Overwrite main branch with checkpoint-27705 weights
Browse files- optimizer.pt +3 -0
- pytorch_model.bin +1 -1
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +1244 -0
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ec27e3c16d71555fb1466c9d71b3dfb5a9e2c18fbfed33624c9fd4958b6d76a
|
| 3 |
+
size 124642443
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 62314258
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1993b39cf8cb41537be7b1ac701aade8f6fe7b5701d1f4fe301e7fce4ee62c88
|
| 3 |
size 62314258
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c4daf92986419a40c249557c6cde2b16e9b9d8bf7edd0389ff117a2ae6d12ca5
|
| 3 |
+
size 14645
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:72d4c8aa281667dff58085c289eca3da10682792ba8a018e903e710e3c3e6527
|
| 3 |
+
size 1465
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1244 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 27705,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.02707092582566324,
|
| 14 |
+
"grad_norm": 0.9713733196258545,
|
| 15 |
+
"learning_rate": 9.910124526258799e-05,
|
| 16 |
+
"loss": 3.4945,
|
| 17 |
+
"step": 250
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.05414185165132648,
|
| 21 |
+
"grad_norm": 0.9521434307098389,
|
| 22 |
+
"learning_rate": 9.819888106839921e-05,
|
| 23 |
+
"loss": 3.5097,
|
| 24 |
+
"step": 500
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05414185165132648,
|
| 28 |
+
"eval_loss": 2.95473575592041,
|
| 29 |
+
"eval_runtime": 87.829,
|
| 30 |
+
"eval_samples_per_second": 112.15,
|
| 31 |
+
"eval_steps_per_second": 7.014,
|
| 32 |
+
"step": 500
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.08121277747698971,
|
| 36 |
+
"grad_norm": 0.9334385395050049,
|
| 37 |
+
"learning_rate": 9.729651687421044e-05,
|
| 38 |
+
"loss": 3.4922,
|
| 39 |
+
"step": 750
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.10828370330265295,
|
| 43 |
+
"grad_norm": 1.2239603996276855,
|
| 44 |
+
"learning_rate": 9.639415268002166e-05,
|
| 45 |
+
"loss": 3.477,
|
| 46 |
+
"step": 1000
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10828370330265295,
|
| 50 |
+
"eval_loss": 2.9319565296173096,
|
| 51 |
+
"eval_runtime": 87.8964,
|
| 52 |
+
"eval_samples_per_second": 112.064,
|
| 53 |
+
"eval_steps_per_second": 7.008,
|
| 54 |
+
"step": 1000
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.1353546291283162,
|
| 58 |
+
"grad_norm": 1.1727226972579956,
|
| 59 |
+
"learning_rate": 9.549178848583288e-05,
|
| 60 |
+
"loss": 3.493,
|
| 61 |
+
"step": 1250
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.16242555495397942,
|
| 65 |
+
"grad_norm": 0.966973602771759,
|
| 66 |
+
"learning_rate": 9.458942429164411e-05,
|
| 67 |
+
"loss": 3.5013,
|
| 68 |
+
"step": 1500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.16242555495397942,
|
| 72 |
+
"eval_loss": 2.9275825023651123,
|
| 73 |
+
"eval_runtime": 87.899,
|
| 74 |
+
"eval_samples_per_second": 112.06,
|
| 75 |
+
"eval_steps_per_second": 7.008,
|
| 76 |
+
"step": 1500
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.18949648077964265,
|
| 80 |
+
"grad_norm": 0.9813922047615051,
|
| 81 |
+
"learning_rate": 9.368706009745533e-05,
|
| 82 |
+
"loss": 3.4907,
|
| 83 |
+
"step": 1750
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.2165674066053059,
|
| 87 |
+
"grad_norm": 1.027085542678833,
|
| 88 |
+
"learning_rate": 9.278469590326656e-05,
|
| 89 |
+
"loss": 3.4588,
|
| 90 |
+
"step": 2000
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.2165674066053059,
|
| 94 |
+
"eval_loss": 2.9044992923736572,
|
| 95 |
+
"eval_runtime": 87.8776,
|
| 96 |
+
"eval_samples_per_second": 112.088,
|
| 97 |
+
"eval_steps_per_second": 7.01,
|
| 98 |
+
"step": 2000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.24363833243096913,
|
| 102 |
+
"grad_norm": 1.0764214992523193,
|
| 103 |
+
"learning_rate": 9.188233170907778e-05,
|
| 104 |
+
"loss": 3.4531,
|
| 105 |
+
"step": 2250
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.2707092582566324,
|
| 109 |
+
"grad_norm": 1.0297119617462158,
|
| 110 |
+
"learning_rate": 9.0979967514889e-05,
|
| 111 |
+
"loss": 3.4445,
|
| 112 |
+
"step": 2500
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.2707092582566324,
|
| 116 |
+
"eval_loss": 2.8855738639831543,
|
| 117 |
+
"eval_runtime": 87.8728,
|
| 118 |
+
"eval_samples_per_second": 112.094,
|
| 119 |
+
"eval_steps_per_second": 7.01,
|
| 120 |
+
"step": 2500
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.2977801840822956,
|
| 124 |
+
"grad_norm": 0.9697523713111877,
|
| 125 |
+
"learning_rate": 9.007760332070024e-05,
|
| 126 |
+
"loss": 3.4349,
|
| 127 |
+
"step": 2750
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.32485110990795885,
|
| 131 |
+
"grad_norm": 0.9611329436302185,
|
| 132 |
+
"learning_rate": 8.917523912651147e-05,
|
| 133 |
+
"loss": 3.4213,
|
| 134 |
+
"step": 3000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.32485110990795885,
|
| 138 |
+
"eval_loss": 2.8725759983062744,
|
| 139 |
+
"eval_runtime": 87.9054,
|
| 140 |
+
"eval_samples_per_second": 112.052,
|
| 141 |
+
"eval_steps_per_second": 7.008,
|
| 142 |
+
"step": 3000
