |
+
"epoch": 1.23,
|
| 1998 |
+
"learning_rate": 3.76984185798343e-05,
|
| 1999 |
+
"loss": 1.7559,
|
| 2000 |
"step": 166000
|
| 2001 |
},
|
| 2002 |
{
|
| 2003 |
+
"epoch": 1.23,
|
| 2004 |
+
"learning_rate": 3.766136562374947e-05,
|
| 2005 |
+
"loss": 1.7691,
|
| 2006 |
"step": 166500
|
| 2007 |
},
|
| 2008 |
{
|
| 2009 |
+
"epoch": 1.24,
|
| 2010 |
+
"learning_rate": 3.762431266766463e-05,
|
| 2011 |
+
"loss": 1.7535,
|
| 2012 |
"step": 167000
|
| 2013 |
},
|
| 2014 |
{
|
| 2015 |
+
"epoch": 1.24,
|
| 2016 |
+
"learning_rate": 3.7587259711579795e-05,
|
| 2017 |
+
"loss": 1.7727,
|
| 2018 |
"step": 167500
|
| 2019 |
},
|
| 2020 |
{
|
| 2021 |
+
"epoch": 1.24,
|
| 2022 |
+
"learning_rate": 3.755020675549495e-05,
|
| 2023 |
+
"loss": 1.7932,
|
| 2024 |
"step": 168000
|
| 2025 |
},
|
| 2026 |
{
|
| 2027 |
+
"epoch": 1.25,
|
| 2028 |
+
"learning_rate": 3.7513153799410116e-05,
|
| 2029 |
+
"loss": 1.7775,
|
| 2030 |
"step": 168500
|
| 2031 |
},
|
| 2032 |
{
|
| 2033 |
+
"epoch": 1.25,
|
| 2034 |
+
"learning_rate": 3.7476100843325286e-05,
|
| 2035 |
+
"loss": 1.7774,
|
| 2036 |
+
"step": 169000
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 1.26,
|
| 2040 |
+
"learning_rate": 3.743904788724045e-05,
|
| 2041 |
+
"loss": 1.7898,
|
| 2042 |
+
"step": 169500
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 1.26,
|
| 2046 |
+
"learning_rate": 3.7401994931155607e-05,
|
| 2047 |
+
"loss": 1.7823,
|
| 2048 |
+
"step": 170000
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 1.26,
|
| 2052 |
+
"learning_rate": 3.736494197507077e-05,
|
| 2053 |
+
"loss": 1.8264,
|
| 2054 |
+
"step": 170500
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 1.27,
|
| 2058 |
+
"learning_rate": 3.7327889018985934e-05,
|
| 2059 |
+
"loss": 1.792,
|
| 2060 |
+
"step": 171000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 1.27,
|
| 2064 |
+
"learning_rate": 3.7290836062901104e-05,
|
| 2065 |
+
"loss": 1.8131,
|
| 2066 |
+
"step": 171500
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 1.27,
|
| 2070 |
+
"learning_rate": 3.725378310681626e-05,
|
| 2071 |
+
"loss": 1.8023,
|
| 2072 |
+
"step": 172000
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 1.28,
|
| 2076 |
+
"learning_rate": 3.7216730150731425e-05,
|
| 2077 |
+
"loss": 1.8306,
|
| 2078 |
+
"step": 172500
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 1.28,
|
| 2082 |
+
"learning_rate": 3.717967719464659e-05,
|
| 2083 |
+
"loss": 1.7956,
|
| 2084 |
+
"step": 173000
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 1.29,
|
| 2088 |
+
"learning_rate": 3.714262423856175e-05,
|
| 2089 |
+
"loss": 1.8114,
|
| 2090 |
+
"step": 173500
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 1.29,
|
| 2094 |
+
"learning_rate": 3.710557128247692e-05,
|
| 2095 |
+
"loss": 1.8154,
|
| 2096 |
+
"step": 174000
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 1.29,
|
| 2100 |
+
"learning_rate": 3.706851832639208e-05,
|
| 2101 |
+
"loss": 1.8034,
|
| 2102 |
+
"step": 174500
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 1.3,
|
| 2106 |
+
"learning_rate": 3.703146537030724e-05,
|
| 2107 |
+
"loss": 1.8171,
|
| 2108 |
+
"step": 175000
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 1.3,
|
| 2112 |
+
"learning_rate": 3.699441241422241e-05,
|
| 2113 |
+
"loss": 1.8111,
|
| 2114 |
+
"step": 175500
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 1.3,
|
| 2118 |
+
"learning_rate": 3.695735945813757e-05,
|
| 2119 |
+
"loss": 1.8063,
|
| 2120 |
+
"step": 176000
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 1.31,
|
| 2124 |
+
"learning_rate": 3.6920306502052734e-05,
|
| 2125 |
+
"loss": 1.828,
|
| 2126 |
+
"step": 176500
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 1.31,
|
| 2130 |
+
"learning_rate": 3.68832535459679e-05,
|
| 2131 |
+
"loss": 1.8398,
|
| 2132 |
+
"step": 177000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 1.32,
|
| 2136 |
+
"learning_rate": 3.684620058988306e-05,
|
| 2137 |
+
"loss": 1.8362,
|
| 2138 |
+
"step": 177500
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 1.32,
|
| 2142 |
+
"learning_rate": 3.6809147633798225e-05,
|
| 2143 |
+
"loss": 1.8502,
|
| 2144 |
+
"step": 178000
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 1.32,
|
| 2148 |
+
"learning_rate": 3.6772094677713396e-05,
|
| 2149 |
+
"loss": 1.8581,
|
| 2150 |
+
"step": 178500
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 1.33,
|
| 2154 |
+
"learning_rate": 3.673504172162855e-05,
|
| 2155 |
+
"loss": 1.8401,
