Commit
·
5cb5bc3
1
Parent(s):
a6bfe6c
Upload trainer_state.json
Browse files- trainer_state.json +1212 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1212 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.687523259161477,
|
| 5 |
+
"global_step": 810000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"learning_rate": 4.9824459158667857e-05,
|
| 13 |
+
"loss": 1.0594,
|
| 14 |
+
"step": 5000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.07,
|
| 18 |
+
"learning_rate": 4.964891831733571e-05,
|
| 19 |
+
"loss": 1.0381,
|
| 20 |
+
"step": 10000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.11,
|
| 24 |
+
"learning_rate": 4.947337747600357e-05,
|
| 25 |
+
"loss": 1.014,
|
| 26 |
+
"step": 15000
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.14,
|
| 30 |
+
"learning_rate": 4.9297836634671426e-05,
|
| 31 |
+
"loss": 0.9997,
|
| 32 |
+
"step": 20000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.18,
|
| 36 |
+
"learning_rate": 4.912229579333928e-05,
|
| 37 |
+
"loss": 0.9936,
|
| 38 |
+
"step": 25000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.21,
|
| 42 |
+
"learning_rate": 4.8946754952007134e-05,
|
| 43 |
+
"loss": 0.9765,
|
| 44 |
+
"step": 30000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.25,
|
| 48 |
+
"learning_rate": 4.877121411067499e-05,
|
| 49 |
+
"loss": 0.975,
|
| 50 |
+
"step": 35000
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.28,
|
| 54 |
+
"learning_rate": 4.859567326934284e-05,
|
| 55 |
+
"loss": 0.9627,
|
| 56 |
+
"step": 40000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.32,
|
| 60 |
+
"learning_rate": 4.84201324280107e-05,
|
| 61 |
+
"loss": 0.9641,
|
| 62 |
+
"step": 45000
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.35,
|
| 66 |
+
"learning_rate": 4.8244591586678564e-05,
|
| 67 |
+
"loss": 0.9562,
|
| 68 |
+
"step": 50000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.39,
|
| 72 |
+
"learning_rate": 4.806905074534642e-05,
|
| 73 |
+
"loss": 0.9634,
|
| 74 |
+
"step": 55000
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.42,
|
| 78 |
+
"learning_rate": 4.789350990401427e-05,
|
| 79 |
+
"loss": 0.9532,
|
| 80 |
+
"step": 60000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.46,
|
| 84 |
+
"learning_rate": 4.771796906268213e-05,
|
| 85 |
+
"loss": 0.9539,
|
| 86 |
+
"step": 65000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.49,
|
| 90 |
+
"learning_rate": 4.754242822134998e-05,
|
| 91 |
+
"loss": 0.9599,
|
| 92 |
+
"step": 70000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.53,
|
| 96 |
+
"learning_rate": 4.7366887380017835e-05,
|
| 97 |
+
"loss": 0.9437,
|
| 98 |
+
"step": 75000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.56,
|
| 102 |
+
"learning_rate": 4.7191346538685696e-05,
|
| 103 |
+
"loss": 0.9395,
|
| 104 |
+
"step": 80000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.6,
|
| 108 |
+
"learning_rate": 4.701580569735355e-05,
|
| 109 |
+
"loss": 0.9385,
|
| 110 |
+
"step": 85000
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.63,
|
| 114 |
+
"learning_rate": 4.6840264856021404e-05,
|
| 115 |
+
"loss": 0.947,
|
| 116 |
+
"step": 90000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.67,
|
| 120 |
+
"learning_rate": 4.666472401468926e-05,
|
| 121 |
+
"loss": 0.9402,
|
| 122 |
+
"step": 95000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.7,
|
| 126 |
+
"learning_rate": 4.648918317335711e-05,
|
| 127 |
+
"loss": 0.9317,
|
| 128 |
+
"step": 100000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.74,
|
| 132 |
+
"learning_rate": 4.631364233202497e-05,
|
| 133 |
+
"loss": 0.9316,
|
| 134 |
+
"step": 105000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.77,
|
| 138 |
+
"learning_rate": 4.613810149069283e-05,
|
| 139 |
+
"loss": 0.9399,
|
| 140 |
+
"step": 110000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.81,
|
| 144 |
+
"learning_rate": 4.596256064936068e-05,
|
| 145 |
+
"loss": 0.9395,
|
| 146 |
+
"step": 115000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.84,
|
| 150 |
+
"learning_rate": 4.5787019808028536e-05,
|
| 151 |
+
"loss": 0.9303,
|
| 152 |
+
"step": 120000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.88,
|
| 156 |
+
"learning_rate": 4.561147896669639e-05,
|
| 157 |
+
"loss": 0.9304,
|
| 158 |
+
"step": 125000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.91,
