Upload log_history.json with huggingface_hub
Browse files- log_history.json +1301 -0
log_history.json
ADDED
|
@@ -0,0 +1,1301 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"loss": 0.2659,
|
| 4 |
+
"grad_norm": 3.4286012649536133,
|
| 5 |
+
"learning_rate": 1.9715261958997724e-05,
|
| 6 |
+
"epoch": 0.2847380410022779,
|
| 7 |
+
"step": 500
|
| 8 |
+
},
|
| 9 |
+
{
|
| 10 |
+
"loss": 0.1081,
|
| 11 |
+
"grad_norm": 0.7319045066833496,
|
| 12 |
+
"learning_rate": 1.9430523917995446e-05,
|
| 13 |
+
"epoch": 0.5694760820045558,
|
| 14 |
+
"step": 1000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"loss": 0.0777,
|
| 18 |
+
"grad_norm": 3.7819042205810547,
|
| 19 |
+
"learning_rate": 1.9145785876993168e-05,
|
| 20 |
+
"epoch": 0.8542141230068337,
|
| 21 |
+
"step": 1500
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"eval_loss": 0.07203580439090729,
|
| 25 |
+
"eval_overall": {
|
| 26 |
+
"precision": 0.8912693298969072,
|
| 27 |
+
"recall": 0.931167956916863,
|
| 28 |
+
"f1": 0.9107818930041152,
|
| 29 |
+
"accuracy": 0.9814269735680226
|
| 30 |
+
},
|
| 31 |
+
"eval_per_label": {
|
| 32 |
+
"LOC": {
|
| 33 |
+
"precision": 0.9239187076602398,
|
| 34 |
+
"recall": 0.965160587915079,
|
| 35 |
+
"f1": 0.9440894568690097,
|
| 36 |
+
"number": 1837
|
| 37 |
+
},
|
| 38 |
+
"MISC": {
|
| 39 |
+
"precision": 0.7727272727272727,
|
| 40 |
+
"recall": 0.8297180043383948,
|
| 41 |
+
"f1": 0.8002092050209204,
|
| 42 |
+
"number": 922
|
| 43 |
+
},
|
| 44 |
+
"ORG": {
|
| 45 |
+
"precision": 0.8624823695345557,
|
| 46 |
+
"recall": 0.9120059656972409,
|
| 47 |
+
"f1": 0.8865530989488946,
|
| 48 |
+
"number": 1341
|
| 49 |
+
},
|
| 50 |
+
"PER": {
|
| 51 |
+
"precision": 0.9420520999468368,
|
| 52 |
+
"recall": 0.9619978284473398,
|
| 53 |
+
"f1": 0.9519204942250873,
|
| 54 |
+
"number": 1842
|
| 55 |
+
}
|
| 56 |
+
},
|
| 57 |
+
"eval_runtime": 7.4533,
|
| 58 |
+
"eval_samples_per_second": 436.049,
|
| 59 |
+
"eval_steps_per_second": 54.607,
|
| 60 |
+
"epoch": 1.0,
|
| 61 |
+
"step": 1756
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"loss": 0.0631,
|
| 65 |
+
"grad_norm": 0.11871737241744995,
|
| 66 |
+
"learning_rate": 1.886104783599089e-05,
|
| 67 |
+
"epoch": 1.1389521640091116,
|
| 68 |
+
"step": 2000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"loss": 0.0464,
|
| 72 |
+
"grad_norm": 0.2318667322397232,
|
| 73 |
+
"learning_rate": 1.8576309794988612e-05,
|
| 74 |
+
"epoch": 1.4236902050113895,
|
| 75 |
+
"step": 2500
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"loss": 0.0442,
|
| 79 |
+
"grad_norm": 0.35079020261764526,
|
| 80 |
+
"learning_rate": 1.8291571753986334e-05,
|
| 81 |
+
"epoch": 1.7084282460136673,
|
| 82 |
+
"step": 3000
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"loss": 0.0411,
|
| 86 |
+
"grad_norm": 0.06829982250928879,
|
| 87 |
+
"learning_rate": 1.8006833712984056e-05,
|
| 88 |
+
"epoch": 1.9931662870159452,
|
| 89 |
+
"step": 3500
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"eval_loss": 0.07787470519542694,
|
| 93 |
+
"eval_overall": {
|
| 94 |
+
"precision": 0.9304723885562209,
|
| 95 |
+
"recall": 0.9414338606529788,
|
| 96 |
+
"f1": 0.9359210306173666,
|
| 97 |
+
"accuracy": 0.9842968152116324
|
| 98 |
+
},
|
| 99 |
+
"eval_per_label": {
|
| 100 |
+
"LOC": {
|
| 101 |
+
"precision": 0.9511015583019882,
|
| 102 |
+
"recall": 0.9635274904735982,
|
| 103 |
+
"f1": 0.9572742022714981,
|
| 104 |
+
"number": 1837
|
| 105 |
+
},
|
| 106 |
+
"MISC": {
|
| 107 |
+
"precision": 0.8725910064239829,
|
| 108 |
+
"recall": 0.8839479392624728,
|
| 109 |
+
"f1": 0.8782327586206896,
|
| 110 |
+
"number": 922
|
| 111 |
+
},
|
| 112 |
+
"ORG": {
|
| 113 |
+
"precision": 0.9050822122571002,
|
| 114 |
+
"recall": 0.9030574198359433,
|
| 115 |
+
"f1": 0.9040686823441583,
|
| 116 |
+
"number": 1341
|
| 117 |
+
},
|
| 118 |
+
"PER": {
|
| 119 |
+
"precision": 0.9568919638105375,
|
| 120 |
+
"recall": 0.9761129207383279,
|
| 121 |
+
"f1": 0.9664068798710024,
|
| 122 |
+
"number": 1842
|
| 123 |
+
}
|
| 124 |
+
},
|
| 125 |
+
"eval_runtime": 7.5011,
|
| 126 |
+
"eval_samples_per_second": 433.267,
|
| 127 |
+
"eval_steps_per_second": 54.258,
|
| 128 |
+
"epoch": 2.0,
|
| 129 |
+
"step": 3512
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"loss": 0.0258,
|
| 133 |
+
"grad_norm": 4.330985069274902,
|
| 134 |
+
"learning_rate": 1.7722095671981778e-05,
|
| 135 |
+
"epoch": 2.277904328018223,
|
| 136 |
+
"step": 4000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"loss": 0.0284,
|
| 140 |
+
"grad_norm": 0.06637139618396759,
|
| 141 |
+
"learning_rate": 1.74373576309795e-05,
|
| 142 |
+
"epoch": 2.562642369020501,
|
| 143 |
+
"step": 4500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"loss": 0.0269,
|
| 147 |
+
"grad_norm": 11.263030052185059,
|
| 148 |
+
"learning_rate": 1.7152619589977222e-05,
|
| 149 |
+
"epoch": 2.847380410022779,
|
| 150 |
+
"step": 5000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"eval_loss": 0.07202505320310593,
|
| 154 |
+
"eval_overall": {
|
| 155 |
+
"precision": 0.9318106587222774,
|
| 156 |
+
"recall": 0.9474924267923258,
|
| 157 |
+
"f1": 0.9395861148197597,
|
| 158 |
+
"accuracy": 0.9857685288750221
|
| 159 |
+
},
|
| 160 |
+
"eval_per_label": {
|
| 161 |
+
"LOC": {
|
| 162 |
+
"precision": 0.9626218851570965,
|
| 163 |
+
