Upload checkpoint-3894
Browse files
checkpoint-3894/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e655ca3280ae7cf497e0d5e52e69637dfe9716ac94aadf78cf38f033e6d14e05
|
| 3 |
+
size 16389
|
checkpoint-3894/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0618be5d24ad595cd8c17d9f6e1f935832c234d211a8f5d643d1d2b00db9bc34
|
| 3 |
+
size 1465
|
checkpoint-3894/trainer_state.json
ADDED
|
@@ -0,0 +1,2595 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 3894,
|
| 3 |
+
"best_metric": 2.92698335647583,
|
| 4 |
+
"best_model_checkpoint": "outputs/chess-explainer/checkpoint-3894",
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 3894,
|
| 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,
|
| 14 |
+
"step": 0,
|
| 15 |
+
"train/loss_commentary": 17.28484344482422
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0,
|
| 19 |
+
"step": 0,
|
| 20 |
+
"train/loss_grounding": 16.196508407592773
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.0007704160246533128,
|
| 24 |
+
"grad_norm": 375.1965026855469,
|
| 25 |
+
"learning_rate": 0.0,
|
| 26 |
+
"loss": 16.959257125854492,
|
| 27 |
+
"step": 1
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01926040061633282,
|
| 31 |
+
"grad_norm": 63.75334930419922,
|
| 32 |
+
"learning_rate": 1.230769230769231e-05,
|
| 33 |
+
"loss": 13.329045613606771,
|
| 34 |
+
"step": 25
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.01926040061633282,
|
| 38 |
+
"step": 25,
|
| 39 |
+
"train/loss_commentary": 10.003331184387207
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.01926040061633282,
|
| 43 |
+
"step": 25,
|
| 44 |
+
"train/loss_grounding": 8.758746147155762
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03852080123266564,
|
| 48 |
+
"grad_norm": 14.34142780303955,
|
| 49 |
+
"learning_rate": 2.512820512820513e-05,
|
| 50 |
+
"loss": 7.129249267578125,
|
| 51 |
+
"step": 50
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03852080123266564,
|
| 55 |
+
"step": 50,
|
| 56 |
+
"train/loss_commentary": 5.0110392570495605
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.03852080123266564,
|
| 60 |
+
"step": 50,
|
| 61 |
+
"train/loss_grounding": 3.773664951324463
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.05778120184899846,
|
| 65 |
+
"grad_norm": 10.976011276245117,
|
| 66 |
+
"learning_rate": 3.794871794871795e-05,
|
| 67 |
+
"loss": 4.540987243652344,
|
| 68 |
+
"step": 75
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.05778120184899846,
|
| 72 |
+
"step": 75,
|
| 73 |
+
"train/loss_commentary": 4.010482311248779
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.05778120184899846,
|
| 77 |
+
"step": 75,
|
| 78 |
+
"train/loss_grounding": 0.6454176902770996
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 0.07704160246533127,
|
| 82 |
+
"grad_norm": 10.42539119720459,
|
| 83 |
+
"learning_rate": 5.0769230769230766e-05,
|
| 84 |
+
"loss": 3.945438232421875,
|
| 85 |
+
"step": 100
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 0.07704160246533127,
|
| 89 |
+
"step": 100,
|
| 90 |
+
"train/loss_commentary": 3.898587703704834
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.07704160246533127,
|
| 94 |
+
"step": 100,
|
| 95 |
+
"train/loss_grounding": 1.0235992670059204
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.0963020030816641,
|
| 99 |
+
"grad_norm": 9.775985717773438,
|
| 100 |
+
"learning_rate": 6.358974358974359e-05,
|
| 101 |
+
"loss": 3.758442687988281,
|
| 102 |
+
"step": 125
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.0963020030816641,
|
| 106 |
+
"step": 125,
|
| 107 |
+
"train/loss_commentary": 3.6387410163879395
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.0963020030816641,
|
| 111 |
+
"step": 125,
|
| 112 |
+
"train/loss_grounding": 0.6389902234077454
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.11556240369799692,
|
| 116 |
+
"grad_norm": 10.101550102233887,
|
| 117 |
+
"learning_rate": 7.641025641025641e-05,
|
| 118 |
+
"loss": 3.5899114990234375,
|
| 119 |
+
"step": 150
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.11556240369799692,
|
| 123 |
+
"step": 150,