|
| 2156 |
+
"step": 179000
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 1.33,
|
| 2160 |
+
"learning_rate": 3.6697988765543716e-05,
|
| 2161 |
+
"loss": 1.8324,
|
| 2162 |
+
"step": 179500
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 1.33,
|
| 2166 |
+
"learning_rate": 3.666093580945888e-05,
|
| 2167 |
+
"loss": 1.8523,
|
| 2168 |
+
"step": 180000
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 1.34,
|
| 2172 |
+
"learning_rate": 1.6559707133435105e-05,
|
| 2173 |
+
"loss": 1.817,
|
| 2174 |
+
"step": 180500
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 1.34,
|
| 2178 |
+
"learning_rate": 1.6467074743223014e-05,
|
| 2179 |
+
"loss": 1.8734,
|
| 2180 |
+
"step": 181000
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 1.35,
|
| 2184 |
+
"learning_rate": 1.6374442353010923e-05,
|
| 2185 |
+
"loss": 1.812,
|
| 2186 |
+
"step": 181500
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 1.35,
|
| 2190 |
+
"learning_rate": 1.6281809962798833e-05,
|
| 2191 |
+
"loss": 1.8239,
|
| 2192 |
+
"step": 182000
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.35,
|
| 2196 |
+
"learning_rate": 1.6189177572586742e-05,
|
| 2197 |
+
"loss": 1.8156,
|
| 2198 |
+
"step": 182500
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 1.36,
|
| 2202 |
+
"learning_rate": 1.609654518237465e-05,
|
| 2203 |
+
"loss": 1.8019,
|
| 2204 |
+
"step": 183000
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 1.36,
|
| 2208 |
+
"learning_rate": 1.600391279216256e-05,
|
| 2209 |
+
"loss": 1.8153,
|
| 2210 |
+
"step": 183500
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 1.36,
|
| 2214 |
+
"learning_rate": 1.591128040195047e-05,
|
| 2215 |
+
"loss": 1.8063,
|
| 2216 |
+
"step": 184000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 1.37,
|
| 2220 |
+
"learning_rate": 1.581864801173838e-05,
|
| 2221 |
+
"loss": 1.8251,
|
| 2222 |
+
"step": 184500
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 1.37,
|
| 2226 |
+
"learning_rate": 1.5726015621526284e-05,
|
| 2227 |
+
"loss": 1.8205,
|
| 2228 |
+
"step": 185000
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 1.37,
|
| 2232 |
+
"learning_rate": 1.5633383231314193e-05,
|
| 2233 |
+
"loss": 1.805,
|
| 2234 |
+
"step": 185500
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 1.38,
|
| 2238 |
+
"learning_rate": 1.5540750841102102e-05,
|
| 2239 |
+
"loss": 1.839,
|
| 2240 |
+
"step": 186000
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 1.38,
|
| 2244 |
+
"learning_rate": 1.544811845089001e-05,
|
| 2245 |
+
"loss": 1.8183,
|
| 2246 |
+
"step": 186500
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 1.39,
|
| 2250 |
+
"learning_rate": 1.535548606067792e-05,
|
| 2251 |
+
"loss": 1.8262,
|
| 2252 |
+
"step": 187000
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 1.39,
|
| 2256 |
+
"learning_rate": 1.526285367046583e-05,
|
| 2257 |
+
"loss": 1.7968,
|
| 2258 |
+
"step": 187500
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 1.39,
|
| 2262 |
+
"learning_rate": 1.5170221280253737e-05,
|
| 2263 |
+
"loss": 1.822,
|
| 2264 |
+
"step": 188000
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 1.4,
|
| 2268 |
+
"learning_rate": 1.507758889004165e-05,
|
| 2269 |
+
"loss": 1.8239,
|
| 2270 |
+
"step": 188500
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 1.4,
|
| 2274 |
+
"learning_rate": 1.4984956499829559e-05,
|
| 2275 |
+
"loss": 1.8336,
|
| 2276 |
+
"step": 189000
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.4,
|
| 2280 |
+
"learning_rate": 1.4892324109617466e-05,
|
| 2281 |
+
"loss": 1.8248,
|
| 2282 |
+
"step": 189500
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 1.41,
|
| 2286 |
+
"learning_rate": 1.4799691719405376e-05,
|
| 2287 |
+
"loss": 1.8142,
|
| 2288 |
+
"step": 190000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 1.41,
|
| 2292 |
+
"learning_rate": 1.4707059329193285e-05,
|
| 2293 |
+
"loss": 1.8187,
|
| 2294 |
+
"step": 190500
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 1.42,
|
| 2298 |
+
"learning_rate": 1.4614426938981194e-05,
|
| 2299 |
+
"loss": 1.8219,
|
| 2300 |
+
"step": 191000
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 1.42,
|
| 2304 |
+
"learning_rate": 1.4521794548769101e-05,
|
| 2305 |
+
"loss": 1.8274,
|
| 2306 |
+
"step": 191500
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 1.42,
|
| 2310 |
+
"learning_rate": 1.442916215855701e-05,
|
| 2311 |
+
"loss": 1.7992,
|
| 2312 |
+
"step": 192000