|
| 162 |
+
"learning_rate": 4.5435938125364244e-05,
|
| 163 |
+
"loss": 0.9126,
|
| 164 |
+
"step": 130000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.95,
|
| 168 |
+
"learning_rate": 4.5260397284032105e-05,
|
| 169 |
+
"loss": 0.9165,
|
| 170 |
+
"step": 135000
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.98,
|
| 174 |
+
"learning_rate": 4.5084856442699966e-05,
|
| 175 |
+
"loss": 0.9311,
|
| 176 |
+
"step": 140000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.02,
|
| 180 |
+
"learning_rate": 4.490931560136782e-05,
|
| 181 |
+
"loss": 0.9139,
|
| 182 |
+
"step": 145000
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.05,
|
| 186 |
+
"learning_rate": 4.4733774760035675e-05,
|
| 187 |
+
"loss": 0.9097,
|
| 188 |
+
"step": 150000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.09,
|
| 192 |
+
"learning_rate": 4.455823391870353e-05,
|
| 193 |
+
"loss": 0.9086,
|
| 194 |
+
"step": 155000
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.12,
|
| 198 |
+
"learning_rate": 4.438269307737138e-05,
|
| 199 |
+
"loss": 0.9172,
|
| 200 |
+
"step": 160000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.16,
|
| 204 |
+
"learning_rate": 4.420715223603924e-05,
|
| 205 |
+
"loss": 0.9106,
|
| 206 |
+
"step": 165000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.19,
|
| 210 |
+
"learning_rate": 4.40316113947071e-05,
|
| 211 |
+
"loss": 0.9001,
|
| 212 |
+
"step": 170000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.23,
|
| 216 |
+
"learning_rate": 4.385607055337495e-05,
|
| 217 |
+
"loss": 0.9108,
|
| 218 |
+
"step": 175000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.26,
|
| 222 |
+
"learning_rate": 4.3680529712042806e-05,
|
| 223 |
+
"loss": 0.911,
|
| 224 |
+
"step": 180000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.3,
|
| 228 |
+
"learning_rate": 4.350498887071066e-05,
|
| 229 |
+
"loss": 0.9007,
|
| 230 |
+
"step": 185000
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.33,
|
| 234 |
+
"learning_rate": 4.3329448029378515e-05,
|
| 235 |
+
"loss": 0.8908,
|
| 236 |
+
"step": 190000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.37,
|
| 240 |
+
"learning_rate": 4.315390718804637e-05,
|
| 241 |
+
"loss": 0.897,
|
| 242 |
+
"step": 195000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.4,
|
| 246 |
+
"learning_rate": 4.297836634671423e-05,
|
| 247 |
+
"loss": 0.902,
|
| 248 |
+
"step": 200000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.44,
|
| 252 |
+
"learning_rate": 4.2802825505382084e-05,
|
| 253 |
+
"loss": 0.8918,
|
| 254 |
+
"step": 205000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.47,
|
| 258 |
+
"learning_rate": 4.262728466404994e-05,
|
| 259 |
+
"loss": 0.889,
|
| 260 |
+
"step": 210000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.51,
|
| 264 |
+
"learning_rate": 4.245174382271779e-05,
|
| 265 |
+
"loss": 0.8929,
|
| 266 |
+
"step": 215000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.54,
|
| 270 |
+
"learning_rate": 4.2276202981385646e-05,
|
| 271 |
+
"loss": 0.8974,
|
| 272 |
+
"step": 220000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.58,
|
| 276 |
+
"learning_rate": 4.210066214005351e-05,
|
| 277 |
+
"loss": 0.8932,
|
| 278 |
+
"step": 225000
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.61,
|
| 282 |
+
"learning_rate": 4.192512129872136e-05,
|
| 283 |
+
"loss": 0.8901,
|
| 284 |
+
"step": 230000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.65,
|
| 288 |
+
"learning_rate": 4.174958045738922e-05,
|
| 289 |
+
"loss": 0.8849,
|
| 290 |
+
"step": 235000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.69,
|
| 294 |
+
"learning_rate": 4.1574039616057077e-05,
|
| 295 |
+
"loss": 0.8801,
|
| 296 |
+
"step": 240000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.72,
|
| 300 |
+
"learning_rate": 4.139849877472493e-05,
|
| 301 |
+
"loss": 0.8807,
|
| 302 |
+
"step": 245000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.76,
|
| 306 |
+
"learning_rate": 4.1222957933392785e-05,
|
| 307 |
+
"loss": 0.8847,
|
| 308 |
+
"step": 250000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.79,
|
| 312 |
+
"learning_rate": 4.104741709206064e-05,
|
| 313 |
+
"loss": 0.8753,
|
| 314 |
+
"step": 255000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.83,
|
| 318 |
+
"learning_rate": 4.08718762507285e-05,