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3519220357336221,
|
| 146 |
+
"grad_norm": 1.000977873802185,
|
| 147 |
+
"learning_rate": 8.827287493232269e-05,
|
| 148 |
+
"loss": 3.4239,
|
| 149 |
+
"step": 3250
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.3789929615592853,
|
| 153 |
+
"grad_norm": 1.0573837757110596,
|
| 154 |
+
"learning_rate": 8.737051073813391e-05,
|
| 155 |
+
"loss": 3.4015,
|
| 156 |
+
"step": 3500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.3789929615592853,
|
| 160 |
+
"eval_loss": 2.8721706867218018,
|
| 161 |
+
"eval_runtime": 87.8848,
|
| 162 |
+
"eval_samples_per_second": 112.078,
|
| 163 |
+
"eval_steps_per_second": 7.009,
|
| 164 |
+
"step": 3500
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.4060638873849486,
|
| 168 |
+
"grad_norm": 0.9532322883605957,
|
| 169 |
+
"learning_rate": 8.646814654394514e-05,
|
| 170 |
+
"loss": 3.401,
|
| 171 |
+
"step": 3750
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4331348132106118,
|
| 175 |
+
"grad_norm": 1.0036464929580688,
|
| 176 |
+
"learning_rate": 8.556578234975636e-05,
|
| 177 |
+
"loss": 3.3984,
|
| 178 |
+
"step": 4000
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.4331348132106118,
|
| 182 |
+
"eval_loss": 2.863417863845825,
|
| 183 |
+
"eval_runtime": 87.8316,
|
| 184 |
+
"eval_samples_per_second": 112.146,
|
| 185 |
+
"eval_steps_per_second": 7.013,
|
| 186 |
+
"step": 4000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.46020573903627504,
|
| 190 |
+
"grad_norm": 0.8860335350036621,
|
| 191 |
+
"learning_rate": 8.466341815556758e-05,
|
| 192 |
+
"loss": 3.4081,
|
| 193 |
+
"step": 4250
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.48727666486193827,
|
| 197 |
+
"grad_norm": 1.4852999448776245,
|
| 198 |
+
"learning_rate": 8.376105396137882e-05,
|
| 199 |
+
"loss": 3.4105,
|
| 200 |
+
"step": 4500
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.48727666486193827,
|
| 204 |
+
"eval_loss": 2.8409109115600586,
|
| 205 |
+
"eval_runtime": 87.9025,
|
| 206 |
+
"eval_samples_per_second": 112.056,
|
| 207 |
+
"eval_steps_per_second": 7.008,
|
| 208 |
+
"step": 4500
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.5143475906876015,
|
| 212 |
+
"grad_norm": 0.936912477016449,
|
| 213 |
+
"learning_rate": 8.285868976719005e-05,
|
| 214 |
+
"loss": 3.3952,
|
| 215 |
+
"step": 4750
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.5414185165132648,
|
| 219 |
+
"grad_norm": 1.0409356355667114,
|
| 220 |
+
"learning_rate": 8.195632557300127e-05,
|
| 221 |
+
"loss": 3.4039,
|
| 222 |
+
"step": 5000
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.5414185165132648,
|
| 226 |
+
"eval_loss": 2.8329713344573975,
|
| 227 |
+
"eval_runtime": 87.8287,
|
| 228 |
+
"eval_samples_per_second": 112.15,
|
| 229 |
+
"eval_steps_per_second": 7.014,
|
| 230 |
+
"step": 5000
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.568489442338928,
|
| 234 |
+
"grad_norm": 0.9606480002403259,
|
| 235 |
+
"learning_rate": 8.10539613788125e-05,
|
| 236 |
+
"loss": 3.3819,
|
| 237 |
+
"step": 5250
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.5955603681645912,
|
| 241 |
+
"grad_norm": 1.142527461051941,
|
| 242 |
+
"learning_rate": 8.015159718462372e-05,
|
| 243 |
+
"loss": 3.3898,
|
| 244 |
+
"step": 5500
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.5955603681645912,
|
| 248 |
+
"eval_loss": 2.8344321250915527,
|
| 249 |
+
"eval_runtime": 87.8057,
|
| 250 |
+
"eval_samples_per_second": 112.18,
|
| 251 |
+
"eval_steps_per_second": 7.015,
|
| 252 |
+
"step": 5500
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.6226312939902545,
|
| 256 |
+
"grad_norm": 1.2294474840164185,
|
| 257 |
+
"learning_rate": 7.924923299043494e-05,
|
| 258 |
+
"loss": 3.3769,
|
| 259 |
+
"step": 5750
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.6497022198159177,
|
| 263 |
+
"grad_norm": 1.0300960540771484,
|
| 264 |
+
"learning_rate": 7.834686879624617e-05,
|
| 265 |
+
"loss": 3.3621,
|
| 266 |
+
"step": 6000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.6497022198159177,
|
| 270 |
+
"eval_loss": 2.8137757778167725,
|
| 271 |
+
"eval_runtime": 87.8367,
|
| 272 |
+
"eval_samples_per_second": 112.14,
|
| 273 |
+
"eval_steps_per_second": 7.013,
|
| 274 |
+
"step": 6000
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.676773145641581,
|
| 278 |
+
"grad_norm": 1.076827883720398,
|
| 279 |
+
"learning_rate": 7.74445046020574e-05,
|
| 280 |
+
"loss": 3.3704,
|
| 281 |
+
"step": 6250
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.7038440714672441,
|
| 285 |
+
"grad_norm": 1.4118528366088867,
|
| 286 |
+
"learning_rate": 7.654214040786863e-05,
|
| 287 |
+
"loss": 3.367,
|
| 288 |
+
"step": 6500
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.7038440714672441,
|
| 292 |
+
"eval_loss": 2.8218181133270264,
|
| 293 |
+
"eval_runtime": 87.9044,
|
| 294 |
+
"eval_samples_per_second": 112.053,
|
| 295 |
+
"eval_steps_per_second": 7.008,
|
| 296 |
+
"step": 6500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.7309149972929074,
|
| 300 |
+
"grad_norm": 1.0568158626556396,
|
| 301 |
+
"learning_rate": 7.563977621367985e-05,