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 1.43,
|
| 2316 |
+
"learning_rate": 1.433652976834492e-05,
|
| 2317 |
+
"loss": 1.8345,
|
| 2318 |
+
"step": 192500
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.43,
|
| 2322 |
+
"learning_rate": 1.4243897378132829e-05,
|
| 2323 |
+
"loss": 1.8225,
|
| 2324 |
+
"step": 193000
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 1.43,
|
| 2328 |
+
"learning_rate": 1.4151264987920736e-05,
|
| 2329 |
+
"loss": 1.8417,
|
| 2330 |
+
"step": 193500
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 1.44,
|
| 2334 |
+
"learning_rate": 1.4058632597708645e-05,
|
| 2335 |
+
"loss": 1.8025,
|
| 2336 |
+
"step": 194000
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 1.44,
|
| 2340 |
+
"learning_rate": 1.3966000207496555e-05,
|
| 2341 |
+
"loss": 1.818,
|
| 2342 |
+
"step": 194500
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 1.45,
|
| 2346 |
+
"learning_rate": 1.3873367817284464e-05,
|
| 2347 |
+
"loss": 1.803,
|
| 2348 |
+
"step": 195000
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 1.45,
|
| 2352 |
+
"learning_rate": 1.3780735427072371e-05,
|
| 2353 |
+
"loss": 1.8012,
|
| 2354 |
+
"step": 195500
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 1.45,
|
| 2358 |
+
"learning_rate": 1.368810303686028e-05,
|
| 2359 |
+
"loss": 1.8183,
|
| 2360 |
+
"step": 196000
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.46,
|
| 2364 |
+
"learning_rate": 1.359547064664819e-05,
|
| 2365 |
+
"loss": 1.822,
|
| 2366 |
+
"step": 196500
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 1.46,
|
| 2370 |
+
"learning_rate": 1.3502838256436099e-05,
|
| 2371 |
+
"loss": 1.8482,
|
| 2372 |
+
"step": 197000
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 1.46,
|
| 2376 |
+
"learning_rate": 1.3410205866224008e-05,
|
| 2377 |
+
"loss": 1.8322,
|
| 2378 |
+
"step": 197500
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 1.47,
|
| 2382 |
+
"learning_rate": 1.3317573476011915e-05,
|
| 2383 |
+
"loss": 1.8187,
|
| 2384 |
+
"step": 198000
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 1.47,
|
| 2388 |
+
"learning_rate": 1.3224941085799824e-05,
|
| 2389 |
+
"loss": 1.8166,
|
| 2390 |
+
"step": 198500
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 1.47,
|
| 2394 |
+
"learning_rate": 1.3132308695587734e-05,
|
| 2395 |
+
"loss": 1.8481,
|
| 2396 |
+
"step": 199000
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 1.48,
|
| 2400 |
+
"learning_rate": 1.3039676305375643e-05,
|
| 2401 |
+
"loss": 1.8319,
|
| 2402 |
+
"step": 199500
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.48,
|
| 2406 |
+
"learning_rate": 1.2947043915163554e-05,
|
| 2407 |
+
"loss": 1.8159,
|
| 2408 |
+
"step": 200000
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 1.49,
|
| 2412 |
+
"learning_rate": 1.2854411524951463e-05,
|
| 2413 |
+
"loss": 1.8474,
|
| 2414 |
+
"step": 200500
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 1.49,
|
| 2418 |
+
"learning_rate": 1.2761779134739372e-05,
|
| 2419 |
+
"loss": 1.8305,
|
| 2420 |
+
"step": 201000
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 1.49,
|
| 2424 |
+
"learning_rate": 1.266914674452728e-05,
|
| 2425 |
+
"loss": 1.827,
|
| 2426 |
+
"step": 201500
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 1.5,
|
| 2430 |
+
"learning_rate": 1.2576514354315189e-05,
|
| 2431 |
+
"loss": 1.8219,
|
| 2432 |
+
"step": 202000
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 1.5,
|
| 2436 |
+
"learning_rate": 1.2483881964103098e-05,
|
| 2437 |
+
"loss": 1.8142,
|
| 2438 |
+
"step": 202500
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 1.5,
|
| 2442 |
+
"learning_rate": 1.2391249573891007e-05,
|
| 2443 |
+
"loss": 1.8161,
|
| 2444 |
+
"step": 203000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 1.51,
|
| 2448 |
+
"learning_rate": 1.2298617183678914e-05,
|
| 2449 |
+
"loss": 1.8336,
|
| 2450 |
+
"step": 203500
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 1.51,
|
| 2454 |
+
"learning_rate": 1.2205984793466823e-05,
|
| 2455 |
+
"loss": 1.816,
|
| 2456 |
+
"step": 204000
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 1.52,
|
| 2460 |
+
"learning_rate": 1.2113352403254733e-05,
|
| 2461 |
+
"loss": 1.8304,
|
| 2462 |
+
"step": 204500
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 1.52,
|
| 2466 |
+
"learning_rate": 1.2020720013042642e-05,
|
| 2467 |
+
"loss": 1.8101,
|
| 2468 |
+
"step": 205000
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 1.52,
|
| 2472 |
+