|
| 124 |
+
"train/loss_commentary": 3.8033623695373535
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.11556240369799692,
|
| 128 |
+
"step": 150,
|
| 129 |
+
"train/loss_grounding": 0.4842790365219116
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.13482280431432975,
|
| 133 |
+
"grad_norm": 8.930421829223633,
|
| 134 |
+
"learning_rate": 8.923076923076924e-05,
|
| 135 |
+
"loss": 3.533555908203125,
|
| 136 |
+
"step": 175
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.13482280431432975,
|
| 140 |
+
"step": 175,
|
| 141 |
+
"train/loss_commentary": 3.418592691421509
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.13482280431432975,
|
| 145 |
+
"step": 175,
|
| 146 |
+
"train/loss_grounding": 0.45324766635894775
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.15408320493066255,
|
| 150 |
+
"grad_norm": 8.682564735412598,
|
| 151 |
+
"learning_rate": 0.00010205128205128205,
|
| 152 |
+
"loss": 3.473717041015625,
|
| 153 |
+
"step": 200
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.15408320493066255,
|
| 157 |
+
"step": 200,
|
| 158 |
+
"train/loss_commentary": 3.751358985900879
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.15408320493066255,
|
| 162 |
+
"step": 200,
|
| 163 |
+
"train/loss_grounding": 1.2230968475341797
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.17334360554699538,
|
| 167 |
+
"grad_norm": 8.672431945800781,
|
| 168 |
+
"learning_rate": 0.00011487179487179487,
|
| 169 |
+
"loss": 3.4303240966796875,
|
| 170 |
+
"step": 225
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.17334360554699538,
|
| 174 |
+
"step": 225,
|
| 175 |
+
"train/loss_commentary": 3.6430795192718506
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.17334360554699538,
|
| 179 |
+
"step": 225,
|
| 180 |
+
"train/loss_grounding": 0.13306836783885956
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.1926040061633282,
|
| 184 |
+
"grad_norm": 7.599946975708008,
|
| 185 |
+
"learning_rate": 0.0001276923076923077,
|
| 186 |
+
"loss": 3.370179443359375,
|
| 187 |
+
"step": 250
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.1926040061633282,
|
| 191 |
+
"step": 250,
|
| 192 |
+
"train/loss_commentary": 3.0655417442321777
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.1926040061633282,
|
| 196 |
+
"step": 250,
|
| 197 |
+
"train/loss_grounding": 0.16639356315135956
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.211864406779661,
|
| 201 |
+
"grad_norm": 7.84333610534668,
|
| 202 |
+
"learning_rate": 0.0001405128205128205,
|
| 203 |
+
"loss": 3.3502032470703127,
|
| 204 |
+
"step": 275
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.211864406779661,
|
| 208 |
+
"step": 275,
|
| 209 |
+
"train/loss_commentary": 3.4781105518341064
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.211864406779661,
|
| 213 |
+
"step": 275,
|
| 214 |
+
"train/loss_grounding": 0.16037797927856445
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.23112480739599384,
|
| 218 |
+
"grad_norm": 8.064518928527832,
|
| 219 |
+
"learning_rate": 0.00015333333333333334,
|
| 220 |
+
"loss": 3.3035980224609376,
|
| 221 |
+
"step": 300
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.23112480739599384,
|
| 225 |
+
"step": 300,
|
| 226 |
+
"train/loss_commentary": 3.5626707077026367
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.25038520801232667,
|
| 230 |
+
"grad_norm": 6.348459243774414,
|
| 231 |
+
"learning_rate": 0.00016615384615384617,
|
| 232 |
+
"loss": 3.298873291015625,
|
| 233 |
+
"step": 325
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.25038520801232667,
|
| 237 |
+
"step": 325,
|
| 238 |
+
"train/loss_commentary": 2.9077095985412598
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 0.25038520801232667,
|
| 242 |
+
"step": 325,
|
| 243 |
+
"train/loss_grounding": 0.613104522228241
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.2696456086286595,
|
| 247 |
+
"grad_norm": 7.342203617095947,
|
| 248 |
+
"learning_rate": 0.00017897435897435898,
|
| 249 |
+
"loss": 3.2783953857421877,
|
| 250 |
+
"step": 350
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.2696456086286595,