"recall": 0.9673380511703865,
|
| 164 |
+
"f1": 0.9649742058104807,
|
| 165 |
+
"number": 1837
|
| 166 |
+
},
|
| 167 |
+
"MISC": {
|
| 168 |
+
"precision": 0.8943355119825708,
|
| 169 |
+
"recall": 0.8904555314533622,
|
| 170 |
+
"f1": 0.8923913043478261,
|
| 171 |
+
"number": 922
|
| 172 |
+
},
|
| 173 |
+
"ORG": {
|
| 174 |
+
"precision": 0.8772791023842917,
|
| 175 |
+
"recall": 0.9328859060402684,
|
| 176 |
+
"f1": 0.9042284062161186,
|
| 177 |
+
"number": 1341
|
| 178 |
+
},
|
| 179 |
+
"PER": {
|
| 180 |
+
"precision": 0.9616630669546437,
|
| 181 |
+
"recall": 0.9668838219326819,
|
| 182 |
+
"f1": 0.9642663779101246,
|
| 183 |
+
"number": 1842
|
| 184 |
+
}
|
| 185 |
+
},
|
| 186 |
+
"eval_runtime": 6.5043,
|
| 187 |
+
"eval_samples_per_second": 499.669,
|
| 188 |
+
"eval_steps_per_second": 62.574,
|
| 189 |
+
"epoch": 3.0,
|
| 190 |
+
"step": 5268
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"loss": 0.0226,
|
| 194 |
+
"grad_norm": 0.7242124676704407,
|
| 195 |
+
"learning_rate": 1.6867881548974945e-05,
|
| 196 |
+
"epoch": 3.132118451025057,
|
| 197 |
+
"step": 5500
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"loss": 0.0165,
|
| 201 |
+
"grad_norm": 0.006902824155986309,
|
| 202 |
+
"learning_rate": 1.6583143507972667e-05,
|
| 203 |
+
"epoch": 3.416856492027335,
|
| 204 |
+
"step": 6000
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"loss": 0.0176,
|
| 208 |
+
"grad_norm": 0.031127002090215683,
|
| 209 |
+
"learning_rate": 1.629840546697039e-05,
|
| 210 |
+
"epoch": 3.7015945330296125,
|
| 211 |
+
"step": 6500
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"loss": 0.0196,
|
| 215 |
+
"grad_norm": 0.0038718737196177244,
|
| 216 |
+
"learning_rate": 1.601366742596811e-05,
|
| 217 |
+
"epoch": 3.9863325740318905,
|
| 218 |
+
"step": 7000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"eval_loss": 0.08569859713315964,
|
| 222 |
+
"eval_overall": {
|
| 223 |
+
"precision": 0.930635838150289,
|
| 224 |
+
"recall": 0.9483338943116796,
|
| 225 |
+
"f1": 0.9394015170459282,
|
| 226 |
+
"accuracy": 0.9850621063165951
|
| 227 |
+
},
|
| 228 |
+
"eval_per_label": {
|
| 229 |
+
"LOC": {
|
| 230 |
+
"precision": 0.9533011272141707,
|
| 231 |
+
"recall": 0.9667936853565596,
|
| 232 |
+
"f1": 0.96,
|
| 233 |
+
"number": 1837
|
| 234 |
+
},
|
| 235 |
+
"MISC": {
|
| 236 |
+
"precision": 0.8623188405797102,
|
| 237 |
+
"recall": 0.903470715835141,
|
| 238 |
+
"f1": 0.8824152542372882,
|
| 239 |
+
"number": 922
|
| 240 |
+
},
|
| 241 |
+
"ORG": {
|
| 242 |
+
"precision": 0.8995664739884393,
|
| 243 |
+
"recall": 0.9284116331096197,
|
| 244 |
+
"f1": 0.9137614678899083,
|
| 245 |
+
"number": 1341
|
| 246 |
+
},
|
| 247 |
+
"PER": {
|
| 248 |
+
"precision": 0.9668838219326819,
|
| 249 |
+
"recall": 0.9668838219326819,
|
| 250 |
+
"f1": 0.9668838219326819,
|
| 251 |
+
"number": 1842
|
| 252 |
+
}
|
| 253 |
+
},
|
| 254 |
+
"eval_runtime": 7.5249,
|
| 255 |
+
"eval_samples_per_second": 431.899,
|
| 256 |
+
"eval_steps_per_second": 54.087,
|
| 257 |
+
"epoch": 4.0,
|
| 258 |
+
"step": 7024
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"loss": 0.0138,
|
| 262 |
+
"grad_norm": 0.17987537384033203,
|
| 263 |
+
"learning_rate": 1.5728929384965833e-05,
|
| 264 |
+
"epoch": 4.271070615034168,
|
| 265 |
+
"step": 7500
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"loss": 0.0134,
|
| 269 |
+
"grad_norm": 1.4385559558868408,
|
| 270 |
+
"learning_rate": 1.5444191343963555e-05,
|
| 271 |
+
"epoch": 4.555808656036446,
|
| 272 |
+
"step": 8000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"loss": 0.0139,
|
| 276 |
+
"grad_norm": 0.02356315404176712,
|
| 277 |
+
"learning_rate": 1.5159453302961277e-05,
|
| 278 |
+
"epoch": 4.840546697038724,
|
| 279 |
+
"step": 8500
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"eval_loss": 0.08417785912752151,
|
| 283 |
+
"eval_overall": {
|
| 284 |
+
"precision": 0.9310686015831134,
|
| 285 |
+
"recall": 0.9501851228542578,
|
| 286 |
+
"f1": 0.9405297351324339,
|
| 287 |
+
"accuracy": 0.9851062577264967
|
| 288 |
+
},
|
| 289 |
+
"eval_per_label": {
|
| 290 |
+
"LOC": {
|
| 291 |
+
"precision": 0.9627228525121556,
|
| 292 |
+
"recall": 0.9700598802395209,
|
| 293 |
+
"f1": 0.9663774403470715,
|
| 294 |
+
"number": 1837
|
| 295 |
+
},
|
| 296 |
+
"MISC": {
|
| 297 |
+
"precision": 0.8386454183266933,
|
| 298 |
+
"recall": 0.913232104121475,
|
| 299 |
+
"f1": 0.8743509865005192,
|
| 300 |
+
"number": 922
|
| 301 |
+
},
|
| 302 |
+
"ORG": {
|
| 303 |
+
"precision": 0.9070110701107011,
|
| 304 |
+
"recall": 0.9164802386278896,
|
| 305 |
+
"f1": 0.9117210682492581,
|
| 306 |
+
"number": 1341
|
| 307 |
+
},
|
| 308 |
+
"PER": {
|
| 309 |
+
"precision": 0.9670981661272924,
|
| 310 |
+
"recall": 0.9733984799131379,
|
| 311 |
+
"f1": 0.9702380952380952,
|
| 312 |
+
"number": 1842
|
| 313 |
+
}
|
| 314 |
+
},
|
| 315 |
+
"eval_runtime": 7.5038,
|
| 316 |
+
"eval_samples_per_second": 433.114,
|
| 317 |
+
"eval_steps_per_second": 54.239,
|
| 318 |
+
"epoch": 5.0,
|
| 319 |
+
"step": 8780
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"loss": 0.0092,
|
| 323 |
+
"grad_norm": 0.7477717399597168,
|
| 324 |
+
"learning_rate": 1.4874715261958999e-05,
|
| 325 |
+
"epoch": 5.125284738041002,
|
| 326 |
+
"step": 9000
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"loss": 0.006,