"learning_rate": 1.192808762283055e-05,
|
| 2473 |
+
"loss": 1.8219,
|
| 2474 |
+
"step": 205500
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 1.53,
|
| 2478 |
+
"learning_rate": 1.1835455232618458e-05,
|
| 2479 |
+
"loss": 1.8271,
|
| 2480 |
+
"step": 206000
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 1.53,
|
| 2484 |
+
"learning_rate": 1.1742822842406368e-05,
|
| 2485 |
+
"loss": 1.8542,
|
| 2486 |
+
"step": 206500
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 1.53,
|
| 2490 |
+
"learning_rate": 1.1650190452194277e-05,
|
| 2491 |
+
"loss": 1.8504,
|
| 2492 |
+
"step": 207000
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 1.54,
|
| 2496 |
+
"learning_rate": 1.1557558061982184e-05,
|
| 2497 |
+
"loss": 1.8177,
|
| 2498 |
+
"step": 207500
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 1.54,
|
| 2502 |
+
"learning_rate": 1.1464925671770093e-05,
|
| 2503 |
+
"loss": 1.8297,
|
| 2504 |
+
"step": 208000
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 1.55,
|
| 2508 |
+
"learning_rate": 1.1372293281558004e-05,
|
| 2509 |
+
"loss": 1.8154,
|
| 2510 |
+
"step": 208500
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 1.55,
|
| 2514 |
+
"learning_rate": 1.1279660891345913e-05,
|
| 2515 |
+
"loss": 1.8371,
|
| 2516 |
+
"step": 209000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 1.55,
|
| 2520 |
+
"learning_rate": 1.118702850113382e-05,
|
| 2521 |
+
"loss": 1.8386,
|
| 2522 |
+
"step": 209500
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 1.56,
|
| 2526 |
+
"learning_rate": 1.109439611092173e-05,
|
| 2527 |
+
"loss": 1.8414,
|
| 2528 |
+
"step": 210000
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 1.56,
|
| 2532 |
+
"learning_rate": 1.1001763720709639e-05,
|
| 2533 |
+
"loss": 1.8708,
|
| 2534 |
+
"step": 210500
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 1.56,
|
| 2538 |
+
"learning_rate": 1.0909131330497548e-05,
|
| 2539 |
+
"loss": 1.8323,
|
| 2540 |
+
"step": 211000
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 1.57,
|
| 2544 |
+
"learning_rate": 1.0816498940285456e-05,
|
| 2545 |
+
"loss": 1.8587,
|
| 2546 |
+
"step": 211500
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 1.57,
|
| 2550 |
+
"learning_rate": 1.0723866550073365e-05,
|
| 2551 |
+
"loss": 1.8364,
|
| 2552 |
+
"step": 212000
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 1.57,
|
| 2556 |
+
"learning_rate": 1.0631234159861274e-05,
|
| 2557 |
+
"loss": 1.8504,
|
| 2558 |
+
"step": 212500
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 1.58,
|
| 2562 |
+
"learning_rate": 1.0538601769649183e-05,
|
| 2563 |
+
"loss": 1.8394,
|
| 2564 |
+
"step": 213000
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 1.58,
|
| 2568 |
+
"learning_rate": 1.044596937943709e-05,
|
| 2569 |
+
"loss": 1.811,
|
| 2570 |
+
"step": 213500
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 1.59,
|
| 2574 |
+
"learning_rate": 1.0353336989225001e-05,
|
| 2575 |
+
"loss": 1.8517,
|
| 2576 |
+
"step": 214000
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 1.59,
|
| 2580 |
+
"learning_rate": 1.026070459901291e-05,
|
| 2581 |
+
"loss": 1.8225,
|
| 2582 |
+
"step": 214500
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 1.59,
|
| 2586 |
+
"learning_rate": 1.016807220880082e-05,
|
| 2587 |
+
"loss": 1.8488,
|
| 2588 |
+
"step": 215000
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 1.6,
|
| 2592 |
+
"learning_rate": 1.0075439818588727e-05,
|
| 2593 |
+
"loss": 1.8289,
|
| 2594 |
+
"step": 215500
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 1.6,
|
| 2598 |
+
"learning_rate": 9.982807428376636e-06,
|
| 2599 |
+
"loss": 1.8847,
|
| 2600 |
+
"step": 216000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 1.6,
|
| 2604 |
+
"learning_rate": 9.890175038164546e-06,
|
| 2605 |
+
"loss": 1.8436,
|
| 2606 |
+
"step": 216500
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 1.61,
|
| 2610 |
+
"learning_rate": 9.797542647952455e-06,
|
| 2611 |
+
"loss": 1.8496,
|
| 2612 |
+
"step": 217000
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 1.61,
|
| 2616 |
+
"learning_rate": 9.704910257740362e-06,
|
| 2617 |
+
"loss": 1.8299,
|
| 2618 |
+
"step": 217500
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 1.62,
|
| 2622 |
+
"learning_rate": 9.612277867528271e-06,
|
| 2623 |
+
"loss": 1.8302,
|
| 2624 |
+
"step": 218000
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 1.62,
|
| 2628 |
+
"learning_rate": 9.51964547731618e-06,
|
| 2629 |
+