|
| 254 |
+
"step": 350,
|
| 255 |
+
"train/loss_commentary": 3.4672110080718994
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.2696456086286595,
|
| 259 |
+
"step": 350,
|
| 260 |
+
"train/loss_grounding": 0.5436829328536987
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.2889060092449923,
|
| 264 |
+
"grad_norm": 6.451893329620361,
|
| 265 |
+
"learning_rate": 0.00019179487179487182,
|
| 266 |
+
"loss": 3.2628683471679687,
|
| 267 |
+
"step": 375
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.2889060092449923,
|
| 271 |
+
"step": 375,
|
| 272 |
+
"train/loss_commentary": 3.084796667098999
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.2889060092449923,
|
| 276 |
+
"step": 375,
|
| 277 |
+
"train/loss_grounding": 0.7972221970558167
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.3081664098613251,
|
| 281 |
+
"grad_norm": 7.1052374839782715,
|
| 282 |
+
"learning_rate": 0.0001999997478243634,
|
| 283 |
+
"loss": 3.2742852783203125,
|
| 284 |
+
"step": 400
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.3081664098613251,
|
| 288 |
+
"step": 400,
|
| 289 |
+
"train/loss_commentary": 3.220027208328247
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.3081664098613251,
|
| 293 |
+
"step": 400,
|
| 294 |
+
"train/loss_grounding": 0.3418666422367096
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.3274268104776579,
|
| 298 |
+
"grad_norm": 6.784857749938965,
|
| 299 |
+
"learning_rate": 0.0001999964010690141,
|
| 300 |
+
"loss": 3.2773907470703123,
|
| 301 |
+
"step": 425
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.3274268104776579,
|
| 305 |
+
"step": 425,
|
| 306 |
+
"train/loss_commentary": 3.7263131141662598
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.3274268104776579,
|
| 310 |
+
"step": 425,
|
| 311 |
+
"train/loss_grounding": 0.5110027194023132
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.34668721109399075,
|
| 315 |
+
"grad_norm": 6.888815402984619,
|
| 316 |
+
"learning_rate": 0.00019998916286537489,
|
| 317 |
+
"loss": 3.2511141967773436,
|
| 318 |
+
"step": 450
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.34668721109399075,
|
| 322 |
+
"step": 450,
|
| 323 |
+
"train/loss_commentary": 2.9739327430725098
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 0.34668721109399075,
|
| 327 |
+
"step": 450,
|
| 328 |
+
"train/loss_grounding": 0.5739176869392395
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.3659476117103236,
|
| 332 |
+
"grad_norm": 5.947560787200928,
|
| 333 |
+
"learning_rate": 0.00019997803349512685,
|
| 334 |
+
"loss": 3.2557708740234377,
|
| 335 |
+
"step": 475
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.3659476117103236,
|
| 339 |
+
"step": 475,
|
| 340 |
+
"train/loss_commentary": 2.7628390789031982
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.3659476117103236,
|
| 344 |
+
"step": 475,
|
| 345 |
+
"train/loss_grounding": 0.4165380299091339
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.3852080123266564,
|
| 349 |
+
"grad_norm": 5.334803104400635,
|
| 350 |
+
"learning_rate": 0.00019996301339137928,
|
| 351 |
+
"loss": 3.2507778930664064,
|
| 352 |
+
"step": 500
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.3852080123266564,
|
| 356 |
+
"step": 500,
|
| 357 |
+
"train/loss_commentary": 3.5731141567230225
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.3852080123266564,
|
| 361 |
+
"step": 500,
|
| 362 |
+
"train/loss_grounding": 0.26948994398117065
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.40446841294298924,
|
| 366 |
+
"grad_norm": 5.836182594299316,
|
| 367 |
+
"learning_rate": 0.00019994410313865278,
|
| 368 |
+
"loss": 3.190909423828125,
|
| 369 |
+
"step": 525
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.40446841294298924,
|
| 373 |
+
"step": 525,
|
| 374 |
+
"train/loss_commentary": 2.594053268432617
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.40446841294298924,
|
| 378 |
+
"step": 525,
|
| 379 |
+
"train/loss_grounding": 0.8509034514427185
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.423728813559322,