|
| 302 |
+
"loss": 3.3509,
|
| 303 |
+
"step": 6750
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.7579859231185706,
|
| 307 |
+
"grad_norm": 1.0436064004898071,
|
| 308 |
+
"learning_rate": 7.473741201949108e-05,
|
| 309 |
+
"loss": 3.3496,
|
| 310 |
+
"step": 7000
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.7579859231185706,
|
| 314 |
+
"eval_loss": 2.8048973083496094,
|
| 315 |
+
"eval_runtime": 87.8448,
|
| 316 |
+
"eval_samples_per_second": 112.13,
|
| 317 |
+
"eval_steps_per_second": 7.012,
|
| 318 |
+
"step": 7000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7850568489442339,
|
| 322 |
+
"grad_norm": 0.9466687440872192,
|
| 323 |
+
"learning_rate": 7.38350478253023e-05,
|
| 324 |
+
"loss": 3.3381,
|
| 325 |
+
"step": 7250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.8121277747698972,
|
| 329 |
+
"grad_norm": 0.9336101412773132,
|
| 330 |
+
"learning_rate": 7.293268363111352e-05,
|
| 331 |
+
"loss": 3.3394,
|
| 332 |
+
"step": 7500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.8121277747698972,
|
| 336 |
+
"eval_loss": 2.80117130279541,
|
| 337 |
+
"eval_runtime": 87.8737,
|
| 338 |
+
"eval_samples_per_second": 112.093,
|
| 339 |
+
"eval_steps_per_second": 7.01,
|
| 340 |
+
"step": 7500
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.8391987005955603,
|
| 344 |
+
"grad_norm": 1.0283994674682617,
|
| 345 |
+
"learning_rate": 7.203031943692475e-05,
|
| 346 |
+
"loss": 3.3379,
|
| 347 |
+
"step": 7750
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.8662696264212236,
|
| 351 |
+
"grad_norm": 0.9297059774398804,
|
| 352 |
+
"learning_rate": 7.112795524273597e-05,
|
| 353 |
+
"loss": 3.3196,
|
| 354 |
+
"step": 8000
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.8662696264212236,
|
| 358 |
+
"eval_loss": 2.8071343898773193,
|
| 359 |
+
"eval_runtime": 87.7691,
|
| 360 |
+
"eval_samples_per_second": 112.226,
|
| 361 |
+
"eval_steps_per_second": 7.018,
|
| 362 |
+
"step": 8000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8933405522468868,
|
| 366 |
+
"grad_norm": 1.1025398969650269,
|
| 367 |
+
"learning_rate": 7.02255910485472e-05,
|
| 368 |
+
"loss": 3.3135,
|
| 369 |
+
"step": 8250
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.9204114780725501,
|
| 373 |
+
"grad_norm": 0.9545220136642456,
|
| 374 |
+
"learning_rate": 6.932322685435842e-05,
|
| 375 |
+
"loss": 3.3235,
|
| 376 |
+
"step": 8500
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.9204114780725501,
|
| 380 |
+
"eval_loss": 2.783069610595703,
|
| 381 |
+
"eval_runtime": 88.0325,
|
| 382 |
+
"eval_samples_per_second": 111.891,
|
| 383 |
+
"eval_steps_per_second": 6.997,
|
| 384 |
+
"step": 8500
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.9474824038982134,
|
| 388 |
+
"grad_norm": 1.0582618713378906,
|
| 389 |
+
"learning_rate": 6.842086266016964e-05,
|
| 390 |
+
"loss": 3.3076,
|
| 391 |
+
"step": 8750
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9745533297238765,
|
| 395 |
+
"grad_norm": 0.9848728775978088,
|
| 396 |
+
"learning_rate": 6.751849846598087e-05,
|
| 397 |
+
"loss": 3.3111,
|
| 398 |
+
"step": 9000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9745533297238765,
|
| 402 |
+
"eval_loss": 2.7730352878570557,
|
| 403 |
+
"eval_runtime": 87.8671,
|
| 404 |
+
"eval_samples_per_second": 112.101,
|
| 405 |
+
"eval_steps_per_second": 7.011,
|
| 406 |
+
"step": 9000
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.0016242555495398,
|
| 410 |
+
"grad_norm": 1.0229711532592773,
|
| 411 |
+
"learning_rate": 6.661613427179209e-05,
|
| 412 |
+
"loss": 3.3042,
|
| 413 |
+
"step": 9250
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.028695181375203,
|
| 417 |
+
"grad_norm": 1.1280988454818726,
|
| 418 |
+
"learning_rate": 6.571377007760331e-05,
|
| 419 |
+
"loss": 3.3033,
|
| 420 |
+
"step": 9500
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.028695181375203,
|
| 424 |
+
"eval_loss": 2.7774875164031982,
|
| 425 |
+
"eval_runtime": 85.5937,
|
| 426 |
+
"eval_samples_per_second": 115.079,
|
| 427 |
+
"eval_steps_per_second": 7.197,
|
| 428 |
+
"step": 9500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.0557661072008662,
|
| 432 |
+
"grad_norm": 1.2061575651168823,
|
| 433 |
+
"learning_rate": 6.481140588341454e-05,
|
| 434 |
+
"loss": 3.2889,
|
| 435 |
+
"step": 9750
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 1.0828370330265296,
|
| 439 |
+
"grad_norm": 1.1995490789413452,
|
| 440 |
+
"learning_rate": 6.390904168922578e-05,
|
| 441 |
+
"loss": 3.2796,
|
| 442 |
+
"step": 10000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0828370330265296,
|
| 446 |
+
"eval_loss": 2.773512363433838,
|
| 447 |
+
"eval_runtime": 86.6895,
|
| 448 |
+
"eval_samples_per_second": 113.624,
|
| 449 |
+
"eval_steps_per_second": 7.106,
|
| 450 |
+
"step": 10000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.1099079588521927,
|
| 454 |
+
"grad_norm": 1.0674035549163818,
|
| 455 |
+
"learning_rate": 6.3006677495037e-05,
|
| 456 |
+
"loss": 3.2893,
|
| 457 |
+
"step": 10250
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.136978884677856,
|
| 461 |
+