"loss": 1.8361,
|
| 2630 |
+
"step": 218500
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 1.62,
|
| 2634 |
+
"learning_rate": 9.42701308710409e-06,
|
| 2635 |
+
"loss": 1.8672,
|
| 2636 |
+
"step": 219000
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 1.63,
|
| 2640 |
+
"learning_rate": 9.334380696891999e-06,
|
| 2641 |
+
"loss": 1.8126,
|
| 2642 |
+
"step": 219500
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 1.63,
|
| 2646 |
+
"learning_rate": 9.241748306679908e-06,
|
| 2647 |
+
"loss": 1.8328,
|
| 2648 |
+
"step": 220000
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 1.63,
|
| 2652 |
+
"learning_rate": 9.149115916467817e-06,
|
| 2653 |
+
"loss": 1.8235,
|
| 2654 |
+
"step": 220500
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 1.64,
|
| 2658 |
+
"learning_rate": 9.056483526255726e-06,
|
| 2659 |
+
"loss": 1.834,
|
| 2660 |
+
"step": 221000
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 1.64,
|
| 2664 |
+
"learning_rate": 8.963851136043634e-06,
|
| 2665 |
+
"loss": 1.8367,
|
| 2666 |
+
"step": 221500
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 1.65,
|
| 2670 |
+
"learning_rate": 8.871218745831543e-06,
|
| 2671 |
+
"loss": 1.8178,
|
| 2672 |
+
"step": 222000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 1.65,
|
| 2676 |
+
"learning_rate": 8.778586355619452e-06,
|
| 2677 |
+
"loss": 1.827,
|
| 2678 |
+
"step": 222500
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 1.65,
|
| 2682 |
+
"learning_rate": 8.685953965407361e-06,
|
| 2683 |
+
"loss": 1.8176,
|
| 2684 |
+
"step": 223000
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 1.66,
|
| 2688 |
+
"learning_rate": 8.593321575195269e-06,
|
| 2689 |
+
"loss": 1.8043,
|
| 2690 |
+
"step": 223500
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 1.66,
|
| 2694 |
+
"learning_rate": 8.500689184983178e-06,
|
| 2695 |
+
"loss": 1.824,
|
| 2696 |
+
"step": 224000
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 1.66,
|
| 2700 |
+
"learning_rate": 8.408056794771087e-06,
|
| 2701 |
+
"loss": 1.8549,
|
| 2702 |
+
"step": 224500
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 1.67,
|
| 2706 |
+
"learning_rate": 8.315424404558996e-06,
|
| 2707 |
+
"loss": 1.8264,
|
| 2708 |
+
"step": 225000
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 1.67,
|
| 2712 |
+
"learning_rate": 8.222792014346905e-06,
|
| 2713 |
+
"loss": 1.8622,
|
| 2714 |
+
"step": 225500
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 1.67,
|
| 2718 |
+
"learning_rate": 8.130159624134814e-06,
|
| 2719 |
+
"loss": 1.8566,
|
| 2720 |
+
"step": 226000
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 1.68,
|
| 2724 |
+
"learning_rate": 8.037527233922724e-06,
|
| 2725 |
+
"loss": 1.8803,
|
| 2726 |
+
"step": 226500
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 1.68,
|
| 2730 |
+
"learning_rate": 7.944894843710633e-06,
|
| 2731 |
+
"loss": 1.8482,
|
| 2732 |
+
"step": 227000
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 1.69,
|
| 2736 |
+
"learning_rate": 7.85226245349854e-06,
|
| 2737 |
+
"loss": 1.8063,
|
| 2738 |
+
"step": 227500
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 1.69,
|
| 2742 |
+
"learning_rate": 7.75963006328645e-06,
|
| 2743 |
+
"loss": 1.8298,
|
| 2744 |
+
"step": 228000
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 1.69,
|
| 2748 |
+
"learning_rate": 7.666997673074358e-06,
|
| 2749 |
+
"loss": 1.8222,
|
| 2750 |
+
"step": 228500
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 1.7,
|
| 2754 |
+
"learning_rate": 7.574365282862267e-06,
|
| 2755 |
+
"loss": 1.8158,
|
| 2756 |
+
"step": 229000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 1.7,
|
| 2760 |
+
"learning_rate": 7.481732892650176e-06,
|
| 2761 |
+
"loss": 1.8452,
|
| 2762 |
+
"step": 229500
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 1.7,
|
| 2766 |
+
"learning_rate": 7.389100502438084e-06,
|
| 2767 |
+
"loss": 1.8256,
|
| 2768 |
+
"step": 230000
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 1.71,
|
| 2772 |
+
"learning_rate": 7.296468112225993e-06,
|
| 2773 |
+
"loss": 1.7952,
|
| 2774 |
+
"step": 230500
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 1.71,
|
| 2778 |
+
"learning_rate": 7.203835722013903e-06,
|
| 2779 |
+
"loss": 1.8425,
|
| 2780 |
+
"step": 231000
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 1.72,
|
| 2784 |
+
"learning_rate": 7.111203331801812e-06,
|
| 2785 |
+
"loss": 1.838,
|
| 2786 |
+
"step": 231500