|
| 383 |
+
"grad_norm": 5.920278072357178,
|
| 384 |
+
"learning_rate": 0.00019992130347285653,
|
| 385 |
+
"loss": 3.202198486328125,
|
| 386 |
+
"step": 550
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.423728813559322,
|
| 390 |
+
"step": 550,
|
| 391 |
+
"train/loss_commentary": 3.6618714332580566
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.423728813559322,
|
| 395 |
+
"step": 550,
|
| 396 |
+
"train/loss_grounding": 1.6018493175506592
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.44298921417565484,
|
| 400 |
+
"grad_norm": 6.209907054901123,
|
| 401 |
+
"learning_rate": 0.0001998946152812597,
|
| 402 |
+
"loss": 3.1561126708984375,
|
| 403 |
+
"step": 575
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.44298921417565484,
|
| 407 |
+
"step": 575,
|
| 408 |
+
"train/loss_commentary": 3.1945908069610596
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.44298921417565484,
|
| 412 |
+
"step": 575,
|
| 413 |
+
"train/loss_grounding": 0.3114471733570099
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.4622496147919877,
|
| 417 |
+
"grad_norm": 5.610293388366699,
|
| 418 |
+
"learning_rate": 0.0001998640396024568,
|
| 419 |
+
"loss": 3.1833868408203125,
|
| 420 |
+
"step": 600
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.4622496147919877,
|
| 424 |
+
"step": 600,
|
| 425 |
+
"train/loss_commentary": 3.5386414527893066
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.4622496147919877,
|
| 429 |
+
"step": 600,
|
| 430 |
+
"train/loss_grounding": 0.07982733845710754
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.4815100154083205,
|
| 434 |
+
"grad_norm": 5.260606288909912,
|
| 435 |
+
"learning_rate": 0.00019982957762632739,
|
| 436 |
+
"loss": 3.1605029296875,
|
| 437 |
+
"step": 625
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.4815100154083205,
|
| 441 |
+
"step": 625,
|
| 442 |
+
"train/loss_commentary": 3.175926923751831
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.4815100154083205,
|
| 446 |
+
"step": 625,
|
| 447 |
+
"train/loss_grounding": 0.26400333642959595
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.5007704160246533,
|
| 451 |
+
"grad_norm": 6.099313259124756,
|
| 452 |
+
"learning_rate": 0.00019979123069398976,
|
| 453 |
+
"loss": 3.1203402709960937,
|
| 454 |
+
"step": 650
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 0.5007704160246533,
|
| 458 |
+
"step": 650,
|
| 459 |
+
"train/loss_commentary": 3.215717315673828
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.5200308166409862,
|
| 463 |
+
"grad_norm": 5.70773458480835,
|
| 464 |
+
"learning_rate": 0.0001997490002977486,
|
| 465 |
+
"loss": 3.1090533447265627,
|
| 466 |
+
"step": 675
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.5200308166409862,
|
| 470 |
+
"step": 675,
|
| 471 |
+
"train/loss_commentary": 3.2540628910064697
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.5200308166409862,
|
| 475 |
+
"step": 675,
|
| 476 |
+
"train/loss_grounding": 0.2826128304004669
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.539291217257319,
|
| 480 |
+
"grad_norm": 5.444095134735107,
|
| 481 |
+
"learning_rate": 0.00019970288808103713,
|
| 482 |
+
"loss": 3.1352139282226563,
|
| 483 |
+
"step": 700
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.539291217257319,
|
| 487 |
+
"step": 700,
|
| 488 |
+
"train/loss_commentary": 2.9782607555389404
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.539291217257319,
|
| 492 |
+
"step": 700,
|
| 493 |
+
"train/loss_grounding": 0.28730711340904236
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.5585516178736518,
|
| 497 |
+
"grad_norm": 6.300431728363037,
|
| 498 |
+
"learning_rate": 0.00019965289583835297,
|
| 499 |
+
"loss": 3.0966998291015626,
|
| 500 |
+
"step": 725
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.5585516178736518,
|
| 504 |
+
"step": 725,
|
| 505 |
+
"train/loss_commentary": 2.881274461746216
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.5585516178736518,
|
| 509 |
+
"step": 725,
|
| 510 |
+
"train/loss_grounding": 0.7637962102890015