|
| 330 |
+
"grad_norm": 0.04322722181677818,
|
| 331 |
+
"learning_rate": 1.4589977220956721e-05,
|
| 332 |
+
"epoch": 5.41002277904328,
|
| 333 |
+
"step": 9500
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"loss": 0.0073,
|
| 337 |
+
"grad_norm": 0.051729559898376465,
|
| 338 |
+
"learning_rate": 1.4305239179954442e-05,
|
| 339 |
+
"epoch": 5.694760820045558,
|
| 340 |
+
"step": 10000
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"loss": 0.0079,
|
| 344 |
+
"grad_norm": 0.012189100496470928,
|
| 345 |
+
"learning_rate": 1.4020501138952165e-05,
|
| 346 |
+
"epoch": 5.979498861047836,
|
| 347 |
+
"step": 10500
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"eval_loss": 0.09725591540336609,
|
| 351 |
+
"eval_overall": {
|
| 352 |
+
"precision": 0.9291845493562232,
|
| 353 |
+
"recall": 0.9473241332884551,
|
| 354 |
+
"f1": 0.9381666666666667,
|
| 355 |
+
"accuracy": 0.9844734208512392
|
| 356 |
+
},
|
| 357 |
+
"eval_per_label": {
|
| 358 |
+
"LOC": {
|
| 359 |
+
"precision": 0.9546424759871932,
|
| 360 |
+
"recall": 0.9738704409363091,
|
| 361 |
+
"f1": 0.9641606036108865,
|
| 362 |
+
"number": 1837
|
| 363 |
+
},
|
| 364 |
+
"MISC": {
|
| 365 |
+
"precision": 0.858739837398374,
|
| 366 |
+
"recall": 0.9164859002169198,
|
| 367 |
+
"f1": 0.8866736621196223,
|
| 368 |
+
"number": 922
|
| 369 |
+
},
|
| 370 |
+
"ORG": {
|
| 371 |
+
"precision": 0.8921852387843705,
|
| 372 |
+
"recall": 0.9194630872483222,
|
| 373 |
+
"f1": 0.9056188027910393,
|
| 374 |
+
"number": 1341
|
| 375 |
+
},
|
| 376 |
+
"PER": {
|
| 377 |
+
"precision": 0.9691969196919692,
|
| 378 |
+
"recall": 0.9565689467969598,
|
| 379 |
+
"f1": 0.9628415300546448,
|
| 380 |
+
"number": 1842
|
| 381 |
+
}
|
| 382 |
+
},
|
| 383 |
+
"eval_runtime": 7.4822,
|
| 384 |
+
"eval_samples_per_second": 434.362,
|
| 385 |
+
"eval_steps_per_second": 54.395,
|
| 386 |
+
"epoch": 6.0,
|
| 387 |
+
"step": 10536
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"loss": 0.008,
|
| 391 |
+
"grad_norm": 0.0029869996942579746,
|
| 392 |
+
"learning_rate": 1.3735763097949887e-05,
|
| 393 |
+
"epoch": 6.264236902050114,
|
| 394 |
+
"step": 11000
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"loss": 0.0066,
|
| 398 |
+
"grad_norm": 0.018263721838593483,
|
| 399 |
+
"learning_rate": 1.3451025056947608e-05,
|
| 400 |
+
"epoch": 6.548974943052392,
|
| 401 |
+
"step": 11500
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"loss": 0.0057,
|
| 405 |
+
"grad_norm": 0.020874306559562683,
|
| 406 |
+
"learning_rate": 1.3166287015945332e-05,
|
| 407 |
+
"epoch": 6.83371298405467,
|
| 408 |
+
"step": 12000
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"eval_loss": 0.097112737596035,
|
| 412 |
+
"eval_overall": {
|
| 413 |
+
"precision": 0.9343584656084656,
|
| 414 |
+
"recall": 0.9510265903736116,
|
| 415 |
+
"f1": 0.9426188490408675,
|
| 416 |
+
"accuracy": 0.9859745687878966
|
| 417 |
+
},
|
| 418 |
+
"eval_per_label": {
|
| 419 |
+
"LOC": {
|
| 420 |
+
"precision": 0.9689373297002725,
|
| 421 |
+
"recall": 0.9678824169842134,
|
| 422 |
+
"f1": 0.968409586056645,
|
| 423 |
+
"number": 1837
|
| 424 |
+
},
|
| 425 |
+
"MISC": {
|
| 426 |
+
"precision": 0.8805031446540881,
|
| 427 |
+
"recall": 0.911062906724512,
|
| 428 |
+
"f1": 0.8955223880597015,
|
| 429 |
+
"number": 922
|
| 430 |
+
},
|
| 431 |
+
"ORG": {
|
| 432 |
+
"precision": 0.8851063829787233,
|
| 433 |
+
"recall": 0.930648769574944,
|
| 434 |
+
"f1": 0.9073064340239914,
|
| 435 |
+
"number": 1341
|
| 436 |
+
},
|
| 437 |
+
"PER": {
|
| 438 |
+
"precision": 0.965386695511087,
|
| 439 |
+
"recall": 0.9690553745928339,
|
| 440 |
+
"f1": 0.967217556217827,
|
| 441 |
+
"number": 1842
|
| 442 |
+
}
|
| 443 |
+
},
|
| 444 |
+
"eval_runtime": 7.4789,
|
| 445 |
+
"eval_samples_per_second": 434.555,
|
| 446 |
+
"eval_steps_per_second": 54.42,
|
| 447 |
+
"epoch": 7.0,
|
| 448 |
+
"step": 12292
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"loss": 0.0031,
|
| 452 |
+
"grad_norm": 0.0007487820694223046,
|
| 453 |
+
"learning_rate": 1.2881548974943054e-05,
|
| 454 |
+
"epoch": 7.118451025056948,
|
| 455 |
+
"step": 12500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"loss": 0.0037,
|
| 459 |
+
"grad_norm": 15.724783897399902,
|
| 460 |
+
"learning_rate": 1.2596810933940776e-05,
|
| 461 |
+
"epoch": 7.403189066059226,
|
| 462 |
+
"step": 13000
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"loss": 0.0061,
|
| 466 |
+
"grad_norm": 0.614613950252533,
|
| 467 |
+
"learning_rate": 1.2312072892938498e-05,
|
| 468 |
+
"epoch": 7.687927107061503,
|
| 469 |
+
"step": 13500
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"loss": 0.0036,
|
| 473 |
+
"grad_norm": 0.0012391641503199935,
|
| 474 |
+
"learning_rate": 1.2027334851936218e-05,
|
| 475 |
+
"epoch": 7.972665148063781,
|
| 476 |
+
"step": 14000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"eval_loss": 0.10294844955205917,
|
| 480 |
+
"eval_overall": {
|
| 481 |
+
"precision": 0.9285831285831286,
|
| 482 |
+
"recall": 0.9540558734432851,
|
| 483 |
+
"f1": 0.9411471735701834,
|
| 484 |
+
"accuracy": 0.9852387119562018
|
| 485 |
+
},
|
| 486 |
+
"eval_per_label": {
|
| 487 |
+
"LOC": {
|
| 488 |
+
"precision": 0.9664864864864865,
|
| 489 |
+
"recall": 0.9733260751224823,
|
| 490 |
+
"f1": 0.9698942229454841,
|
| 491 |