"grad_norm": 1.193318486213684,
|
| 462 |
+
"learning_rate": 6.210431330084822e-05,
|
| 463 |
+
"loss": 3.2957,
|
| 464 |
+
"step": 10500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.136978884677856,
|
| 468 |
+
"eval_loss": 2.768921375274658,
|
| 469 |
+
"eval_runtime": 86.6142,
|
| 470 |
+
"eval_samples_per_second": 113.723,
|
| 471 |
+
"eval_steps_per_second": 7.112,
|
| 472 |
+
"step": 10500
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.1640498105035193,
|
| 476 |
+
"grad_norm": 1.0185372829437256,
|
| 477 |
+
"learning_rate": 6.120194910665945e-05,
|
| 478 |
+
"loss": 3.2763,
|
| 479 |
+
"step": 10750
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.1911207363291825,
|
| 483 |
+
"grad_norm": 0.9858824014663696,
|
| 484 |
+
"learning_rate": 6.029958491247067e-05,
|
| 485 |
+
"loss": 3.2935,
|
| 486 |
+
"step": 11000
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.1911207363291825,
|
| 490 |
+
"eval_loss": 2.7698659896850586,
|
| 491 |
+
"eval_runtime": 86.7972,
|
| 492 |
+
"eval_samples_per_second": 113.483,
|
| 493 |
+
"eval_steps_per_second": 7.097,
|
| 494 |
+
"step": 11000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.2181916621548456,
|
| 498 |
+
"grad_norm": 1.0175237655639648,
|
| 499 |
+
"learning_rate": 5.9397220718281895e-05,
|
| 500 |
+
"loss": 3.2765,
|
| 501 |
+
"step": 11250
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.245262587980509,
|
| 505 |
+
"grad_norm": 0.9483986496925354,
|
| 506 |
+
"learning_rate": 5.849485652409312e-05,
|
| 507 |
+
"loss": 3.2714,
|
| 508 |
+
"step": 11500
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.245262587980509,
|
| 512 |
+
"eval_loss": 2.758622884750366,
|
| 513 |
+
"eval_runtime": 87.0052,
|
| 514 |
+
"eval_samples_per_second": 113.212,
|
| 515 |
+
"eval_steps_per_second": 7.08,
|
| 516 |
+
"step": 11500
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 1.2723335138061722,
|
| 520 |
+
"grad_norm": 1.1170575618743896,
|
| 521 |
+
"learning_rate": 5.759249232990436e-05,
|
| 522 |
+
"loss": 3.2834,
|
| 523 |
+
"step": 11750
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 1.2994044396318354,
|
| 527 |
+
"grad_norm": 1.1345117092132568,
|
| 528 |
+
"learning_rate": 5.669012813571558e-05,
|
| 529 |
+
"loss": 3.2633,
|
| 530 |
+
"step": 12000
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 1.2994044396318354,
|
| 534 |
+
"eval_loss": 2.760470390319824,
|
| 535 |
+
"eval_runtime": 87.1216,
|
| 536 |
+
"eval_samples_per_second": 113.06,
|
| 537 |
+
"eval_steps_per_second": 7.071,
|
| 538 |
+
"step": 12000
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.3264753654574988,
|
| 542 |
+
"grad_norm": 0.9705148339271545,
|
| 543 |
+
"learning_rate": 5.5787763941526805e-05,
|
| 544 |
+
"loss": 3.2687,
|
| 545 |
+
"step": 12250
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.353546291283162,
|
| 549 |
+
"grad_norm": 1.2483344078063965,
|
| 550 |
+
"learning_rate": 5.488539974733803e-05,
|
| 551 |
+
"loss": 3.2641,
|
| 552 |
+
"step": 12500
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.353546291283162,
|
| 556 |
+
"eval_loss": 2.750833034515381,
|
| 557 |
+
"eval_runtime": 87.0498,
|
| 558 |
+
"eval_samples_per_second": 113.154,
|
| 559 |
+
"eval_steps_per_second": 7.076,
|
| 560 |
+
"step": 12500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.3806172171088251,
|
| 564 |
+
"grad_norm": 0.9966532588005066,
|
| 565 |
+
"learning_rate": 5.398303555314925e-05,
|
| 566 |
+
"loss": 3.2619,
|
| 567 |
+
"step": 12750
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.4076881429344883,
|
| 571 |
+
"grad_norm": 1.0225645303726196,
|
| 572 |
+
"learning_rate": 5.3080671358960476e-05,
|
| 573 |
+
"loss": 3.2647,
|
| 574 |
+
"step": 13000
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.4076881429344883,
|
| 578 |
+
"eval_loss": 2.745871067047119,
|
| 579 |
+
"eval_runtime": 87.2201,
|
| 580 |
+
"eval_samples_per_second": 112.933,
|
| 581 |
+
"eval_steps_per_second": 7.063,
|
| 582 |
+
"step": 13000
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.4347590687601515,
|
| 586 |
+
"grad_norm": 0.9746631383895874,
|
| 587 |
+
"learning_rate": 5.21783071647717e-05,
|
| 588 |
+
"loss": 3.2489,
|
| 589 |
+
"step": 13250
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.4618299945858149,
|
| 593 |
+
"grad_norm": 1.0829297304153442,
|
| 594 |
+
"learning_rate": 5.1275942970582924e-05,
|
| 595 |
+
"loss": 3.2456,
|
| 596 |
+
"step": 13500
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.4618299945858149,
|
| 600 |
+
"eval_loss": 2.743380546569824,
|
| 601 |
+
"eval_runtime": 86.2917,
|
| 602 |
+
"eval_samples_per_second": 114.148,
|
| 603 |
+
"eval_steps_per_second": 4.392,
|
| 604 |
+
"step": 13500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.488900920411478,
|
| 608 |
+
"grad_norm": 1.0580862760543823,
|
| 609 |
+
"learning_rate": 5.037357877639416e-05,
|
| 610 |
+
"loss": 3.2568,
|
| 611 |
+
"step": 13750
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.5159718462371412,
|
| 615 |
+
"grad_norm": 0.9426606893539429,
|
| 616 |
+
"learning_rate": 4.9471214582205386e-05,
|
| 617 |
+
"loss": 3.2413,
|
| 618 |
+