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 1.72,
|
| 2790 |
+
"learning_rate": 7.018570941589721e-06,
|
| 2791 |
+
"loss": 1.8279,
|
| 2792 |
+
"step": 232000
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 1.72,
|
| 2796 |
+
"learning_rate": 6.925938551377629e-06,
|
| 2797 |
+
"loss": 1.8274,
|
| 2798 |
+
"step": 232500
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 1.73,
|
| 2802 |
+
"learning_rate": 6.833306161165538e-06,
|
| 2803 |
+
"loss": 1.8322,
|
| 2804 |
+
"step": 233000
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 1.73,
|
| 2808 |
+
"learning_rate": 6.7406737709534474e-06,
|
| 2809 |
+
"loss": 1.8451,
|
| 2810 |
+
"step": 233500
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 1.73,
|
| 2814 |
+
"learning_rate": 6.648041380741356e-06,
|
| 2815 |
+
"loss": 1.8469,
|
| 2816 |
+
"step": 234000
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 1.74,
|
| 2820 |
+
"learning_rate": 6.555408990529265e-06,
|
| 2821 |
+
"loss": 1.8336,
|
| 2822 |
+
"step": 234500
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 1.74,
|
| 2826 |
+
"learning_rate": 6.462776600317173e-06,
|
| 2827 |
+
"loss": 1.8538,
|
| 2828 |
+
"step": 235000
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 1.75,
|
| 2832 |
+
"learning_rate": 6.370144210105082e-06,
|
| 2833 |
+
"loss": 1.8638,
|
| 2834 |
+
"step": 235500
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 1.75,
|
| 2838 |
+
"learning_rate": 6.277511819892991e-06,
|
| 2839 |
+
"loss": 1.8497,
|
| 2840 |
+
"step": 236000
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 1.75,
|
| 2844 |
+
"learning_rate": 6.184879429680901e-06,
|
| 2845 |
+
"loss": 1.8547,
|
| 2846 |
+
"step": 236500
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 1.76,
|
| 2850 |
+
"learning_rate": 6.092247039468809e-06,
|
| 2851 |
+
"loss": 1.8248,
|
| 2852 |
+
"step": 237000
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 1.76,
|
| 2856 |
+
"learning_rate": 5.999614649256718e-06,
|
| 2857 |
+
"loss": 1.8354,
|
| 2858 |
+
"step": 237500
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 1.76,
|
| 2862 |
+
"learning_rate": 5.906982259044627e-06,
|
| 2863 |
+
"loss": 1.8425,
|
| 2864 |
+
"step": 238000
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 1.77,
|
| 2868 |
+
"learning_rate": 5.8143498688325356e-06,
|
| 2869 |
+
"loss": 1.8571,
|
| 2870 |
+
"step": 238500
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 1.77,
|
| 2874 |
+
"learning_rate": 5.721717478620445e-06,
|
| 2875 |
+
"loss": 1.8436,
|
| 2876 |
+
"step": 239000
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 1.77,
|
| 2880 |
+
"learning_rate": 5.629085088408354e-06,
|
| 2881 |
+
"loss": 1.863,
|
| 2882 |
+
"step": 239500
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 1.78,
|
| 2886 |
+
"learning_rate": 5.536452698196262e-06,
|
| 2887 |
+
"loss": 1.8557,
|
| 2888 |
+
"step": 240000
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 1.78,
|
| 2892 |
+
"learning_rate": 5.443820307984171e-06,
|
| 2893 |
+
"loss": 1.8647,
|
| 2894 |
+
"step": 240500
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 1.79,
|
| 2898 |
+
"learning_rate": 5.3511879177720805e-06,
|
| 2899 |
+
"loss": 1.8469,
|
| 2900 |
+
"step": 241000
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 1.79,
|
| 2904 |
+
"learning_rate": 5.258555527559989e-06,
|
| 2905 |
+
"loss": 1.8427,
|
| 2906 |
+
"step": 241500
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 1.79,
|
| 2910 |
+
"learning_rate": 5.165923137347898e-06,
|
| 2911 |
+
"loss": 1.8385,
|
| 2912 |
+
"step": 242000
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 1.8,
|
| 2916 |
+
"learning_rate": 5.073290747135806e-06,
|
| 2917 |
+
"loss": 1.8224,
|
| 2918 |
+
"step": 242500
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 1.8,
|
| 2922 |
+
"learning_rate": 4.980658356923715e-06,
|
| 2923 |
+
"loss": 1.8266,
|
| 2924 |
+
"step": 243000
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 1.8,
|
| 2928 |
+
"learning_rate": 4.8880259667116246e-06,
|
| 2929 |
+
"loss": 1.8374,
|
| 2930 |
+
"step": 243500
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 1.81,
|
| 2934 |
+
"learning_rate": 4.795393576499534e-06,
|
| 2935 |
+
"loss": 1.8227,
|
| 2936 |
+
"step": 244000
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 1.81,
|
| 2940 |
+
"learning_rate": 4.702761186287442e-06,
|
| 2941 |
+
"loss": 1.9014,
|
| 2942 |
+
"step": 244500
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 1.82,
|
| 2946 |
+