+
"number": 1837
|
| 492 |
+
},
|
| 493 |
+
"MISC": {
|
| 494 |
+
"precision": 0.8497983870967742,
|
| 495 |
+
"recall": 0.9143167028199566,
|
| 496 |
+
"f1": 0.8808777429467085,
|
| 497 |
+
"number": 922
|
| 498 |
+
},
|
| 499 |
+
"ORG": {
|
| 500 |
+
"precision": 0.9019607843137255,
|
| 501 |
+
"recall": 0.9261744966442953,
|
| 502 |
+
"f1": 0.9139072847682119,
|
| 503 |
+
"number": 1341
|
| 504 |
+
},
|
| 505 |
+
"PER": {
|
| 506 |
+
"precision": 0.9522799575821845,
|
| 507 |
+
"recall": 0.9750271444082519,
|
| 508 |
+
"f1": 0.9635193133047211,
|
| 509 |
+
"number": 1842
|
| 510 |
+
}
|
| 511 |
+
},
|
| 512 |
+
"eval_runtime": 7.6218,
|
| 513 |
+
"eval_samples_per_second": 426.41,
|
| 514 |
+
"eval_steps_per_second": 53.4,
|
| 515 |
+
"epoch": 8.0,
|
| 516 |
+
"step": 14048
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"loss": 0.0042,
|
| 520 |
+
"grad_norm": 0.15165293216705322,
|
| 521 |
+
"learning_rate": 1.1742596810933942e-05,
|
| 522 |
+
"epoch": 8.257403189066059,
|
| 523 |
+
"step": 14500
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"loss": 0.0038,
|
| 527 |
+
"grad_norm": 0.01020512543618679,
|
| 528 |
+
"learning_rate": 1.1457858769931664e-05,
|
| 529 |
+
"epoch": 8.542141230068337,
|
| 530 |
+
"step": 15000
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"loss": 0.0028,
|
| 534 |
+
"grad_norm": 0.00046127362293191254,
|
| 535 |
+
"learning_rate": 1.1173120728929384e-05,
|
| 536 |
+
"epoch": 8.826879271070615,
|
| 537 |
+
"step": 15500
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"eval_loss": 0.11473368108272552,
|
| 541 |
+
"eval_overall": {
|
| 542 |
+
"precision": 0.9368072787427626,
|
| 543 |
+
"recall": 0.9530461124200605,
|
| 544 |
+
"f1": 0.944856928339034,
|
| 545 |
+
"accuracy": 0.9860187201977983
|
| 546 |
+
},
|
| 547 |
+
"eval_per_label": {
|
| 548 |
+
"LOC": {
|
| 549 |
+
"precision": 0.9632034632034632,
|
| 550 |
+
"recall": 0.9689711486118672,
|
| 551 |
+
"f1": 0.966078697421981,
|
| 552 |
+
"number": 1837
|
| 553 |
+
},
|
| 554 |
+
"MISC": {
|
| 555 |
+
"precision": 0.8518145161290323,
|
| 556 |
+
"recall": 0.9164859002169198,
|
| 557 |
+
"f1": 0.8829676071055382,
|
| 558 |
+
"number": 922
|
| 559 |
+
},
|
| 560 |
+
"ORG": {
|
| 561 |
+
"precision": 0.9313207547169812,
|
| 562 |
+
"recall": 0.9202087994034303,
|
| 563 |
+
"f1": 0.9257314328582145,
|
| 564 |
+
"number": 1341
|
| 565 |
+
},
|
| 566 |
+
"PER": {
|
| 567 |
+
"precision": 0.9595744680851064,
|
| 568 |
+
"recall": 0.9793702497285559,
|
| 569 |
+
"f1": 0.969371305749597,
|
| 570 |
+
"number": 1842
|
| 571 |
+
}
|
| 572 |
+
},
|
| 573 |
+
"eval_runtime": 7.5249,
|
| 574 |
+
"eval_samples_per_second": 431.9,
|
| 575 |
+
"eval_steps_per_second": 54.087,
|
| 576 |
+
"epoch": 9.0,
|
| 577 |
+
"step": 15804
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"loss": 0.0049,
|
| 581 |
+
"grad_norm": 0.17741906642913818,
|
| 582 |
+
"learning_rate": 1.0888382687927108e-05,
|
| 583 |
+
"epoch": 9.111617312072893,
|
| 584 |
+
"step": 16000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"loss": 0.003,
|
| 588 |
+
"grad_norm": 0.0018998866435140371,
|
| 589 |
+
"learning_rate": 1.060364464692483e-05,
|
| 590 |
+
"epoch": 9.39635535307517,
|
| 591 |
+
"step": 16500
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"loss": 0.0042,
|
| 595 |
+
"grad_norm": 0.0021700740326195955,
|
| 596 |
+
"learning_rate": 1.0318906605922552e-05,
|
| 597 |
+
"epoch": 9.681093394077449,
|
| 598 |
+
"step": 17000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"loss": 0.0026,
|
| 602 |
+
"grad_norm": 0.014039441011846066,
|
| 603 |
+
"learning_rate": 1.0034168564920275e-05,
|
| 604 |
+
"epoch": 9.965831435079727,
|
| 605 |
+
"step": 17500
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"eval_loss": 0.1184961125254631,
|
| 609 |
+
"eval_overall": {
|
| 610 |
+
"precision": 0.9358889623265036,
|
| 611 |
+
"recall": 0.9532144059239314,
|
| 612 |
+
"f1": 0.9444722361180591,
|
| 613 |
+
"accuracy": 0.9852681462294696
|
| 614 |
+
},
|
| 615 |
+
"eval_per_label": {
|
| 616 |
+
"LOC": {
|
| 617 |
+
"precision": 0.9669197396963124,
|
| 618 |
+
"recall": 0.9706042460533478,
|
| 619 |
+
"f1": 0.9687584895408857,
|
| 620 |
+
"number": 1837
|
| 621 |
+
},
|
| 622 |
+
"MISC": {
|
| 623 |
+
"precision": 0.8826638477801269,
|
| 624 |
+
"recall": 0.9056399132321041,
|
| 625 |
+
"f1": 0.8940042826552462,
|
| 626 |
+
"number": 922
|
| 627 |
+
},
|
| 628 |
+
"ORG": {
|
| 629 |
+
"precision": 0.8961318051575932,
|
| 630 |
+
"recall": 0.9328859060402684,
|
| 631 |
+
"f1": 0.9141395688710267,
|
| 632 |
+
"number": 1341
|
| 633 |
+
},
|
| 634 |
+
"PER": {
|
| 635 |
+
"precision": 0.9619506966773848,
|
| 636 |
+
"recall": 0.9744842562432139,
|
| 637 |
+
"f1": 0.9681769147788565,
|
| 638 |
+
"number": 1842
|
| 639 |
+
}
|
| 640 |
+
},
|
| 641 |
+
"eval_runtime": 7.5803,
|
| 642 |
+
"eval_samples_per_second": 428.744,
|
| 643 |
+
"eval_steps_per_second": 53.692,
|
| 644 |
+
"epoch": 10.0,
|
| 645 |
+
"step": 17560
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"loss": 0.0011,
|
| 649 |
+
"grad_norm": 0.00040622701635584235,
|
| 650 |
+
"learning_rate": 9.749430523917997e-06,
|
| 651 |
+
"epoch": 10.250569476082005,
|
| 652 |
+
"step": 18000
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"loss": 0.001,
|
| 656 |
+
"grad_norm": 0.0004215097869746387,