|
| 319 |
+
"loss": 0.8764,
|
| 320 |
+
"step": 260000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 1.86,
|
| 324 |
+
"learning_rate": 4.0696335409396354e-05,
|
| 325 |
+
"loss": 0.8748,
|
| 326 |
+
"step": 265000
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.9,
|
| 330 |
+
"learning_rate": 4.052079456806421e-05,
|
| 331 |
+
"loss": 0.8789,
|
| 332 |
+
"step": 270000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.93,
|
| 336 |
+
"learning_rate": 4.034525372673206e-05,
|
| 337 |
+
"loss": 0.875,
|
| 338 |
+
"step": 275000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.97,
|
| 342 |
+
"learning_rate": 4.016971288539992e-05,
|
| 343 |
+
"loss": 0.8711,
|
| 344 |
+
"step": 280000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.0,
|
| 348 |
+
"learning_rate": 3.999417204406777e-05,
|
| 349 |
+
"loss": 0.8688,
|
| 350 |
+
"step": 285000
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.04,
|
| 354 |
+
"learning_rate": 3.981863120273563e-05,
|
| 355 |
+
"loss": 0.8594,
|
| 356 |
+
"step": 290000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.07,
|
| 360 |
+
"learning_rate": 3.9643090361403486e-05,
|
| 361 |
+
"loss": 0.8595,
|
| 362 |
+
"step": 295000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.11,
|
| 366 |
+
"learning_rate": 3.946754952007134e-05,
|
| 367 |
+
"loss": 0.8536,
|
| 368 |
+
"step": 300000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.14,
|
| 372 |
+
"learning_rate": 3.9292008678739194e-05,
|
| 373 |
+
"loss": 0.8618,
|
| 374 |
+
"step": 305000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.18,
|
| 378 |
+
"learning_rate": 3.911646783740705e-05,
|
| 379 |
+
"loss": 0.8511,
|
| 380 |
+
"step": 310000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.21,
|
| 384 |
+
"learning_rate": 3.894092699607491e-05,
|
| 385 |
+
"loss": 0.8543,
|
| 386 |
+
"step": 315000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.25,
|
| 390 |
+
"learning_rate": 3.8765386154742763e-05,
|
| 391 |
+
"loss": 0.856,
|
| 392 |
+
"step": 320000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.28,
|
| 396 |
+
"learning_rate": 3.8589845313410624e-05,
|
| 397 |
+
"loss": 0.8608,
|
| 398 |
+
"step": 325000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.32,
|
| 402 |
+
"learning_rate": 3.841430447207848e-05,
|
| 403 |
+
"loss": 0.8449,
|
| 404 |
+
"step": 330000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.35,
|
| 408 |
+
"learning_rate": 3.823876363074633e-05,
|
| 409 |
+
"loss": 0.8527,
|
| 410 |
+
"step": 335000
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.39,
|
| 414 |
+
"learning_rate": 3.806322278941419e-05,
|
| 415 |
+
"loss": 0.8441,
|
| 416 |
+
"step": 340000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.42,
|
| 420 |
+
"learning_rate": 3.788768194808204e-05,
|
| 421 |
+
"loss": 0.8436,
|
| 422 |
+
"step": 345000
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.46,
|
| 426 |
+
"learning_rate": 3.77121411067499e-05,
|
| 427 |
+
"loss": 0.8448,
|
| 428 |
+
"step": 350000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.49,
|
| 432 |
+
"learning_rate": 3.7536600265417756e-05,
|
| 433 |
+
"loss": 0.8348,
|
| 434 |
+
"step": 355000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.53,
|
| 438 |
+
"learning_rate": 3.736105942408561e-05,
|
| 439 |
+
"loss": 0.8466,
|
| 440 |
+
"step": 360000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.56,
|
| 444 |
+
"learning_rate": 3.7185518582753464e-05,
|
| 445 |
+
"loss": 0.8354,
|
| 446 |
+
"step": 365000
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.6,
|
| 450 |
+
"learning_rate": 3.700997774142132e-05,
|
| 451 |
+
"loss": 0.8421,
|
| 452 |
+
"step": 370000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 2.63,
|
| 456 |
+
"learning_rate": 3.683443690008917e-05,
|
| 457 |
+
"loss": 0.8354,
|
| 458 |
+
"step": 375000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.67,
|
| 462 |
+
"learning_rate": 3.6658896058757034e-05,
|
| 463 |
+
"loss": 0.8358,
|
| 464 |
+
"step": 380000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.7,
|
| 468 |
+
"learning_rate": 3.648335521742489e-05,
|
| 469 |
+
"loss": 0.8414,
|
| 470 |
+
"step": 385000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.74,
|
| 474 |
+
"learning_rate": 3.630781437609274e-05,
|
| 475 |
+
"loss": 0.8428,
|
| 476 |