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 0.5778120184899846,
|
| 514 |
+
"grad_norm": 6.088650226593018,
|
| 515 |
+
"learning_rate": 0.00019959902551518844,
|
| 516 |
+
"loss": 3.0824234008789064,
|
| 517 |
+
"step": 750
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 0.5778120184899846,
|
| 521 |
+
"step": 750,
|
| 522 |
+
"train/loss_commentary": 2.540555953979492
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.5778120184899846,
|
| 526 |
+
"step": 750,
|
| 527 |
+
"train/loss_grounding": 0.7935182452201843
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.5970724191063174,
|
| 531 |
+
"grad_norm": 5.334967136383057,
|
| 532 |
+
"learning_rate": 0.00019954127920795474,
|
| 533 |
+
"loss": 3.1100128173828123,
|
| 534 |
+
"step": 775
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.5970724191063174,
|
| 538 |
+
"step": 775,
|
| 539 |
+
"train/loss_commentary": 3.3859386444091797
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.5970724191063174,
|
| 543 |
+
"step": 775,
|
| 544 |
+
"train/loss_grounding": 0.3784239590167999
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 0.6163328197226502,
|
| 548 |
+
"grad_norm": 5.416520118713379,
|
| 549 |
+
"learning_rate": 0.00019947965916390046,
|
| 550 |
+
"loss": 3.081431884765625,
|
| 551 |
+
"step": 800
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.6163328197226502,
|
| 555 |
+
"step": 800,
|
| 556 |
+
"train/loss_commentary": 3.4729719161987305
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.6163328197226502,
|
| 560 |
+
"step": 800,
|
| 561 |
+
"train/loss_grounding": 0.26661574840545654
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.635593220338983,
|
| 565 |
+
"grad_norm": 4.908987045288086,
|
| 566 |
+
"learning_rate": 0.000199414167781024,
|
| 567 |
+
"loss": 3.1401141357421873,
|
| 568 |
+
"step": 825
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.635593220338983,
|
| 572 |
+
"step": 825,
|
| 573 |
+
"train/loss_commentary": 3.0707640647888184
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.635593220338983,
|
| 577 |
+
"step": 825,
|
| 578 |
+
"train/loss_grounding": 1.1153666973114014
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.6548536209553159,
|
| 582 |
+
"grad_norm": 5.562082290649414,
|
| 583 |
+
"learning_rate": 0.00019934480760798042,
|
| 584 |
+
"loss": 3.0947344970703123,
|
| 585 |
+
"step": 850
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.6548536209553159,
|
| 589 |
+
"step": 850,
|
| 590 |
+
"train/loss_commentary": 3.315356492996216
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.6548536209553159,
|
| 594 |
+
"step": 850,
|
| 595 |
+
"train/loss_grounding": 0.37771421670913696
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 0.6741140215716487,
|
| 599 |
+
"grad_norm": 5.242153167724609,
|
| 600 |
+
"learning_rate": 0.00019927158134398215,
|
| 601 |
+
"loss": 3.081480712890625,
|
| 602 |
+
"step": 875
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.6741140215716487,
|
| 606 |
+
"step": 875,
|
| 607 |
+
"train/loss_commentary": 3.304847240447998
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 0.6741140215716487,
|
| 611 |
+
"step": 875,
|
| 612 |
+
"train/loss_grounding": 0.5464853048324585
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.6933744221879815,
|
| 616 |
+
"grad_norm": 4.929018020629883,
|
| 617 |
+
"learning_rate": 0.0001991944918386939,
|
| 618 |
+
"loss": 3.067822265625,
|
| 619 |
+
"step": 900
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.6933744221879815,
|
| 623 |
+
"step": 900,
|
| 624 |
+
"train/loss_commentary": 2.8900187015533447
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.7126348228043143,
|
| 628 |
+
"grad_norm": 5.884594917297363,
|
| 629 |
+
"learning_rate": 0.00019911354209212193,
|
| 630 |
+
"loss": 3.033839111328125,
|
| 631 |
+
"step": 925
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.7126348228043143,
|
| 635 |
+
"step": 925,
|
| 636 |
+
"train/loss_commentary": 3.20168137550354
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.7126348228043143,
|
| 640 |
+
"step": 925,
|
| 641 |
+
"train/loss_grounding": 0.3656123876571655