|
| 657 |
+
"learning_rate": 9.464692482915719e-06,
|
| 658 |
+
"epoch": 10.535307517084282,
|
| 659 |
+
"step": 18500
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"loss": 0.002,
|
| 663 |
+
"grad_norm": 0.0020023963879793882,
|
| 664 |
+
"learning_rate": 9.17995444191344e-06,
|
| 665 |
+
"epoch": 10.82004555808656,
|
| 666 |
+
"step": 19000
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"eval_loss": 0.11947210878133774,
|
| 670 |
+
"eval_overall": {
|
| 671 |
+
"precision": 0.9384717168375786,
|
| 672 |
+
"recall": 0.9548973409626389,
|
| 673 |
+
"f1": 0.9466132799466134,
|
| 674 |
+
"accuracy": 0.9857685288750221
|
| 675 |
+
},
|
| 676 |
+
"eval_per_label": {
|
| 677 |
+
"LOC": {
|
| 678 |
+
"precision": 0.9658906334596643,
|
| 679 |
+
"recall": 0.9711486118671747,
|
| 680 |
+
"f1": 0.9685124864277959,
|
| 681 |
+
"number": 1837
|
| 682 |
+
},
|
| 683 |
+
"MISC": {
|
| 684 |
+
"precision": 0.8954108858057631,
|
| 685 |
+
"recall": 0.9099783080260304,
|
| 686 |
+
"f1": 0.9026358257127487,
|
| 687 |
+
"number": 922
|
| 688 |
+
},
|
| 689 |
+
"ORG": {
|
| 690 |
+
"precision": 0.9075812274368231,
|
| 691 |
+
"recall": 0.9373601789709173,
|
| 692 |
+
"f1": 0.922230374174615,
|
| 693 |
+
"number": 1341
|
| 694 |
+
},
|
| 695 |
+
"PER": {
|
| 696 |
+
"precision": 0.9557805007991476,
|
| 697 |
+
"recall": 0.9739413680781759,
|
| 698 |
+
"f1": 0.9647754772788384,
|
| 699 |
+
"number": 1842
|
| 700 |
+
}
|
| 701 |
+
},
|
| 702 |
+
"eval_runtime": 7.5556,
|
| 703 |
+
"eval_samples_per_second": 430.142,
|
| 704 |
+
"eval_steps_per_second": 53.867,
|
| 705 |
+
"epoch": 11.0,
|
| 706 |
+
"step": 19316
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"loss": 0.0028,
|
| 710 |
+
"grad_norm": 0.030722877010703087,
|
| 711 |
+
"learning_rate": 8.895216400911163e-06,
|
| 712 |
+
"epoch": 11.104783599088838,
|
| 713 |
+
"step": 19500
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"loss": 0.0033,
|
| 717 |
+
"grad_norm": 0.004747629631310701,
|
| 718 |
+
"learning_rate": 8.610478359908885e-06,
|
| 719 |
+
"epoch": 11.389521640091116,
|
| 720 |
+
"step": 20000
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"loss": 0.0018,
|
| 724 |
+
"grad_norm": 0.0014466517604887486,
|
| 725 |
+
"learning_rate": 8.325740318906607e-06,
|
| 726 |
+
"epoch": 11.674259681093394,
|
| 727 |
+
"step": 20500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"loss": 0.0004,
|
| 731 |
+
"grad_norm": 0.005812987219542265,
|
| 732 |
+
"learning_rate": 8.041002277904329e-06,
|
| 733 |
+
"epoch": 11.958997722095672,
|
| 734 |
+
"step": 21000
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"eval_loss": 0.12146918475627899,
|
| 738 |
+
"eval_overall": {
|
| 739 |
+
"precision": 0.9431799302209669,
|
| 740 |
+
"recall": 0.9554022214742511,
|
| 741 |
+
"f1": 0.9492517348047823,
|
| 742 |
+
"accuracy": 0.9866074056631542
|
| 743 |
+
},
|
| 744 |
+
"eval_per_label": {
|
| 745 |
+
"LOC": {
|
| 746 |
+
"precision": 0.971195652173913,
|
| 747 |
+
"recall": 0.9727817093086554,
|
| 748 |
+
"f1": 0.9719880337231438,
|
| 749 |
+
"number": 1837
|
| 750 |
+
},
|
| 751 |
+
"MISC": {
|
| 752 |
+
"precision": 0.9096844396082698,
|
| 753 |
+
"recall": 0.9067245119305857,
|
| 754 |
+
"f1": 0.9082020640956002,
|
| 755 |
+
"number": 922
|
| 756 |
+
},
|
| 757 |
+
"ORG": {
|
| 758 |
+
"precision": 0.9121522693997072,
|
| 759 |
+
"recall": 0.9291573452647278,
|
| 760 |
+
"f1": 0.9205762837089029,
|
| 761 |
+
"number": 1341
|
| 762 |
+
},
|
| 763 |
+
"PER": {
|
| 764 |
+
"precision": 0.9545934530095037,
|
| 765 |
+
"recall": 0.9815418023887079,
|
| 766 |
+
"f1": 0.9678800856531049,
|
| 767 |
+
"number": 1842
|
| 768 |
+
}
|
| 769 |
+
},
|
| 770 |
+
"eval_runtime": 7.7732,
|
| 771 |
+
"eval_samples_per_second": 418.106,
|
| 772 |
+
"eval_steps_per_second": 52.36,
|
| 773 |
+
"epoch": 12.0,
|
| 774 |
+
"step": 21072
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"loss": 0.0006,
|
| 778 |
+
"grad_norm": 0.0022160038352012634,
|
| 779 |
+
"learning_rate": 7.75626423690205e-06,
|
| 780 |
+
"epoch": 12.24373576309795,
|
| 781 |
+
"step": 21500
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"loss": 0.0008,
|
| 785 |
+
"grad_norm": 0.00019432637782301754,
|
| 786 |
+
"learning_rate": 7.471526195899773e-06,
|
| 787 |
+
"epoch": 12.528473804100228,
|
| 788 |
+
"step": 22000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"loss": 0.0011,
|
| 792 |
+
"grad_norm": 1.6397913694381714,
|
| 793 |
+
"learning_rate": 7.186788154897495e-06,
|
| 794 |
+
"epoch": 12.813211845102506,
|
| 795 |
+
"step": 22500
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"eval_loss": 0.12505799531936646,
|
| 799 |
+
"eval_overall": {
|
| 800 |
+
"precision": 0.9386808087504143,
|
| 801 |
+
"recall": 0.9532144059239314,
|
| 802 |
+
"f1": 0.9458917835671342,
|
| 803 |
+
"accuracy": 0.9860923058809677
|
| 804 |
+
},
|
| 805 |
+
"eval_per_label": {
|
| 806 |
+
"LOC": {
|
| 807 |
+
"precision": 0.9675148890092041,
|
| 808 |
+
"recall": 0.9727817093086554,
|
| 809 |
+
"f1": 0.9701411509229099,
|
| 810 |
+
"number": 1837
|
| 811 |
+
},
|
| 812 |
+
"MISC": {
|
| 813 |
+
"precision": 0.8782791185729276,
|
| 814 |
+
"recall": 0.9078091106290672,
|
| 815 |
+
"f1": 0.8927999999999999,
|
| 816 |
+
"number": 922
|
| 817 |
+
},
|
| 818 |
+
"ORG": {
|
| 819 |
+