"learning_rate": 4.610128796075351e-06,
|
| 2947 |
+
"loss": 1.8057,
|
| 2948 |
+
"step": 245000
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 1.82,
|
| 2952 |
+
"learning_rate": 4.5174964058632595e-06,
|
| 2953 |
+
"loss": 1.8626,
|
| 2954 |
+
"step": 245500
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 1.82,
|
| 2958 |
+
"learning_rate": 4.424864015651169e-06,
|
| 2959 |
+
"loss": 1.8353,
|
| 2960 |
+
"step": 246000
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 1.83,
|
| 2964 |
+
"learning_rate": 4.332231625439078e-06,
|
| 2965 |
+
"loss": 1.8497,
|
| 2966 |
+
"step": 246500
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 1.83,
|
| 2970 |
+
"learning_rate": 4.239599235226987e-06,
|
| 2971 |
+
"loss": 1.8413,
|
| 2972 |
+
"step": 247000
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 1.83,
|
| 2976 |
+
"learning_rate": 4.146966845014895e-06,
|
| 2977 |
+
"loss": 1.8455,
|
| 2978 |
+
"step": 247500
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 1.84,
|
| 2982 |
+
"learning_rate": 4.054334454802804e-06,
|
| 2983 |
+
"loss": 1.8563,
|
| 2984 |
+
"step": 248000
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 1.84,
|
| 2988 |
+
"learning_rate": 3.961702064590713e-06,
|
| 2989 |
+
"loss": 1.8326,
|
| 2990 |
+
"step": 248500
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 1.85,
|
| 2994 |
+
"learning_rate": 3.869069674378622e-06,
|
| 2995 |
+
"loss": 1.862,
|
| 2996 |
+
"step": 249000
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 1.85,
|
| 3000 |
+
"learning_rate": 3.7764372841665314e-06,
|
| 3001 |
+
"loss": 1.8478,
|
| 3002 |
+
"step": 249500
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 1.85,
|
| 3006 |
+
"learning_rate": 3.68380489395444e-06,
|
| 3007 |
+
"loss": 1.8536,
|
| 3008 |
+
"step": 250000
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 1.86,
|
| 3012 |
+
"learning_rate": 3.591172503742349e-06,
|
| 3013 |
+
"loss": 1.8461,
|
| 3014 |
+
"step": 250500
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 1.86,
|
| 3018 |
+
"learning_rate": 3.4985401135302576e-06,
|
| 3019 |
+
"loss": 1.8516,
|
| 3020 |
+
"step": 251000
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 1.86,
|
| 3024 |
+
"learning_rate": 3.4059077233181664e-06,
|
| 3025 |
+
"loss": 1.8275,
|
| 3026 |
+
"step": 251500
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 1.87,
|
| 3030 |
+
"learning_rate": 3.313275333106075e-06,
|
| 3031 |
+
"loss": 1.842,
|
| 3032 |
+
"step": 252000
|
| 3033 |
+
},
|
| 3034 |
+
{
|
| 3035 |
+
"epoch": 1.87,
|
| 3036 |
+
"learning_rate": 3.2206429428939847e-06,
|
| 3037 |
+
"loss": 1.8754,
|
| 3038 |
+
"step": 252500
|
| 3039 |
+
},
|
| 3040 |
+
{
|
| 3041 |
+
"epoch": 1.87,
|
| 3042 |
+
"learning_rate": 3.1280105526818934e-06,
|
| 3043 |
+
"loss": 1.8505,
|
| 3044 |
+
"step": 253000
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 1.88,
|
| 3048 |
+
"learning_rate": 3.035378162469802e-06,
|
| 3049 |
+
"loss": 1.8354,
|
| 3050 |
+
"step": 253500
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 1.88,
|
| 3054 |
+
"learning_rate": 2.942745772257711e-06,
|
| 3055 |
+
"loss": 1.871,
|
| 3056 |
+
"step": 254000
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 1.89,
|
| 3060 |
+
"learning_rate": 2.85011338204562e-06,
|
| 3061 |
+
"loss": 1.8274,
|
| 3062 |
+
"step": 254500
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 1.89,
|
| 3066 |
+
"learning_rate": 2.7574809918335287e-06,
|
| 3067 |
+
"loss": 1.8193,
|
| 3068 |
+
"step": 255000
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 1.89,
|
| 3072 |
+
"learning_rate": 2.6648486016214375e-06,
|
| 3073 |
+
"loss": 1.8336,
|
| 3074 |
+
"step": 255500
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 1.9,
|
| 3078 |
+
"learning_rate": 2.5722162114093466e-06,
|
| 3079 |
+
"loss": 1.8599,
|
| 3080 |
+
"step": 256000
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 1.9,
|
| 3084 |
+
"learning_rate": 2.4795838211972554e-06,
|
| 3085 |
+
"loss": 1.8584,
|
| 3086 |
+
"step": 256500
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 1.9,
|
| 3090 |
+
"learning_rate": 2.386951430985164e-06,
|
| 3091 |
+
"loss": 1.8074,
|
| 3092 |
+
"step": 257000
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 1.91,
|
| 3096 |
+
"learning_rate": 2.2943190407730732e-06,
|
| 3097 |
+
"loss": 1.8509,
|
| 3098 |
+
"step": 257500
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 1.91,
|
| 3102 |
+
"learning_rate": 2.201686650560982e-06,
|
| 3103 |
+
"loss": 1.8579,
|