"step": 14000
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.5159718462371412,
|
| 622 |
+
"eval_loss": 2.7432825565338135,
|
| 623 |
+
"eval_runtime": 86.3742,
|
| 624 |
+
"eval_samples_per_second": 114.039,
|
| 625 |
+
"eval_steps_per_second": 4.388,
|
| 626 |
+
"step": 14000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.5430427720628046,
|
| 630 |
+
"grad_norm": 1.001214623451233,
|
| 631 |
+
"learning_rate": 4.856885038801661e-05,
|
| 632 |
+
"loss": 3.2606,
|
| 633 |
+
"step": 14250
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.5701136978884678,
|
| 637 |
+
"grad_norm": 1.0681531429290771,
|
| 638 |
+
"learning_rate": 4.7666486193827834e-05,
|
| 639 |
+
"loss": 3.2395,
|
| 640 |
+
"step": 14500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.5701136978884678,
|
| 644 |
+
"eval_loss": 2.7390074729919434,
|
| 645 |
+
"eval_runtime": 86.2807,
|
| 646 |
+
"eval_samples_per_second": 114.162,
|
| 647 |
+
"eval_steps_per_second": 4.393,
|
| 648 |
+
"step": 14500
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.597184623714131,
|
| 652 |
+
"grad_norm": 1.1323778629302979,
|
| 653 |
+
"learning_rate": 4.676412199963906e-05,
|
| 654 |
+
"loss": 3.2469,
|
| 655 |
+
"step": 14750
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.6242555495397943,
|
| 659 |
+
"grad_norm": 1.0971120595932007,
|
| 660 |
+
"learning_rate": 4.586175780545028e-05,
|
| 661 |
+
"loss": 3.2362,
|
| 662 |
+
"step": 15000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.6242555495397943,
|
| 666 |
+
"eval_loss": 2.742565393447876,
|
| 667 |
+
"eval_runtime": 86.3237,
|
| 668 |
+
"eval_samples_per_second": 114.105,
|
| 669 |
+
"eval_steps_per_second": 4.39,
|
| 670 |
+
"step": 15000
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.6513264753654575,
|
| 674 |
+
"grad_norm": 0.8698033094406128,
|
| 675 |
+
"learning_rate": 4.4959393611261505e-05,
|
| 676 |
+
"loss": 3.2348,
|
| 677 |
+
"step": 15250
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 1.6783974011911207,
|
| 681 |
+
"grad_norm": 1.107132077217102,
|
| 682 |
+
"learning_rate": 4.405702941707273e-05,
|
| 683 |
+
"loss": 3.2453,
|
| 684 |
+
"step": 15500
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.6783974011911207,
|
| 688 |
+
"eval_loss": 2.73408842086792,
|
| 689 |
+
"eval_runtime": 86.2136,
|
| 690 |
+
"eval_samples_per_second": 114.251,
|
| 691 |
+
"eval_steps_per_second": 4.396,
|
| 692 |
+
"step": 15500
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.705468327016784,
|
| 696 |
+
"grad_norm": 1.015751838684082,
|
| 697 |
+
"learning_rate": 4.315466522288395e-05,
|
| 698 |
+
"loss": 3.242,
|
| 699 |
+
"step": 15750
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.7325392528424473,
|
| 703 |
+
"grad_norm": 1.1737513542175293,
|
| 704 |
+
"learning_rate": 4.2252301028695184e-05,
|
| 705 |
+
"loss": 3.2309,
|
| 706 |
+
"step": 16000
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.7325392528424473,
|
| 710 |
+
"eval_loss": 2.729858160018921,
|
| 711 |
+
"eval_runtime": 86.1955,
|
| 712 |
+
"eval_samples_per_second": 114.275,
|
| 713 |
+
"eval_steps_per_second": 4.397,
|
| 714 |
+
"step": 16000
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.7596101786681104,
|
| 718 |
+
"grad_norm": 1.0125904083251953,
|
| 719 |
+
"learning_rate": 4.134993683450641e-05,
|
| 720 |
+
"loss": 3.2313,
|
| 721 |
+
"step": 16250
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.7866811044937738,
|
| 725 |
+
"grad_norm": 1.0134389400482178,
|
| 726 |
+
"learning_rate": 4.044757264031763e-05,
|
| 727 |
+
"loss": 3.227,
|
| 728 |
+
"step": 16500
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.7866811044937738,
|
| 732 |
+
"eval_loss": 2.7314817905426025,
|
| 733 |
+
"eval_runtime": 90.6312,
|
| 734 |
+
"eval_samples_per_second": 108.682,
|
| 735 |
+
"eval_steps_per_second": 3.63,
|
| 736 |
+
"step": 16500
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.8137520303194368,
|
| 740 |
+
"grad_norm": 0.9060974717140198,
|
| 741 |
+
"learning_rate": 3.954520844612886e-05,
|
| 742 |
+
"loss": 3.2259,
|
| 743 |
+
"step": 16750
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.8408229561451002,
|
| 747 |
+
"grad_norm": 1.01957368850708,
|
| 748 |
+
"learning_rate": 3.8642844251940086e-05,
|
| 749 |
+
"loss": 3.2254,
|
| 750 |
+
"step": 17000
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.8408229561451002,
|
| 754 |
+
"eval_loss": 2.7291440963745117,
|
| 755 |
+
"eval_runtime": 90.6696,
|
| 756 |
+
"eval_samples_per_second": 108.636,
|
| 757 |
+
"eval_steps_per_second": 3.629,
|
| 758 |
+
"step": 17000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.8678938819707636,
|
| 762 |
+
"grad_norm": 0.9996930360794067,
|
| 763 |
+
"learning_rate": 3.774048005775131e-05,
|
| 764 |
+
"loss": 3.2189,
|
| 765 |
+
"step": 17250
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.8949648077964265,
|
| 769 |
+
"grad_norm": 1.0718858242034912,
|
| 770 |
+
"learning_rate": 3.6838115863562534e-05,
|
| 771 |
+
"loss": 3.208,
|
| 772 |
+
"step": 17500
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.8949648077964265,
|
| 776 |
+
"eval_loss": 2.724837303161621,