"precision": 0.9241741741741741,
|
| 820 |
+
"recall": 0.9179716629381058,
|
| 821 |
+
"f1": 0.9210624766180322,
|
| 822 |
+
"number": 1341
|
| 823 |
+
},
|
| 824 |
+
"PER": {
|
| 825 |
+
"precision": 0.9511041009463722,
|
| 826 |
+
"recall": 0.9820846905537459,
|
| 827 |
+
"f1": 0.9663461538461537,
|
| 828 |
+
"number": 1842
|
| 829 |
+
}
|
| 830 |
+
},
|
| 831 |
+
"eval_runtime": 7.5826,
|
| 832 |
+
"eval_samples_per_second": 428.613,
|
| 833 |
+
"eval_steps_per_second": 53.676,
|
| 834 |
+
"epoch": 13.0,
|
| 835 |
+
"step": 22828
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"loss": 0.0015,
|
| 839 |
+
"grad_norm": 0.00018138765881303698,
|
| 840 |
+
"learning_rate": 6.9020501138952166e-06,
|
| 841 |
+
"epoch": 13.097949886104784,
|
| 842 |
+
"step": 23000
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"loss": 0.0007,
|
| 846 |
+
"grad_norm": 0.00034025911008939147,
|
| 847 |
+
"learning_rate": 6.617312072892939e-06,
|
| 848 |
+
"epoch": 13.382687927107062,
|
| 849 |
+
"step": 23500
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"loss": 0.0006,
|
| 853 |
+
"grad_norm": 0.00041584973223507404,
|
| 854 |
+
"learning_rate": 6.3325740318906616e-06,
|
| 855 |
+
"epoch": 13.66742596810934,
|
| 856 |
+
"step": 24000
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"loss": 0.0008,
|
| 860 |
+
"grad_norm": 0.0006968477973714471,
|
| 861 |
+
"learning_rate": 6.047835990888384e-06,
|
| 862 |
+
"epoch": 13.952164009111618,
|
| 863 |
+
"step": 24500
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"eval_loss": 0.12853111326694489,
|
| 867 |
+
"eval_overall": {
|
| 868 |
+
"precision": 0.9430043203722167,
|
| 869 |
+
"recall": 0.9550656344665096,
|
| 870 |
+
"f1": 0.9489966555183946,
|
| 871 |
+
"accuracy": 0.986342497203744
|
| 872 |
+
},
|
| 873 |
+
"eval_per_label": {
|
| 874 |
+
"LOC": {
|
| 875 |
+
"precision": 0.9736986301369863,
|
| 876 |
+
"recall": 0.9673380511703865,
|
| 877 |
+
"f1": 0.9705079191698526,
|
| 878 |
+
"number": 1837
|
| 879 |
+
},
|
| 880 |
+
"MISC": {
|
| 881 |
+
"precision": 0.8892438764643238,
|
| 882 |
+
"recall": 0.9056399132321041,
|
| 883 |
+
"f1": 0.8973670069854918,
|
| 884 |
+
"number": 922
|
| 885 |
+
},
|
| 886 |
+
"ORG": {
|
| 887 |
+
"precision": 0.9200293470286134,
|
| 888 |
+
"recall": 0.9351230425055929,
|
| 889 |
+
"f1": 0.9275147928994084,
|
| 890 |
+
"number": 1341
|
| 891 |
+
},
|
| 892 |
+
"PER": {
|
| 893 |
+
"precision": 0.9566367001586462,
|
| 894 |
+
"recall": 0.9820846905537459,
|
| 895 |
+
"f1": 0.9691936780069649,
|
| 896 |
+
"number": 1842
|
| 897 |
+
}
|
| 898 |
+
},
|
| 899 |
+
"eval_runtime": 7.5688,
|
| 900 |
+
"eval_samples_per_second": 429.396,
|
| 901 |
+
"eval_steps_per_second": 53.774,
|
| 902 |
+
"epoch": 14.0,
|
| 903 |
+
"step": 24584
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"loss": 0.0006,
|
| 907 |
+
"grad_norm": 0.00020477738871704787,
|
| 908 |
+
"learning_rate": 5.763097949886105e-06,
|
| 909 |
+
"epoch": 14.236902050113896,
|
| 910 |
+
"step": 25000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"loss": 0.0007,
|
| 914 |
+
"grad_norm": 0.00032146400189958513,
|
| 915 |
+
"learning_rate": 5.478359908883827e-06,
|
| 916 |
+
"epoch": 14.521640091116174,
|
| 917 |
+
"step": 25500
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"loss": 0.0014,
|
| 921 |
+
"grad_norm": 0.000293695367872715,
|
| 922 |
+
"learning_rate": 5.19362186788155e-06,
|
| 923 |
+
"epoch": 14.806378132118452,
|
| 924 |
+
"step": 26000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"eval_loss": 0.1253676861524582,
|
| 928 |
+
"eval_overall": {
|
| 929 |
+
"precision": 0.938937613767996,
|
| 930 |
+
"recall": 0.9548973409626389,
|
| 931 |
+
"f1": 0.9468502294534834,
|
| 932 |
+
"accuracy": 0.9861364572908695
|
| 933 |
+
},
|
| 934 |
+
"eval_per_label": {
|
| 935 |
+
"LOC": {
|
| 936 |
+
"precision": 0.9607948442534908,
|
| 937 |
+
"recall": 0.9738704409363091,
|
| 938 |
+
"f1": 0.9672884563395511,
|
| 939 |
+
"number": 1837
|
| 940 |
+
},
|
| 941 |
+
"MISC": {
|
| 942 |
+
"precision": 0.8742203742203742,
|
| 943 |
+
"recall": 0.9121475054229935,
|
| 944 |
+
"f1": 0.8927813163481952,
|
| 945 |
+
"number": 922
|
| 946 |
+
},
|
| 947 |
+
"ORG": {
|
| 948 |
+
"precision": 0.9322289156626506,
|
| 949 |
+
"recall": 0.9231916480238628,
|
| 950 |
+
"f1": 0.9276882727613338,
|
| 951 |
+
"number": 1341
|
| 952 |
+
},
|
| 953 |
+
"PER": {
|
| 954 |
+
"precision": 0.9550502379693284,
|
| 955 |
+
"recall": 0.9804560260586319,
|
| 956 |
+
"f1": 0.9675863916421109,
|
| 957 |
+
"number": 1842
|
| 958 |
+
}
|
| 959 |
+
},
|
| 960 |
+
"eval_runtime": 8.6163,
|
| 961 |
+
"eval_samples_per_second": 377.191,
|
| 962 |
+
"eval_steps_per_second": 47.236,
|
| 963 |
+
"epoch": 15.0,
|
| 964 |
+
"step": 26340
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"loss": 0.0008,
|
| 968 |
+
"grad_norm": 0.00019819244334939867,
|
| 969 |
+
"learning_rate": 4.908883826879272e-06,
|
| 970 |
+
"epoch": 15.09111617312073,
|
| 971 |
+
"step": 26500
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"loss": 0.0008,
|
| 975 |
+
"grad_norm": 0.00021079520229250193,
|
| 976 |
+
"learning_rate": 4.624145785876993e-06,
|
| 977 |
+
"epoch": 15.375854214123008,
|
| 978 |
+
"step": 27000
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"loss": 0.0003,
|
| 982 |
+
"grad_norm": 0.00012124140630476177,