| 3104 |
+
"step": 258000
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 1.92,
|
| 3108 |
+
"learning_rate": 2.1090542603488907e-06,
|
| 3109 |
+
"loss": 1.8529,
|
| 3110 |
+
"step": 258500
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 1.92,
|
| 3114 |
+
"learning_rate": 2.0164218701368e-06,
|
| 3115 |
+
"loss": 1.8535,
|
| 3116 |
+
"step": 259000
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 1.92,
|
| 3120 |
+
"learning_rate": 1.9237894799247086e-06,
|
| 3121 |
+
"loss": 1.899,
|
| 3122 |
+
"step": 259500
|
| 3123 |
+
},
|
| 3124 |
+
{
|
| 3125 |
+
"epoch": 1.93,
|
| 3126 |
+
"learning_rate": 1.8311570897126173e-06,
|
| 3127 |
+
"loss": 1.8707,
|
| 3128 |
+
"step": 260000
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 1.93,
|
| 3132 |
+
"learning_rate": 1.7385246995005263e-06,
|
| 3133 |
+
"loss": 1.8482,
|
| 3134 |
+
"step": 260500
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 1.93,
|
| 3138 |
+
"learning_rate": 1.645892309288435e-06,
|
| 3139 |
+
"loss": 1.8445,
|
| 3140 |
+
"step": 261000
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 1.94,
|
| 3144 |
+
"learning_rate": 1.553259919076344e-06,
|
| 3145 |
+
"loss": 1.8472,
|
| 3146 |
+
"step": 261500
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 1.94,
|
| 3150 |
+
"learning_rate": 1.4606275288642529e-06,
|
| 3151 |
+
"loss": 1.838,
|
| 3152 |
+
"step": 262000
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 1.95,
|
| 3156 |
+
"learning_rate": 1.3679951386521616e-06,
|
| 3157 |
+
"loss": 1.8691,
|
| 3158 |
+
"step": 262500
|
| 3159 |
+
},
|
| 3160 |
+
{
|
| 3161 |
+
"epoch": 1.95,
|
| 3162 |
+
"learning_rate": 1.2753627484400705e-06,
|
| 3163 |
+
"loss": 1.8888,
|
| 3164 |
+
"step": 263000
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 1.95,
|
| 3168 |
+
"learning_rate": 1.1827303582279795e-06,
|
| 3169 |
+
"loss": 1.8937,
|
| 3170 |
+
"step": 263500
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 1.96,
|
| 3174 |
+
"learning_rate": 1.0900979680158882e-06,
|
| 3175 |
+
"loss": 1.8475,
|
| 3176 |
+
"step": 264000
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 1.96,
|
| 3180 |
+
"learning_rate": 9.974655778037972e-07,
|
| 3181 |
+
"loss": 1.8725,
|
| 3182 |
+
"step": 264500
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 1.96,
|
| 3186 |
+
"learning_rate": 9.048331875917061e-07,
|
| 3187 |
+
"loss": 1.8697,
|
| 3188 |
+
"step": 265000
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 1.97,
|
| 3192 |
+
"learning_rate": 8.12200797379615e-07,
|
| 3193 |
+
"loss": 1.8696,
|
| 3194 |
+
"step": 265500
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 1.97,
|
| 3198 |
+
"learning_rate": 7.195684071675239e-07,
|
| 3199 |
+
"loss": 1.852,
|
| 3200 |
+
"step": 266000
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 1.97,
|
| 3204 |
+
"learning_rate": 6.269360169554327e-07,
|
| 3205 |
+
"loss": 1.8908,
|
| 3206 |
+
"step": 266500
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 1.98,
|
| 3210 |
+
"learning_rate": 5.343036267433415e-07,
|
| 3211 |
+
"loss": 1.8715,
|
| 3212 |
+
"step": 267000
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 1.98,
|
| 3216 |
+
"learning_rate": 4.416712365312505e-07,
|
| 3217 |
+
"loss": 1.8736,
|
| 3218 |
+
"step": 267500
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 1.99,
|
| 3222 |
+
"learning_rate": 3.490388463191593e-07,
|
| 3223 |
+
"loss": 1.8706,
|
| 3224 |
+
"step": 268000
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 1.99,
|
| 3228 |
+
"learning_rate": 2.564064561070682e-07,
|
| 3229 |
+
"loss": 1.8902,
|
| 3230 |
+
"step": 268500
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 1.99,
|
| 3234 |
+
"learning_rate": 1.637740658949771e-07,
|
| 3235 |
+
"loss": 1.8956,
|
| 3236 |
+
"step": 269000
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 2.0,
|
| 3240 |
+
"learning_rate": 7.114167568288598e-08,
|
| 3241 |
+
"loss": 1.8512,
|
| 3242 |
+
"step": 269500
|
| 3243 |
+
},
|
| 3244 |
+
{
|
| 3245 |
+
"epoch": 2.0,
|
| 3246 |
+
"step": 269884,
|
| 3247 |
+
"total_flos": 7.103408724277985e+17,
|
| 3248 |
+
"train_loss": 0.612412237187797,
|
| 3249 |
+
"train_runtime": 12901.2681,
|
| 3250 |
+
"train_samples_per_second": 20.919,
|
| 3251 |
+
"train_steps_per_second": 20.919
|
| 3252 |
}
|
| 3253 |
],
|
| 3254 |
+
"max_steps": 269884,
|
| 3255 |
+
"num_train_epochs": 2,
|
| 3256 |
+
"total_flos": 7.103408724277985e+17,
|
| 3257 |
"trial_name": null,
|
| 3258 |
"trial_params": null
|
| 3259 |
}
|