|
| 777 |
+
"eval_runtime": 90.2749,
|
| 778 |
+
"eval_samples_per_second": 109.111,
|
| 779 |
+
"eval_steps_per_second": 3.644,
|
| 780 |
+
"step": 17500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.92203573362209,
|
| 784 |
+
"grad_norm": 1.085183024406433,
|
| 785 |
+
"learning_rate": 3.5935751669373765e-05,
|
| 786 |
+
"loss": 3.2267,
|
| 787 |
+
"step": 17750
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.949106659447753,
|
| 791 |
+
"grad_norm": 1.1373517513275146,
|
| 792 |
+
"learning_rate": 3.503338747518499e-05,
|
| 793 |
+
"loss": 3.2044,
|
| 794 |
+
"step": 18000
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.949106659447753,
|
| 798 |
+
"eval_loss": 2.7174291610717773,
|
| 799 |
+
"eval_runtime": 90.6301,
|
| 800 |
+
"eval_samples_per_second": 108.684,
|
| 801 |
+
"eval_steps_per_second": 3.63,
|
| 802 |
+
"step": 18000
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.9761775852734162,
|
| 806 |
+
"grad_norm": 1.235625147819519,
|
| 807 |
+
"learning_rate": 3.413102328099621e-05,
|
| 808 |
+
"loss": 3.2136,
|
| 809 |
+
"step": 18250
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 2.0032485110990796,
|
| 813 |
+
"grad_norm": 1.1179406642913818,
|
| 814 |
+
"learning_rate": 3.322865908680744e-05,
|
| 815 |
+
"loss": 3.2067,
|
| 816 |
+
"step": 18500
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 2.0032485110990796,
|
| 820 |
+
"eval_loss": 2.718658447265625,
|
| 821 |
+
"eval_runtime": 90.6713,
|
| 822 |
+
"eval_samples_per_second": 108.634,
|
| 823 |
+
"eval_steps_per_second": 3.628,
|
| 824 |
+
"step": 18500
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.030319436924743,
|
| 828 |
+
"grad_norm": 1.0269511938095093,
|
| 829 |
+
"learning_rate": 3.232629489261866e-05,
|
| 830 |
+
"loss": 3.2035,
|
| 831 |
+
"step": 18750
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.057390362750406,
|
| 835 |
+
"grad_norm": 0.9589810967445374,
|
| 836 |
+
"learning_rate": 3.1423930698429885e-05,
|
| 837 |
+
"loss": 3.2099,
|
| 838 |
+
"step": 19000
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 2.057390362750406,
|
| 842 |
+
"eval_loss": 2.721846103668213,
|
| 843 |
+
"eval_runtime": 90.6473,
|
| 844 |
+
"eval_samples_per_second": 108.663,
|
| 845 |
+
"eval_steps_per_second": 3.629,
|
| 846 |
+
"step": 19000
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 2.0844612885760694,
|
| 850 |
+
"grad_norm": 1.1196249723434448,
|
| 851 |
+
"learning_rate": 3.052156650424111e-05,
|
| 852 |
+
"loss": 3.2153,
|
| 853 |
+
"step": 19250
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.1115322144017323,
|
| 857 |
+
"grad_norm": 1.0364315509796143,
|
| 858 |
+
"learning_rate": 2.961920231005234e-05,
|
| 859 |
+
"loss": 3.2019,
|
| 860 |
+
"step": 19500
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.1115322144017323,
|
| 864 |
+
"eval_loss": 2.7143051624298096,
|
| 865 |
+
"eval_runtime": 88.2104,
|
| 866 |
+
"eval_samples_per_second": 111.665,
|
| 867 |
+
"eval_steps_per_second": 3.73,
|
| 868 |
+
"step": 19500
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 2.1386031402273957,
|
| 872 |
+
"grad_norm": 1.0353423357009888,
|
| 873 |
+
"learning_rate": 2.8716838115863563e-05,
|
| 874 |
+
"loss": 3.1951,
|
| 875 |
+
"step": 19750
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 2.165674066053059,
|
| 879 |
+
"grad_norm": 1.054973840713501,
|
| 880 |
+
"learning_rate": 2.7814473921674787e-05,
|
| 881 |
+
"loss": 3.2101,
|
| 882 |
+
"step": 20000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 2.165674066053059,
|
| 886 |
+
"eval_loss": 2.721579074859619,
|
| 887 |
+
"eval_runtime": 88.2213,
|
| 888 |
+
"eval_samples_per_second": 111.651,
|
| 889 |
+
"eval_steps_per_second": 3.729,
|
| 890 |
+
"step": 20000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 2.192744991878722,
|
| 894 |
+
"grad_norm": 1.2565995454788208,
|
| 895 |
+
"learning_rate": 2.691210972748601e-05,
|
| 896 |
+
"loss": 3.2131,
|
| 897 |
+
"step": 20250
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 2.2198159177043855,
|
| 901 |
+
"grad_norm": 1.0481986999511719,
|
| 902 |
+
"learning_rate": 2.6009745533297242e-05,
|
| 903 |
+
"loss": 3.1898,
|
| 904 |
+
"step": 20500
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 2.2198159177043855,
|
| 908 |
+
"eval_loss": 2.7174248695373535,
|
| 909 |
+
"eval_runtime": 88.1012,
|
| 910 |
+
"eval_samples_per_second": 111.803,
|
| 911 |
+
"eval_steps_per_second": 3.734,
|
| 912 |
+
"step": 20500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 2.246886843530049,
|
| 916 |
+
"grad_norm": 0.9449788331985474,
|
| 917 |
+
"learning_rate": 2.5107381339108466e-05,
|
| 918 |
+
"loss": 3.191,
|
| 919 |
+
"step": 20750
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 2.273957769355712,
|
| 923 |
+
"grad_norm": 1.0603032112121582,
|
| 924 |
+
"learning_rate": 2.420501714491969e-05,
|
| 925 |
+
"loss": 3.1985,
|
| 926 |
+
"step": 21000
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 2.273957769355712,
|
| 930 |
+
"eval_loss": 2.7089896202087402,
|
| 931 |
+
"eval_runtime": 88.0285,
|
| 932 |
+
"eval_samples_per_second": 111.896,
|
| 933 |
+