+
"step": 390000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.77,
|
| 480 |
+
"learning_rate": 3.6132273534760596e-05,
|
| 481 |
+
"loss": 0.8303,
|
| 482 |
+
"step": 395000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.81,
|
| 486 |
+
"learning_rate": 3.595673269342845e-05,
|
| 487 |
+
"loss": 0.834,
|
| 488 |
+
"step": 400000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.84,
|
| 492 |
+
"learning_rate": 3.578119185209631e-05,
|
| 493 |
+
"loss": 0.8311,
|
| 494 |
+
"step": 405000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.88,
|
| 498 |
+
"learning_rate": 3.5605651010764165e-05,
|
| 499 |
+
"loss": 0.8245,
|
| 500 |
+
"step": 410000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.91,
|
| 504 |
+
"learning_rate": 3.5430110169432026e-05,
|
| 505 |
+
"loss": 0.8211,
|
| 506 |
+
"step": 415000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.95,
|
| 510 |
+
"learning_rate": 3.525456932809988e-05,
|
| 511 |
+
"loss": 0.8154,
|
| 512 |
+
"step": 420000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.98,
|
| 516 |
+
"learning_rate": 3.5079028486767735e-05,
|
| 517 |
+
"loss": 0.8196,
|
| 518 |
+
"step": 425000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 3.02,
|
| 522 |
+
"learning_rate": 3.490348764543559e-05,
|
| 523 |
+
"loss": 0.8188,
|
| 524 |
+
"step": 430000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 3.05,
|
| 528 |
+
"learning_rate": 3.472794680410344e-05,
|
| 529 |
+
"loss": 0.81,
|
| 530 |
+
"step": 435000
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 3.09,
|
| 534 |
+
"learning_rate": 3.45524059627713e-05,
|
| 535 |
+
"loss": 0.8054,
|
| 536 |
+
"step": 440000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 3.12,
|
| 540 |
+
"learning_rate": 3.437686512143916e-05,
|
| 541 |
+
"loss": 0.8156,
|
| 542 |
+
"step": 445000
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 3.16,
|
| 546 |
+
"learning_rate": 3.420132428010701e-05,
|
| 547 |
+
"loss": 0.8032,
|
| 548 |
+
"step": 450000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 3.19,
|
| 552 |
+
"learning_rate": 3.4025783438774866e-05,
|
| 553 |
+
"loss": 0.8109,
|
| 554 |
+
"step": 455000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 3.23,
|
| 558 |
+
"learning_rate": 3.385024259744272e-05,
|
| 559 |
+
"loss": 0.812,
|
| 560 |
+
"step": 460000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 3.27,
|
| 564 |
+
"learning_rate": 3.3674701756110575e-05,
|
| 565 |
+
"loss": 0.8102,
|
| 566 |
+
"step": 465000
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 3.3,
|
| 570 |
+
"learning_rate": 3.3499160914778436e-05,
|
| 571 |
+
"loss": 0.8032,
|
| 572 |
+
"step": 470000
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 3.34,
|
| 576 |
+
"learning_rate": 3.332362007344629e-05,
|
| 577 |
+
"loss": 0.8098,
|
| 578 |
+
"step": 475000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 3.37,
|
| 582 |
+
"learning_rate": 3.3148079232114144e-05,
|
| 583 |
+
"loss": 0.8057,
|
| 584 |
+
"step": 480000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 3.41,
|
| 588 |
+
"learning_rate": 3.2972538390782e-05,
|
| 589 |
+
"loss": 0.8026,
|
| 590 |
+
"step": 485000
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.44,
|
| 594 |
+
"learning_rate": 3.279699754944985e-05,
|
| 595 |
+
"loss": 0.8039,
|
| 596 |
+
"step": 490000
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 3.48,
|
| 600 |
+
"learning_rate": 3.262145670811771e-05,
|
| 601 |
+
"loss": 0.8024,
|
| 602 |
+
"step": 495000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 3.51,
|
| 606 |
+
"learning_rate": 3.244591586678557e-05,
|
| 607 |
+
"loss": 0.8018,
|
| 608 |
+
"step": 500000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 3.55,
|
| 612 |
+
"learning_rate": 3.227037502545343e-05,
|
| 613 |
+
"loss": 0.7953,
|
| 614 |
+
"step": 505000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 3.58,
|
| 618 |
+
"learning_rate": 3.209483418412128e-05,
|
| 619 |
+
"loss": 0.7895,
|
| 620 |
+
"step": 510000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 3.62,
|
| 624 |
+
"learning_rate": 3.191929334278914e-05,
|
| 625 |
+
"loss": 0.7878,
|
| 626 |
+
"step": 515000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 3.65,
|
| 630 |
+
"learning_rate": 3.174375250145699e-05,
|
| 631 |
+
"loss": 0.794,
|
| 632 |
+