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.7318952234206472,
|
| 645 |
+
"grad_norm": 6.620251178741455,
|
| 646 |
+
"learning_rate": 0.00019902873525449706,
|
| 647 |
+
"loss": 3.0728030395507813,
|
| 648 |
+
"step": 950
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.7318952234206472,
|
| 652 |
+
"step": 950,
|
| 653 |
+
"train/loss_commentary": 3.7644381523132324
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.75115562403698,
|
| 657 |
+
"grad_norm": 5.130228042602539,
|
| 658 |
+
"learning_rate": 0.00019894007462615236,
|
| 659 |
+
"loss": 3.0683718872070314,
|
| 660 |
+
"step": 975
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.75115562403698,
|
| 664 |
+
"step": 975,
|
| 665 |
+
"train/loss_commentary": 3.1705219745635986
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.75115562403698,
|
| 669 |
+
"step": 975,
|
| 670 |
+
"train/loss_grounding": 0.06520748138427734
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 0.7704160246533128,
|
| 674 |
+
"grad_norm": 5.158539772033691,
|
| 675 |
+
"learning_rate": 0.0001988475636573945,
|
| 676 |
+
"loss": 3.0548333740234375,
|
| 677 |
+
"step": 1000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.7704160246533128,
|
| 681 |
+
"step": 1000,
|
| 682 |
+
"train/loss_commentary": 2.971273422241211
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.7704160246533128,
|
| 686 |
+
"step": 1000,
|
| 687 |
+
"train/loss_grounding": 0.8938158750534058
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 0.7896764252696457,
|
| 691 |
+
"grad_norm": 5.542895317077637,
|
| 692 |
+
"learning_rate": 0.00019875120594836955,
|
| 693 |
+
"loss": 3.05259521484375,
|
| 694 |
+
"step": 1025
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.7896764252696457,
|
| 698 |
+
"step": 1025,
|
| 699 |
+
"train/loss_commentary": 3.1235668659210205
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.7896764252696457,
|
| 703 |
+
"step": 1025,
|
| 704 |
+
"train/loss_grounding": 1.4225568771362305
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.8089368258859785,
|
| 708 |
+
"grad_norm": 5.495582103729248,
|
| 709 |
+
"learning_rate": 0.00019865100524892287,
|
| 710 |
+
"loss": 3.023851013183594,
|
| 711 |
+
"step": 1050
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.8089368258859785,
|
| 715 |
+
"step": 1050,
|
| 716 |
+
"train/loss_commentary": 2.4219141006469727
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.8089368258859785,
|
| 720 |
+
"step": 1050,
|
| 721 |
+
"train/loss_grounding": 0.78606116771698
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.8281972265023112,
|
| 725 |
+
"grad_norm": 4.9485392570495605,
|
| 726 |
+
"learning_rate": 0.00019854696545845325,
|
| 727 |
+
"loss": 3.0161468505859377,
|
| 728 |
+
"step": 1075
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.8281972265023112,
|
| 732 |
+
"step": 1075,
|
| 733 |
+
"train/loss_commentary": 3.358492851257324
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.8281972265023112,
|
| 737 |
+
"step": 1075,
|
| 738 |
+
"train/loss_grounding": 0.4151490330696106
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.847457627118644,
|
| 742 |
+
"grad_norm": 5.522010803222656,
|
| 743 |
+
"learning_rate": 0.00019843909062576103,
|
| 744 |
+
"loss": 3.043304443359375,
|
| 745 |
+
"step": 1100
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.847457627118644,
|
| 749 |
+
"step": 1100,
|
| 750 |
+
"train/loss_commentary": 3.107300281524658
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 0.8667180277349769,
|
| 754 |
+
"grad_norm": 5.547881603240967,
|
| 755 |
+
"learning_rate": 0.0001983273849488906,
|
| 756 |
+
"loss": 3.019481201171875,
|
| 757 |
+
"step": 1125
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 0.8667180277349769,
|
| 761 |
+
"step": 1125,
|
| 762 |
+
"train/loss_commentary": 3.6284444332122803
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.8667180277349769,
|
| 766 |
+
"step": 1125,
|
| 767 |
+
"train/loss_grounding": 0.32966309785842896
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.8859784283513097,
|
| 771 |
+