"eval_steps_per_second": 3.737,
|
| 934 |
+
"step": 21000
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.301028695181375,
|
| 938 |
+
"grad_norm": 1.096023440361023,
|
| 939 |
+
"learning_rate": 2.3302652950730917e-05,
|
| 940 |
+
"loss": 3.1891,
|
| 941 |
+
"step": 21250
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.3280996210070386,
|
| 945 |
+
"grad_norm": 1.216874599456787,
|
| 946 |
+
"learning_rate": 2.2400288756542144e-05,
|
| 947 |
+
"loss": 3.1974,
|
| 948 |
+
"step": 21500
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.3280996210070386,
|
| 952 |
+
"eval_loss": 2.707638740539551,
|
| 953 |
+
"eval_runtime": 88.1174,
|
| 954 |
+
"eval_samples_per_second": 111.783,
|
| 955 |
+
"eval_steps_per_second": 3.734,
|
| 956 |
+
"step": 21500
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 2.3551705468327016,
|
| 960 |
+
"grad_norm": 1.0820770263671875,
|
| 961 |
+
"learning_rate": 2.1497924562353368e-05,
|
| 962 |
+
"loss": 3.1982,
|
| 963 |
+
"step": 21750
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.382241472658365,
|
| 967 |
+
"grad_norm": 1.223590612411499,
|
| 968 |
+
"learning_rate": 2.0595560368164592e-05,
|
| 969 |
+
"loss": 3.1779,
|
| 970 |
+
"step": 22000
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 2.382241472658365,
|
| 974 |
+
"eval_loss": 2.7064244747161865,
|
| 975 |
+
"eval_runtime": 88.1632,
|
| 976 |
+
"eval_samples_per_second": 111.725,
|
| 977 |
+
"eval_steps_per_second": 3.732,
|
| 978 |
+
"step": 22000
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 2.4093123984840283,
|
| 982 |
+
"grad_norm": 1.0039288997650146,
|
| 983 |
+
"learning_rate": 1.9693196173975816e-05,
|
| 984 |
+
"loss": 3.1909,
|
| 985 |
+
"step": 22250
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 2.4363833243096913,
|
| 989 |
+
"grad_norm": 1.0038957595825195,
|
| 990 |
+
"learning_rate": 1.8790831979787043e-05,
|
| 991 |
+
"loss": 3.1904,
|
| 992 |
+
"step": 22500
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 2.4363833243096913,
|
| 996 |
+
"eval_loss": 2.706280469894409,
|
| 997 |
+
"eval_runtime": 88.2453,
|
| 998 |
+
"eval_samples_per_second": 111.621,
|
| 999 |
+
"eval_steps_per_second": 3.728,
|
| 1000 |
+
"step": 22500
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 2.4634542501353547,
|
| 1004 |
+
"grad_norm": 1.0922913551330566,
|
| 1005 |
+
"learning_rate": 1.7888467785598267e-05,
|
| 1006 |
+
"loss": 3.194,
|
| 1007 |
+
"step": 22750
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.490525175961018,
|
| 1011 |
+
"grad_norm": 1.0635244846343994,
|
| 1012 |
+
"learning_rate": 1.6986103591409495e-05,
|
| 1013 |
+
"loss": 3.1961,
|
| 1014 |
+
"step": 23000
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.490525175961018,
|
| 1018 |
+
"eval_loss": 2.708953619003296,
|
| 1019 |
+
"eval_runtime": 88.2167,
|
| 1020 |
+
"eval_samples_per_second": 111.657,
|
| 1021 |
+
"eval_steps_per_second": 3.729,
|
| 1022 |
+
"step": 23000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 2.517596101786681,
|
| 1026 |
+
"grad_norm": 1.0230063199996948,
|
| 1027 |
+
"learning_rate": 1.608373939722072e-05,
|
| 1028 |
+
"loss": 3.1841,
|
| 1029 |
+
"step": 23250
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 2.5446670276123444,
|
| 1033 |
+
"grad_norm": 1.0655814409255981,
|
| 1034 |
+
"learning_rate": 1.5181375203031944e-05,
|
| 1035 |
+
"loss": 3.2059,
|
| 1036 |
+
"step": 23500
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 2.5446670276123444,
|
| 1040 |
+
"eval_loss": 2.7019364833831787,
|
| 1041 |
+
"eval_runtime": 88.9998,
|
| 1042 |
+
"eval_samples_per_second": 110.674,
|
| 1043 |
+
"eval_steps_per_second": 3.697,
|
| 1044 |
+
"step": 23500
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.5717379534380074,
|
| 1048 |
+
"grad_norm": 1.1926429271697998,
|
| 1049 |
+
"learning_rate": 1.4279011008843171e-05,
|
| 1050 |
+
"loss": 3.187,
|
| 1051 |
+
"step": 23750
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 2.5988088792636708,
|
| 1055 |
+
"grad_norm": 1.09559166431427,
|
| 1056 |
+
"learning_rate": 1.3376646814654395e-05,
|
| 1057 |
+
"loss": 3.1913,
|
| 1058 |
+
"step": 24000
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 2.5988088792636708,
|
| 1062 |
+
"eval_loss": 2.7079548835754395,
|
| 1063 |
+
"eval_runtime": 88.7957,
|
| 1064 |
+
"eval_samples_per_second": 110.929,
|
| 1065 |
+
"eval_steps_per_second": 3.705,
|
| 1066 |
+
"step": 24000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 2.625879805089334,
|
| 1070 |
+
"grad_norm": 1.0533467531204224,
|
| 1071 |
+
"learning_rate": 1.2474282620465621e-05,
|
| 1072 |
+
"loss": 3.189,
|
| 1073 |
+
"step": 24250
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.6529507309149976,
|
| 1077 |
+
"grad_norm": 1.0677307844161987,
|
| 1078 |
+
"learning_rate": 1.1571918426276845e-05,
|
| 1079 |
+
"loss": 3.1908,
|
| 1080 |
+
"step": 24500
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 2.6529507309149976,
|
| 1084 |
+
"eval_loss": 2.7036521434783936,
|
| 1085 |
+
"eval_runtime": 89.1696,
|
| 1086 |
+
"eval_samples_per_second": 110.464,
|
| 1087 |
+
"eval_steps_per_second": 3.69,
|
| 1088 |
+
"step": 24500