|
| 983 |
+
"learning_rate": 4.339407744874715e-06,
|
| 984 |
+
"epoch": 15.660592255125284,
|
| 985 |
+
"step": 27500
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"loss": 0.0008,
|
| 989 |
+
"grad_norm": 0.000182148942258209,
|
| 990 |
+
"learning_rate": 4.054669703872437e-06,
|
| 991 |
+
"epoch": 15.945330296127562,
|
| 992 |
+
"step": 28000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"eval_loss": 0.12416187673807144,
|
| 996 |
+
"eval_overall": {
|
| 997 |
+
"precision": 0.9447862963578912,
|
| 998 |
+
"recall": 0.9560753954897341,
|
| 999 |
+
"f1": 0.9503973232956922,
|
| 1000 |
+
"accuracy": 0.9866957084829575
|
| 1001 |
+
},
|
| 1002 |
+
"eval_per_label": {
|
| 1003 |
+
"LOC": {
|
| 1004 |
+
"precision": 0.9680043383947939,
|
| 1005 |
+
"recall": 0.9716929776810016,
|
| 1006 |
+
"f1": 0.969845150774246,
|
| 1007 |
+
"number": 1837
|
| 1008 |
+
},
|
| 1009 |
+
"MISC": {
|
| 1010 |
+
"precision": 0.8954108858057631,
|
| 1011 |
+
"recall": 0.9099783080260304,
|
| 1012 |
+
"f1": 0.9026358257127487,
|
| 1013 |
+
"number": 922
|
| 1014 |
+
},
|
| 1015 |
+
"ORG": {
|
| 1016 |
+
"precision": 0.9301634472511144,
|
| 1017 |
+
"recall": 0.9336316181953765,
|
| 1018 |
+
"f1": 0.9318943059173799,
|
| 1019 |
+
"number": 1341
|
| 1020 |
+
},
|
| 1021 |
+
"PER": {
|
| 1022 |
+
"precision": 0.957051961823966,
|
| 1023 |
+
"recall": 0.9799131378935939,
|
| 1024 |
+
"f1": 0.9683476394849786,
|
| 1025 |
+
"number": 1842
|
| 1026 |
+
}
|
| 1027 |
+
},
|
| 1028 |
+
"eval_runtime": 8.6628,
|
| 1029 |
+
"eval_samples_per_second": 375.168,
|
| 1030 |
+
"eval_steps_per_second": 46.983,
|
| 1031 |
+
"epoch": 16.0,
|
| 1032 |
+
"step": 28096
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"loss": 0.001,
|
| 1036 |
+
"grad_norm": 0.019741835072636604,
|
| 1037 |
+
"learning_rate": 3.76993166287016e-06,
|
| 1038 |
+
"epoch": 16.23006833712984,
|
| 1039 |
+
"step": 28500
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"loss": 0.0003,
|
| 1043 |
+
"grad_norm": 0.00012148160021752119,
|
| 1044 |
+
"learning_rate": 3.4851936218678815e-06,
|
| 1045 |
+
"epoch": 16.514806378132118,
|
| 1046 |
+
"step": 29000
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"loss": 0.0005,
|
| 1050 |
+
"grad_norm": 0.00011341737263137475,
|
| 1051 |
+
"learning_rate": 3.200455580865604e-06,
|
| 1052 |
+
"epoch": 16.799544419134396,
|
| 1053 |
+
"step": 29500
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"eval_loss": 0.12274094671010971,
|
| 1057 |
+
"eval_overall": {
|
| 1058 |
+
"precision": 0.9438631456568676,
|
| 1059 |
+
"recall": 0.9564119824974756,
|
| 1060 |
+
"f1": 0.9500961297333445,
|
| 1061 |
+
"accuracy": 0.9868428798492965
|
| 1062 |
+
},
|
| 1063 |
+
"eval_per_label": {
|
| 1064 |
+
"LOC": {
|
| 1065 |
+
"precision": 0.971195652173913,
|
| 1066 |
+
"recall": 0.9727817093086554,
|
| 1067 |
+
"f1": 0.9719880337231438,
|
| 1068 |
+
"number": 1837
|
| 1069 |
+
},
|
| 1070 |
+
"MISC": {
|
| 1071 |
+
"precision": 0.8761609907120743,
|
| 1072 |
+
"recall": 0.920824295010846,
|
| 1073 |
+
"f1": 0.8979375991538868,
|
| 1074 |
+
"number": 922
|
| 1075 |
+
},
|
| 1076 |
+
"ORG": {
|
| 1077 |
+
"precision": 0.9307004470938898,
|
| 1078 |
+
"recall": 0.9313944817300522,
|
| 1079 |
+
"f1": 0.9310473350726798,
|
| 1080 |
+
"number": 1341
|
| 1081 |
+
},
|
| 1082 |
+
"PER": {
|
| 1083 |
+
"precision": 0.9614973262032086,
|
| 1084 |
+
"recall": 0.9761129207383279,
|
| 1085 |
+
"f1": 0.9687500000000001,
|
| 1086 |
+
"number": 1842
|
| 1087 |
+
}
|
| 1088 |
+
},
|
| 1089 |
+
"eval_runtime": 8.9038,
|
| 1090 |
+
"eval_samples_per_second": 365.012,
|
| 1091 |
+
"eval_steps_per_second": 45.711,
|
| 1092 |
+
"epoch": 17.0,
|
| 1093 |
+
"step": 29852
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"loss": 0.0007,
|
| 1097 |
+
"grad_norm": 0.00015434053784701973,
|
| 1098 |
+
"learning_rate": 2.9157175398633257e-06,
|
| 1099 |
+
"epoch": 17.084282460136674,
|
| 1100 |
+
"step": 30000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"loss": 0.0,
|
| 1104 |
+
"grad_norm": 0.00021848917822353542,
|
| 1105 |
+
"learning_rate": 2.6309794988610482e-06,
|
| 1106 |
+
"epoch": 17.36902050113895,
|
| 1107 |
+
"step": 30500
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"loss": 0.0002,
|
| 1111 |
+
"grad_norm": 0.0001007779865176417,
|
| 1112 |
+
"learning_rate": 2.34624145785877e-06,
|
| 1113 |
+
"epoch": 17.65375854214123,
|
| 1114 |
+
"step": 31000
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"loss": 0.0002,
|
| 1118 |
+
"grad_norm": 0.00010259783448418602,
|
| 1119 |
+
"learning_rate": 2.061503416856492e-06,
|
| 1120 |
+
"epoch": 17.938496583143507,
|
| 1121 |
+
"step": 31500
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"eval_loss": 0.12293359637260437,
|
| 1125 |
+
"eval_overall": {
|
| 1126 |
+
"precision": 0.9440199335548173,
|
| 1127 |
+
"recall": 0.9564119824974756,
|
| 1128 |
+
"f1": 0.9501755559271025,
|
| 1129 |
+
"accuracy": 0.9870636368988049
|
| 1130 |
+
},
|
| 1131 |
+
"eval_per_label": {
|
| 1132 |
+
"LOC": {
|
| 1133 |
+
"precision": 0.9701249321021184,
|
| 1134 |
+
"recall": 0.9722373434948285,
|
| 1135 |
+
"f1": 0.9711799891245242,
|
| 1136 |
+
"number": 1837
|
| 1137 |
+
},
|
| 1138 |
+
"MISC": {
|
| 1139 |
+
"precision": 0.8810020876826722,
|
| 1140 |
+
"recall": 0.9154013015184381,
|
| 1141 |
+