"step": 520000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.69,
|
| 636 |
+
"learning_rate": 3.1568211660124845e-05,
|
| 637 |
+
"loss": 0.7998,
|
| 638 |
+
"step": 525000
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.72,
|
| 642 |
+
"learning_rate": 3.13926708187927e-05,
|
| 643 |
+
"loss": 0.7819,
|
| 644 |
+
"step": 530000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 3.76,
|
| 648 |
+
"learning_rate": 3.121712997746056e-05,
|
| 649 |
+
"loss": 0.7917,
|
| 650 |
+
"step": 535000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 3.79,
|
| 654 |
+
"learning_rate": 3.1041589136128414e-05,
|
| 655 |
+
"loss": 0.7827,
|
| 656 |
+
"step": 540000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 3.83,
|
| 660 |
+
"learning_rate": 3.086604829479627e-05,
|
| 661 |
+
"loss": 0.7874,
|
| 662 |
+
"step": 545000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 3.86,
|
| 666 |
+
"learning_rate": 3.069050745346412e-05,
|
| 667 |
+
"loss": 0.7788,
|
| 668 |
+
"step": 550000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 3.9,
|
| 672 |
+
"learning_rate": 3.051496661213198e-05,
|
| 673 |
+
"loss": 0.7821,
|
| 674 |
+
"step": 555000
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 3.93,
|
| 678 |
+
"learning_rate": 3.0339425770799834e-05,
|
| 679 |
+
"loss": 0.7828,
|
| 680 |
+
"step": 560000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 3.97,
|
| 684 |
+
"learning_rate": 3.016388492946769e-05,
|
| 685 |
+
"loss": 0.7769,
|
| 686 |
+
"step": 565000
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 4.0,
|
| 690 |
+
"learning_rate": 2.9988344088135546e-05,
|
| 691 |
+
"loss": 0.7779,
|
| 692 |
+
"step": 570000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 4.04,
|
| 696 |
+
"learning_rate": 2.98128032468034e-05,
|
| 697 |
+
"loss": 0.7633,
|
| 698 |
+
"step": 575000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 4.07,
|
| 702 |
+
"learning_rate": 2.963726240547126e-05,
|
| 703 |
+
"loss": 0.765,
|
| 704 |
+
"step": 580000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 4.11,
|
| 708 |
+
"learning_rate": 2.9461721564139115e-05,
|
| 709 |
+
"loss": 0.7614,
|
| 710 |
+
"step": 585000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 4.14,
|
| 714 |
+
"learning_rate": 2.9286180722806973e-05,
|
| 715 |
+
"loss": 0.7669,
|
| 716 |
+
"step": 590000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 4.18,
|
| 720 |
+
"learning_rate": 2.9110639881474827e-05,
|
| 721 |
+
"loss": 0.7721,
|
| 722 |
+
"step": 595000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 4.21,
|
| 726 |
+
"learning_rate": 2.893509904014268e-05,
|
| 727 |
+
"loss": 0.7633,
|
| 728 |
+
"step": 600000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 4.25,
|
| 732 |
+
"learning_rate": 2.875955819881054e-05,
|
| 733 |
+
"loss": 0.7618,
|
| 734 |
+
"step": 605000
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 4.28,
|
| 738 |
+
"learning_rate": 2.8584017357478393e-05,
|
| 739 |
+
"loss": 0.7603,
|
| 740 |
+
"step": 610000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 4.32,
|
| 744 |
+
"learning_rate": 2.8408476516146247e-05,
|
| 745 |
+
"loss": 0.7594,
|
| 746 |
+
"step": 615000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 4.35,
|
| 750 |
+
"learning_rate": 2.8232935674814104e-05,
|
| 751 |
+
"loss": 0.7612,
|
| 752 |
+
"step": 620000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.39,
|
| 756 |
+
"learning_rate": 2.805739483348196e-05,
|
| 757 |
+
"loss": 0.7616,
|
| 758 |
+
"step": 625000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.42,
|
| 762 |
+
"learning_rate": 2.7881853992149816e-05,
|
| 763 |
+
"loss": 0.7628,
|
| 764 |
+
"step": 630000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 4.46,
|
| 768 |
+
"learning_rate": 2.770631315081767e-05,
|
| 769 |
+
"loss": 0.7638,
|
| 770 |
+
"step": 635000
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 4.49,
|
| 774 |
+
"learning_rate": 2.7530772309485525e-05,
|
| 775 |
+
"loss": 0.7469,
|
| 776 |
+
"step": 640000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 4.53,
|
| 780 |
+
"learning_rate": 2.7355231468153382e-05,
|
| 781 |
+
"loss": 0.7477,
|
| 782 |
+
"step": 645000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 4.56,
|
| 786 |
+
"learning_rate": 2.7179690626821236e-05,
|
| 787 |
+
"loss": 0.7501,
|
| 788 |
+