"grad_norm": 5.641401767730713,
|
| 772 |
+
"learning_rate": 0.00019821185277496705,
|
| 773 |
+
"loss": 3.015819091796875,
|
| 774 |
+
"step": 1150
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.8859784283513097,
|
| 778 |
+
"step": 1150,
|
| 779 |
+
"train/loss_commentary": 3.2091832160949707
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.8859784283513097,
|
| 783 |
+
"step": 1150,
|
| 784 |
+
"train/loss_grounding": 0.07454752922058105
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 0.9052388289676425,
|
| 788 |
+
"grad_norm": 6.049923896789551,
|
| 789 |
+
"learning_rate": 0.00019809249860002707,
|
| 790 |
+
"loss": 3.0372503662109374,
|
| 791 |
+
"step": 1175
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 0.9052388289676425,
|
| 795 |
+
"step": 1175,
|
| 796 |
+
"train/loss_commentary": 3.027353525161743
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 0.9244992295839753,
|
| 800 |
+
"grad_norm": 4.996873378753662,
|
| 801 |
+
"learning_rate": 0.00019796932706884376,
|
| 802 |
+
"loss": 2.9694619750976563,
|
| 803 |
+
"step": 1200
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.9244992295839753,
|
| 807 |
+
"step": 1200,
|
| 808 |
+
"train/loss_commentary": 3.4352097511291504
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.9244992295839753,
|
| 812 |
+
"step": 1200,
|
| 813 |
+
"train/loss_grounding": 0.40801095962524414
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.9437596302003082,
|
| 817 |
+
"grad_norm": 5.559181213378906,
|
| 818 |
+
"learning_rate": 0.00019784234297474612,
|
| 819 |
+
"loss": 2.9874993896484376,
|
| 820 |
+
"step": 1225
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 0.9437596302003082,
|
| 824 |
+
"step": 1225,
|
| 825 |
+
"train/loss_commentary": 2.4496941566467285
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.9437596302003082,
|
| 829 |
+
"step": 1225,
|
| 830 |
+
"train/loss_grounding": 0.2623909115791321
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.963020030816641,
|
| 834 |
+
"grad_norm": 4.692135810852051,
|
| 835 |
+
"learning_rate": 0.00019771155125943227,
|
| 836 |
+
"loss": 3.0027130126953123,
|
| 837 |
+
"step": 1250
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.963020030816641,
|
| 841 |
+
"step": 1250,
|
| 842 |
+
"train/loss_commentary": 3.094949722290039
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.963020030816641,
|
| 846 |
+
"step": 1250,
|
| 847 |
+
"train/loss_grounding": 0.29363787174224854
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 0.9822804314329738,
|
| 851 |
+
"grad_norm": 5.5430192947387695,
|
| 852 |
+
"learning_rate": 0.00019757695701277746,
|
| 853 |
+
"loss": 3.001900329589844,
|
| 854 |
+
"step": 1275
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.9822804314329738,
|
| 858 |
+
"step": 1275,
|
| 859 |
+
"train/loss_commentary": 3.620838165283203
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.9822804314329738,
|
| 863 |
+
"step": 1275,
|
| 864 |
+
"train/loss_grounding": 0.8283157348632812
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 1.0,
|
| 868 |
+
"eval_loss": 3.091080665588379,
|
| 869 |
+
"eval_runtime": 23.4779,
|
| 870 |
+
"eval_samples_per_second": 79.862,
|
| 871 |
+
"eval_steps_per_second": 1.278,
|
| 872 |
+
"step": 1298
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 1.0015408320493067,
|
| 876 |
+
"grad_norm": 5.057258605957031,
|
| 877 |
+
"learning_rate": 0.00019743856547263565,
|
| 878 |
+
"loss": 2.971866455078125,
|
| 879 |
+
"step": 1300
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.0015408320493067,
|
| 883 |
+
"step": 1300,
|
| 884 |
+
"train/loss_commentary": 2.974379777908325
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.0015408320493067,
|
| 888 |
+
"step": 1300,
|
| 889 |
+
"train/loss_grounding": 1.1300758123397827
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 1.0208012326656395,
|
| 893 |
+
"grad_norm": 5.1049675941467285,
|
| 894 |
+
"learning_rate": 0.0001972963820246359,
|
| 895 |
+
"loss": 2.8544305419921874,
|
| 896 |
+
"step": 1325
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 1.0208012326656395,