"f1": 0.8978723404255319,
|
| 1142 |
+
"number": 922
|
| 1143 |
+
},
|
| 1144 |
+
"ORG": {
|
| 1145 |
+
"precision": 0.926829268292683,
|
| 1146 |
+
"recall": 0.9351230425055929,
|
| 1147 |
+
"f1": 0.9309576837416481,
|
| 1148 |
+
"number": 1341
|
| 1149 |
+
},
|
| 1150 |
+
"PER": {
|
| 1151 |
+
"precision": 0.9630620985010707,
|
| 1152 |
+
"recall": 0.9766558089033659,
|
| 1153 |
+
"f1": 0.969811320754717,
|
| 1154 |
+
"number": 1842
|
| 1155 |
+
}
|
| 1156 |
+
},
|
| 1157 |
+
"eval_runtime": 8.7313,
|
| 1158 |
+
"eval_samples_per_second": 372.226,
|
| 1159 |
+
"eval_steps_per_second": 46.614,
|
| 1160 |
+
"epoch": 18.0,
|
| 1161 |
+
"step": 31608
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"loss": 0.0001,
|
| 1165 |
+
"grad_norm": 0.0010326108895242214,
|
| 1166 |
+
"learning_rate": 1.7767653758542143e-06,
|
| 1167 |
+
"epoch": 18.223234624145785,
|
| 1168 |
+
"step": 32000
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"loss": 0.0007,
|
| 1172 |
+
"grad_norm": 0.00016551795124541968,
|
| 1173 |
+
"learning_rate": 1.4920273348519363e-06,
|
| 1174 |
+
"epoch": 18.507972665148063,
|
| 1175 |
+
"step": 32500
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"loss": 0.0004,
|
| 1179 |
+
"grad_norm": 0.00024008983746170998,
|
| 1180 |
+
"learning_rate": 1.2072892938496584e-06,
|
| 1181 |
+
"epoch": 18.79271070615034,
|
| 1182 |
+
"step": 33000
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"eval_loss": 0.12236841022968292,
|
| 1186 |
+
"eval_overall": {
|
| 1187 |
+
"precision": 0.944813829787234,
|
| 1188 |
+
"recall": 0.9565802760013463,
|
| 1189 |
+
"f1": 0.9506606455929084,
|
| 1190 |
+
"accuracy": 0.9872255254017779
|
| 1191 |
+
},
|
| 1192 |
+
"eval_per_label": {
|
| 1193 |
+
"LOC": {
|
| 1194 |
+
"precision": 0.971677559912854,
|
| 1195 |
+
"recall": 0.9711486118671747,
|
| 1196 |
+
"f1": 0.9714130138851075,
|
| 1197 |
+
"number": 1837
|
| 1198 |
+
},
|
| 1199 |
+
"MISC": {
|
| 1200 |
+
"precision": 0.8712667353244078,
|
| 1201 |
+
"recall": 0.9175704989154013,
|
| 1202 |
+
"f1": 0.8938193343898574,
|
| 1203 |
+
"number": 922
|
| 1204 |
+
},
|
| 1205 |
+
"ORG": {
|
| 1206 |
+
"precision": 0.9383458646616541,
|
| 1207 |
+
"recall": 0.930648769574944,
|
| 1208 |
+
"f1": 0.9344814676151254,
|
| 1209 |
+
"number": 1341
|
| 1210 |
+
},
|
| 1211 |
+
"PER": {
|
| 1212 |
+
"precision": 0.961149547631719,
|
| 1213 |
+
"recall": 0.9804560260586319,
|
| 1214 |
+
"f1": 0.9707067992475142,
|
| 1215 |
+
"number": 1842
|
| 1216 |
+
}
|
| 1217 |
+
},
|
| 1218 |
+
"eval_runtime": 6.5081,
|
| 1219 |
+
"eval_samples_per_second": 499.381,
|
| 1220 |
+
"eval_steps_per_second": 62.538,
|
| 1221 |
+
"epoch": 19.0,
|
| 1222 |
+
"step": 33364
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"loss": 0.0001,
|
| 1226 |
+
"grad_norm": 0.00016071839490905404,
|
| 1227 |
+
"learning_rate": 9.225512528473805e-07,
|
| 1228 |
+
"epoch": 19.07744874715262,
|
| 1229 |
+
"step": 33500
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"loss": 0.0001,
|
| 1233 |
+
"grad_norm": 9.665234392741695e-05,
|
| 1234 |
+
"learning_rate": 6.378132118451026e-07,
|
| 1235 |
+
"epoch": 19.362186788154897,
|
| 1236 |
+
"step": 34000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"loss": 0.0001,
|
| 1240 |
+
"grad_norm": 0.004949676804244518,
|
| 1241 |
+
"learning_rate": 3.530751708428246e-07,
|
| 1242 |
+
"epoch": 19.646924829157175,
|
| 1243 |
+
"step": 34500
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"loss": 0.0005,
|
| 1247 |
+
"grad_norm": 0.00014002641546539962,
|
| 1248 |
+
"learning_rate": 6.83371298405467e-08,
|
| 1249 |
+
"epoch": 19.931662870159453,
|
| 1250 |
+
"step": 35000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"eval_loss": 0.12258101999759674,
|
| 1254 |
+
"eval_overall": {
|
| 1255 |
+
"precision": 0.9443891102257637,
|
| 1256 |
+
"recall": 0.9574217435207001,
|
| 1257 |
+
"f1": 0.9508607721878657,
|
| 1258 |
+
"accuracy": 0.9872549596750456
|
| 1259 |
+
},
|
| 1260 |
+
"eval_per_label": {
|
| 1261 |
+
"LOC": {
|
| 1262 |
+
"precision": 0.9727371864776445,
|
| 1263 |
+
"recall": 0.9711486118671747,
|
| 1264 |
+
"f1": 0.9719422500681013,
|
| 1265 |
+
"number": 1837
|
| 1266 |
+
},
|
| 1267 |
+
"MISC": {
|
| 1268 |
+
"precision": 0.8746113989637305,
|
| 1269 |
+
"recall": 0.9154013015184381,
|
| 1270 |
+
"f1": 0.8945416004239534,
|
| 1271 |
+
"number": 922
|
| 1272 |
+
},
|
| 1273 |
+
"ORG": {
|
| 1274 |
+
"precision": 0.9336810730253353,
|
| 1275 |
+
"recall": 0.9343773303504848,
|
| 1276 |
+
"f1": 0.9340290719344019,
|
| 1277 |
+
"number": 1341
|
| 1278 |
+
},
|
| 1279 |
+
"PER": {
|
| 1280 |
+
"precision": 0.960169941582581,
|
| 1281 |
+
"recall": 0.9815418023887079,
|
| 1282 |
+
"f1": 0.970738255033557,
|
| 1283 |
+
"number": 1842
|
| 1284 |
+
}
|
| 1285 |
+
},
|
| 1286 |
+
"eval_runtime": 6.7405,
|
| 1287 |
+
"eval_samples_per_second": 482.16,
|
| 1288 |
+
"eval_steps_per_second": 60.381,
|
| 1289 |
+
"epoch": 20.0,
|
| 1290 |
+
"step": 35120
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"train_runtime": 3604.2086,
|
| 1294 |
+
"train_samples_per_second": 77.914,
|
| 1295 |
+
"train_steps_per_second": 9.744,
|
| 1296 |
+
"total_flos": 6145780901050062.0,
|
| 1297 |
+
"train_loss": 0.013757273165407275,
|
| 1298 |
+
"epoch": 20.0,
|
| 1299 |
+
"step": 35120
|
| 1300 |
+
}
|
| 1301 |
+
]
|