"step": 650000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 4.6,
|
| 792 |
+
"learning_rate": 2.700414978548909e-05,
|
| 793 |
+
"loss": 0.7513,
|
| 794 |
+
"step": 655000
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.63,
|
| 798 |
+
"learning_rate": 2.6828608944156948e-05,
|
| 799 |
+
"loss": 0.751,
|
| 800 |
+
"step": 660000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.67,
|
| 804 |
+
"learning_rate": 2.6653068102824802e-05,
|
| 805 |
+
"loss": 0.7491,
|
| 806 |
+
"step": 665000
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.7,
|
| 810 |
+
"learning_rate": 2.6477527261492663e-05,
|
| 811 |
+
"loss": 0.7516,
|
| 812 |
+
"step": 670000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 4.74,
|
| 816 |
+
"learning_rate": 2.6301986420160517e-05,
|
| 817 |
+
"loss": 0.7511,
|
| 818 |
+
"step": 675000
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 4.74,
|
| 822 |
+
"eval_loss": 0.6953830122947693,
|
| 823 |
+
"eval_runtime": 12.1778,
|
| 824 |
+
"eval_samples_per_second": 82.116,
|
| 825 |
+
"eval_steps_per_second": 10.265,
|
| 826 |
+
"step": 675000
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 4.77,
|
| 830 |
+
"learning_rate": 2.6126445578828375e-05,
|
| 831 |
+
"loss": 0.7395,
|
| 832 |
+
"step": 680000
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 4.77,
|
| 836 |
+
"eval_loss": 0.6852219104766846,
|
| 837 |
+
"eval_runtime": 12.2181,
|
| 838 |
+
"eval_samples_per_second": 81.846,
|
| 839 |
+
"eval_steps_per_second": 10.231,
|
| 840 |
+
"step": 680000
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 4.81,
|
| 844 |
+
"learning_rate": 2.595090473749623e-05,
|
| 845 |
+
"loss": 0.7445,
|
| 846 |
+
"step": 685000
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 4.81,
|
| 850 |
+
"eval_loss": 0.7140026688575745,
|
| 851 |
+
"eval_runtime": 12.2246,
|
| 852 |
+
"eval_samples_per_second": 81.802,
|
| 853 |
+
"eval_steps_per_second": 10.225,
|
| 854 |
+
"step": 685000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 4.84,
|
| 858 |
+
"learning_rate": 2.5775363896164083e-05,
|
| 859 |
+
"loss": 0.7448,
|
| 860 |
+
"step": 690000
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 4.84,
|
| 864 |
+
"eval_loss": 0.678001880645752,
|
| 865 |
+
"eval_runtime": 12.1996,
|
| 866 |
+
"eval_samples_per_second": 81.97,
|
| 867 |
+
"eval_steps_per_second": 10.246,
|
| 868 |
+
"step": 690000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 4.88,
|
| 872 |
+
"learning_rate": 2.559982305483194e-05,
|
| 873 |
+
"loss": 0.7392,
|
| 874 |
+
"step": 695000
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 4.88,
|
| 878 |
+
"eval_loss": 0.6525120139122009,
|
| 879 |
+
"eval_runtime": 12.2116,
|
| 880 |
+
"eval_samples_per_second": 81.889,
|
| 881 |
+
"eval_steps_per_second": 10.236,
|
| 882 |
+
"step": 695000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 4.92,
|
| 886 |
+
"learning_rate": 2.5424282213499795e-05,
|
| 887 |
+
"loss": 0.7499,
|
| 888 |
+
"step": 700000
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 4.92,
|
| 892 |
+
"eval_loss": 0.6543171405792236,
|
| 893 |
+
"eval_runtime": 12.2207,
|
| 894 |
+
"eval_samples_per_second": 81.828,
|
| 895 |
+
"eval_steps_per_second": 10.229,
|
| 896 |
+
"step": 700000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 4.95,
|
| 900 |
+
"learning_rate": 2.524874137216765e-05,
|
| 901 |
+
"loss": 0.7393,
|
| 902 |
+
"step": 705000
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 4.95,
|
| 906 |
+
"eval_loss": 0.665242075920105,
|
| 907 |
+
"eval_runtime": 5.3758,
|
| 908 |
+
"eval_samples_per_second": 186.02,
|
| 909 |
+
"eval_steps_per_second": 23.253,
|
| 910 |
+
"step": 705000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 4.99,
|
| 914 |
+
"learning_rate": 2.5073200530835506e-05,
|
| 915 |
+
"loss": 0.7324,
|
| 916 |
+
"step": 710000
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 4.99,
|
| 920 |
+
"eval_loss": 0.6618428826332092,
|
| 921 |
+
"eval_runtime": 5.3906,
|
| 922 |
+
"eval_samples_per_second": 185.507,
|
| 923 |
+
"eval_steps_per_second": 23.188,
|
| 924 |
+
"step": 710000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 5.02,
|
| 928 |
+
"learning_rate": 2.489765968950336e-05,
|
| 929 |
+
"loss": 0.7261,
|
| 930 |
+
"step": 715000
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 5.02,
|
| 934 |
+
"eval_loss": 0.6423526406288147,