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.6800216567406605,
|
| 1092 |
+
"grad_norm": 1.1612430810928345,
|
| 1093 |
+
"learning_rate": 1.066955423208807e-05,
|
| 1094 |
+
"loss": 3.2007,
|
| 1095 |
+
"step": 24750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.707092582566324,
|
| 1099 |
+
"grad_norm": 0.9524983763694763,
|
| 1100 |
+
"learning_rate": 9.767190037899298e-06,
|
| 1101 |
+
"loss": 3.1934,
|
| 1102 |
+
"step": 25000
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.707092582566324,
|
| 1106 |
+
"eval_loss": 2.704775333404541,
|
| 1107 |
+
"eval_runtime": 89.0167,
|
| 1108 |
+
"eval_samples_per_second": 110.653,
|
| 1109 |
+
"eval_steps_per_second": 3.696,
|
| 1110 |
+
"step": 25000
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 2.734163508391987,
|
| 1114 |
+
"grad_norm": 1.1597352027893066,
|
| 1115 |
+
"learning_rate": 8.864825843710522e-06,
|
| 1116 |
+
"loss": 3.1883,
|
| 1117 |
+
"step": 25250
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 2.7612344342176502,
|
| 1121 |
+
"grad_norm": 1.1218962669372559,
|
| 1122 |
+
"learning_rate": 7.962461649521747e-06,
|
| 1123 |
+
"loss": 3.1889,
|
| 1124 |
+
"step": 25500
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 2.7612344342176502,
|
| 1128 |
+
"eval_loss": 2.701597213745117,
|
| 1129 |
+
"eval_runtime": 89.0323,
|
| 1130 |
+
"eval_samples_per_second": 110.634,
|
| 1131 |
+
"eval_steps_per_second": 3.695,
|
| 1132 |
+
"step": 25500
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 2.7883053600433136,
|
| 1136 |
+
"grad_norm": 1.1327321529388428,
|
| 1137 |
+
"learning_rate": 7.060097455332973e-06,
|
| 1138 |
+
"loss": 3.1874,
|
| 1139 |
+
"step": 25750
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 2.8153762858689766,
|
| 1143 |
+
"grad_norm": 1.2180895805358887,
|
| 1144 |
+
"learning_rate": 6.157733261144198e-06,
|
| 1145 |
+
"loss": 3.1883,
|
| 1146 |
+
"step": 26000
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 2.8153762858689766,
|
| 1150 |
+
"eval_loss": 2.703519105911255,
|
| 1151 |
+
"eval_runtime": 88.8695,
|
| 1152 |
+
"eval_samples_per_second": 110.837,
|
| 1153 |
+
"eval_steps_per_second": 3.702,
|
| 1154 |
+
"step": 26000
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.84244721169464,
|
| 1158 |
+
"grad_norm": 0.9983986020088196,
|
| 1159 |
+
"learning_rate": 5.2553690669554234e-06,
|
| 1160 |
+
"loss": 3.1859,
|
| 1161 |
+
"step": 26250
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 2.869518137520303,
|
| 1165 |
+
"grad_norm": 1.367304801940918,
|
| 1166 |
+
"learning_rate": 4.353004872766649e-06,
|
| 1167 |
+
"loss": 3.181,
|
| 1168 |
+
"step": 26500
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 2.869518137520303,
|
| 1172 |
+
"eval_loss": 2.7028160095214844,
|
| 1173 |
+
"eval_runtime": 89.1113,
|
| 1174 |
+
"eval_samples_per_second": 110.536,
|
| 1175 |
+
"eval_steps_per_second": 3.692,
|
| 1176 |
+
"step": 26500
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 2.8965890633459663,
|
| 1180 |
+
"grad_norm": 1.0023523569107056,
|
| 1181 |
+
"learning_rate": 3.450640678577874e-06,
|
| 1182 |
+
"loss": 3.1741,
|
| 1183 |
+
"step": 26750
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 2.9236599891716297,
|
| 1187 |
+
"grad_norm": 1.055477499961853,
|
| 1188 |
+
"learning_rate": 2.5482764843890995e-06,
|
| 1189 |
+
"loss": 3.1904,
|
| 1190 |
+
"step": 27000
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 2.9236599891716297,
|
| 1194 |
+
"eval_loss": 2.702355146408081,
|
| 1195 |
+
"eval_runtime": 89.0095,
|
| 1196 |
+
"eval_samples_per_second": 110.662,
|
| 1197 |
+
"eval_steps_per_second": 3.696,
|
| 1198 |
+
"step": 27000
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 2.950730914997293,
|
| 1202 |
+
"grad_norm": 1.1611043214797974,
|
| 1203 |
+
"learning_rate": 1.645912290200325e-06,
|
| 1204 |
+
"loss": 3.1703,
|
| 1205 |
+
"step": 27250
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 2.977801840822956,
|
| 1209 |
+
"grad_norm": 1.0057547092437744,
|
| 1210 |
+
"learning_rate": 7.435480960115503e-07,
|
| 1211 |
+
"loss": 3.1838,
|
| 1212 |
+
"step": 27500
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 2.977801840822956,
|
| 1216 |
+
"eval_loss": 2.7012827396392822,
|
| 1217 |
+
"eval_runtime": 89.1882,
|
| 1218 |
+
"eval_samples_per_second": 110.441,
|
| 1219 |
+
"eval_steps_per_second": 3.689,
|
| 1220 |
+
"step": 27500
|
| 1221 |
+
}
|
| 1222 |
+
],
|
| 1223 |
+
"logging_steps": 250,
|
| 1224 |
+
"max_steps": 27705,
|
| 1225 |
+
"num_input_tokens_seen": 0,
|
| 1226 |
+
"num_train_epochs": 3,
|
| 1227 |
+
"save_steps": 1000,
|
| 1228 |
+
"stateful_callbacks": {
|
| 1229 |
+
"TrainerControl": {
|
| 1230 |
+
"args": {
|
| 1231 |
+
"should_epoch_stop": false,
|
| 1232 |
+
"should_evaluate": false,
|
| 1233 |
+
"should_log": false,
|
| 1234 |
+
"should_save": true,
|
| 1235 |
+
"should_training_stop": true
|
| 1236 |
+
},
|
| 1237 |
+
"attributes": {}
|
| 1238 |
+
}
|
| 1239 |
+
},
|
| 1240 |
+
"total_flos": 1.0002326224896e+16,
|
| 1241 |
+
"train_batch_size": 16,
|
| 1242 |
+
"trial_name": null,
|
| 1243 |
+
"trial_params": null
|
| 1244 |
+
}
|