|
| 935 |
+
"eval_runtime": 5.3892,
|
| 936 |
+
"eval_samples_per_second": 185.555,
|
| 937 |
+
"eval_steps_per_second": 23.194,
|
| 938 |
+
"step": 715000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 5.06,
|
| 942 |
+
"learning_rate": 2.4722118848171215e-05,
|
| 943 |
+
"loss": 0.7327,
|
| 944 |
+
"step": 720000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 5.06,
|
| 948 |
+
"eval_loss": 0.6585870385169983,
|
| 949 |
+
"eval_runtime": 5.3853,
|
| 950 |
+
"eval_samples_per_second": 185.69,
|
| 951 |
+
"eval_steps_per_second": 23.211,
|
| 952 |
+
"step": 720000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 5.09,
|
| 956 |
+
"learning_rate": 2.4546578006839072e-05,
|
| 957 |
+
"loss": 0.7265,
|
| 958 |
+
"step": 725000
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 5.09,
|
| 962 |
+
"eval_loss": 0.6187921762466431,
|
| 963 |
+
"eval_runtime": 5.3825,
|
| 964 |
+
"eval_samples_per_second": 185.787,
|
| 965 |
+
"eval_steps_per_second": 23.223,
|
| 966 |
+
"step": 725000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 5.13,
|
| 970 |
+
"learning_rate": 2.4371037165506926e-05,
|
| 971 |
+
"loss": 0.7247,
|
| 972 |
+
"step": 730000
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 5.13,
|
| 976 |
+
"eval_loss": 0.6582339406013489,
|
| 977 |
+
"eval_runtime": 5.3823,
|
| 978 |
+
"eval_samples_per_second": 185.796,
|
| 979 |
+
"eval_steps_per_second": 23.224,
|
| 980 |
+
"step": 730000
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 5.16,
|
| 984 |
+
"learning_rate": 2.419549632417478e-05,
|
| 985 |
+
"loss": 0.7265,
|
| 986 |
+
"step": 735000
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 5.16,
|
| 990 |
+
"eval_loss": 0.7226254940032959,
|
| 991 |
+
"eval_runtime": 5.3797,
|
| 992 |
+
"eval_samples_per_second": 185.883,
|
| 993 |
+
"eval_steps_per_second": 23.235,
|
| 994 |
+
"step": 735000
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 5.2,
|
| 998 |
+
"learning_rate": 2.401995548284264e-05,
|
| 999 |
+
"loss": 0.7166,
|
| 1000 |
+
"step": 740000
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 5.2,
|
| 1004 |
+
"eval_loss": 0.6698991656303406,
|
| 1005 |
+
"eval_runtime": 5.3768,
|
| 1006 |
+
"eval_samples_per_second": 185.986,
|
| 1007 |
+
"eval_steps_per_second": 23.248,
|
| 1008 |
+
"step": 740000
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 5.23,
|
| 1012 |
+
"learning_rate": 2.3844414641510496e-05,
|
| 1013 |
+
"loss": 0.7214,
|
| 1014 |
+
"step": 745000
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 5.23,
|
| 1018 |
+
"eval_loss": 0.6653444170951843,
|
| 1019 |
+
"eval_runtime": 5.3777,
|
| 1020 |
+
"eval_samples_per_second": 185.954,
|
| 1021 |
+
"eval_steps_per_second": 23.244,
|
| 1022 |
+
"step": 745000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 5.27,
|
| 1026 |
+
"learning_rate": 2.366887380017835e-05,
|
| 1027 |
+
"loss": 0.7268,
|
| 1028 |
+
"step": 750000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 5.27,
|
| 1032 |
+
"eval_loss": 0.6490678787231445,
|
| 1033 |
+
"eval_runtime": 5.3962,
|
| 1034 |
+
"eval_samples_per_second": 185.314,
|
| 1035 |
+
"eval_steps_per_second": 23.164,
|
| 1036 |
+
"step": 750000
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 5.3,
|
| 1040 |
+
"learning_rate": 2.3493332958846207e-05,
|
| 1041 |
+
"loss": 0.7177,
|
| 1042 |
+
"step": 755000
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 5.3,
|
| 1046 |
+
"eval_loss": 0.6720253825187683,
|
| 1047 |
+
"eval_runtime": 5.3813,
|
| 1048 |
+
"eval_samples_per_second": 185.828,
|
| 1049 |
+
"eval_steps_per_second": 23.229,
|
| 1050 |
+
"step": 755000
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 5.34,
|
| 1054 |
+
"learning_rate": 2.331779211751406e-05,
|
| 1055 |
+
"loss": 0.7173,
|
| 1056 |
+
"step": 760000
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 5.34,
|
| 1060 |
+
"eval_loss": 0.636309027671814,
|
| 1061 |
+
"eval_runtime": 5.3741,
|
| 1062 |
+
"eval_samples_per_second": 186.079,
|
| 1063 |
+
"eval_steps_per_second": 23.26,
|
| 1064 |
+
"step": 760000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 5.37,
|
| 1068 |
+
"learning_rate": 2.3142251276181916e-05,
|
| 1069 |
+
"loss": 0.7222,
|
| 1070 |
+
"step": 765000
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 5.37,
|
| 1074 |
+
"eval_loss": 0.6736326813697815,
|
| 1075 |
+
"eval_runtime": 5.3844,
|
| 1076 |
+
"eval_samples_per_second": 185.723,
|
| 1077 |
+
"eval_steps_per_second": 23.215,
|
| 1078 |
+
"step": 765000
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 5.41,
|
| 1082 |
+
"learning_rate": 2.2966710434849773e-05,
|
| 1083 |
+
"loss": 0.7189,
|
| 1084 |
+
"step": 770000
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 5.41,
|
| 1088 |
+
"eval_loss": 0.6502253413200378,
|
| 1089 |
+
"eval_runtime": 5.3808,
|
| 1090 |
+
"eval_samples_per_second": 185.846,
|
| 1091 |
+
"eval_steps_per_second": 23.231,
|
| 1092 |
+
"step": 770000
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 5.44,
|
| 1096 |
+
"learning_rate": 2.2791169593517627e-05,
|
| 1097 |
+
"loss": 0.7142,
|
| 1098 |
+
"step": 775000
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 5.44,
|
| 1102 |
+
"eval_loss": 0.6675522327423096,
|
| 1103 |
+
"eval_runtime": 5.3769,
|
| 1104 |
+
"eval_samples_per_second": 185.982,
|
| 1105 |
+
"eval_steps_per_second": 23.248,
|
| 1106 |
+
"step": 775000
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 5.48,
|
| 1110 |
+
"learning_rate": 2.2615628752185485e-05,
|
| 1111 |
+
"loss": 0.7123,
|
| 1112 |
+
"step": 780000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 5.48,
|
| 1116 |
+
"eval_loss": 0.7307547330856323,
|
| 1117 |
+
"eval_runtime": 5.3752,
|
| 1118 |
+
"eval_samples_per_second": 186.041,
|
| 1119 |
+
"eval_steps_per_second": 23.255,
|
| 1120 |
+
"step": 780000
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 5.51,
|
| 1124 |
+
"learning_rate": 2.2440087910853343e-05,
|
| 1125 |
+
"loss": 0.7149,
|
| 1126 |
+
"step": 785000
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 5.51,
|
| 1130 |
+
"eval_loss": 0.6528046727180481,
|
| 1131 |
+
"eval_runtime": 5.3874,
|
| 1132 |
+
"eval_samples_per_second": 185.618,
|
| 1133 |
+
"eval_steps_per_second": 23.202,
|
| 1134 |
+
"step": 785000
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 5.55,
|
| 1138 |
+
"learning_rate": 2.2264547069521197e-05,
|
| 1139 |
+
"loss": 0.7111,
|
| 1140 |
+
"step": 790000
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 5.55,
|
| 1144 |
+
"eval_loss": 0.6415424942970276,
|
| 1145 |
+
"eval_runtime": 5.3848,
|
| 1146 |
+
"eval_samples_per_second": 185.708,
|
| 1147 |
+
"eval_steps_per_second": 23.213,
|
| 1148 |
+
"step": 790000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 5.58,
|
| 1152 |
+
"learning_rate": 2.208900622818905e-05,
|
| 1153 |
+
"loss": 0.7126,
|
| 1154 |
+
"step": 795000
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 5.58,
|
| 1158 |
+
"eval_loss": 0.664243221282959,
|
| 1159 |
+
"eval_runtime": 5.3818,
|
| 1160 |
+
"eval_samples_per_second": 185.812,
|
| 1161 |
+
"eval_steps_per_second": 23.226,
|
| 1162 |
+
"step": 795000
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 5.62,
|
| 1166 |
+
"learning_rate": 2.191346538685691e-05,
|
| 1167 |
+
"loss": 0.7075,
|
| 1168 |
+
"step": 800000
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 5.62,
|
| 1172 |
+
"eval_loss": 0.6190058588981628,
|
| 1173 |
+
"eval_runtime": 5.3768,
|
| 1174 |
+
"eval_samples_per_second": 185.985,
|
| 1175 |
+
"eval_steps_per_second": 23.248,
|
| 1176 |
+
"step": 800000
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 5.65,
|
| 1180 |
+
"learning_rate": 2.1737924545524763e-05,
|
| 1181 |
+
"loss": 0.7047,
|
| 1182 |
+
"step": 805000
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 5.65,
|
| 1186 |
+
"eval_loss": 0.645745038986206,
|
| 1187 |
+
"eval_runtime": 5.3814,
|
| 1188 |
+
"eval_samples_per_second": 185.824,
|
| 1189 |
+
"eval_steps_per_second": 23.228,
|
| 1190 |
+
"step": 805000
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 5.69,
|
| 1194 |
+
"learning_rate": 2.1562383704192617e-05,
|
| 1195 |
+
"loss": 0.7002,
|
| 1196 |
+
"step": 810000
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 5.69,
|
| 1200 |
+
"eval_loss": 0.6156101226806641,
|
| 1201 |
+
"eval_runtime": 5.3698,
|
| 1202 |
+
"eval_samples_per_second": 186.226,
|
| 1203 |
+
"eval_steps_per_second": 23.278,
|
| 1204 |
+
"step": 810000
|
| 1205 |
+
}
|
| 1206 |
+
],
|
| 1207 |
+
"max_steps": 1424170,
|
| 1208 |
+
"num_train_epochs": 10,
|
| 1209 |
+
"total_flos": 3.3150958919481446e+17,
|
| 1210 |
+
"trial_name": null,
|
| 1211 |
+
"trial_params": null
|
| 1212 |
+
}
|