|
| 900 |
+
"step": 1325,
|
| 901 |
+
"train/loss_commentary": 2.9629321098327637
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.0208012326656395,
|
| 905 |
+
"step": 1325,
|
| 906 |
+
"train/loss_grounding": 0.3255331814289093
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.0400616332819723,
|
| 910 |
+
"grad_norm": 6.376607418060303,
|
| 911 |
+
"learning_rate": 0.0001971504122019727,
|
| 912 |
+
"loss": 2.80501220703125,
|
| 913 |
+
"step": 1350
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 1.0400616332819723,
|
| 917 |
+
"step": 1350,
|
| 918 |
+
"train/loss_commentary": 2.4742014408111572
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.0400616332819723,
|
| 922 |
+
"step": 1350,
|
| 923 |
+
"train/loss_grounding": 0.2645317614078522
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.0593220338983051,
|
| 927 |
+
"grad_norm": 5.080802917480469,
|
| 928 |
+
"learning_rate": 0.00019700066168519068,
|
| 929 |
+
"loss": 2.86232666015625,
|
| 930 |
+
"step": 1375
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.0593220338983051,
|
| 934 |
+
"step": 1375,
|
| 935 |
+
"train/loss_commentary": 2.502061605453491
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 1.0593220338983051,
|
| 939 |
+
"step": 1375,
|
| 940 |
+
"train/loss_grounding": 0.2997746169567108
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.078582434514638,
|
| 944 |
+
"grad_norm": 5.816366195678711,
|
| 945 |
+
"learning_rate": 0.00019684713630196344,
|
| 946 |
+
"loss": 2.8555743408203127,
|
| 947 |
+
"step": 1400
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.078582434514638,
|
| 951 |
+
"step": 1400,
|
| 952 |
+
"train/loss_commentary": 2.7969703674316406
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 1.078582434514638,
|
| 956 |
+
"step": 1400,
|
| 957 |
+
"train/loss_grounding": 0.4004746377468109
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 1.0978428351309708,
|
| 961 |
+
"grad_norm": 4.9736175537109375,
|
| 962 |
+
"learning_rate": 0.000196689842026867,
|
| 963 |
+
"loss": 2.845473327636719,
|
| 964 |
+
"step": 1425
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 1.0978428351309708,
|
| 968 |
+
"step": 1425,
|
| 969 |
+
"train/loss_commentary": 3.4462995529174805
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.0978428351309708,
|
| 973 |
+
"step": 1425,
|
| 974 |
+
"train/loss_grounding": 0.09122885763645172
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.1171032357473036,
|
| 978 |
+
"grad_norm": 5.362197399139404,
|
| 979 |
+
"learning_rate": 0.00019652878498114688,
|
| 980 |
+
"loss": 2.887198486328125,
|
| 981 |
+
"step": 1450
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.1171032357473036,
|
| 985 |
+
"step": 1450,
|
| 986 |
+
"train/loss_commentary": 3.1834933757781982
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.1171032357473036,
|
| 990 |
+
"step": 1450,
|
| 991 |
+
"train/loss_grounding": 0.0760427713394165
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 1.1363636363636362,
|
| 995 |
+
"grad_norm": 6.323387622833252,
|
| 996 |
+
"learning_rate": 0.00019636397143248045,
|
| 997 |
+
"loss": 2.895880126953125,
|
| 998 |
+
"step": 1475
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 1.1363636363636362,
|
| 1002 |
+
"step": 1475,
|
| 1003 |
+
"train/loss_commentary": 2.9457924365997314
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.1363636363636362,
|
| 1007 |
+
"step": 1475,
|
| 1008 |
+
"train/loss_grounding": 0.9209918975830078
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 1.1556240369799693,
|
| 1012 |
+
"grad_norm": 5.502486705780029,
|
| 1013 |
+
"learning_rate": 0.00019619540779473253,
|
| 1014 |
+
"loss": 2.8752532958984376,
|
| 1015 |
+
"step": 1500
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 1.1556240369799693,
|
| 1019 |
+
"step": 1500,
|
| 1020 |
+
"train/loss_commentary": 2.99944806098938
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.1556240369799693,
|
| 1024 |
+
"step": 1500,
|
| 1025 |
+
"train/loss_grounding": 0.1289810836315155
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 1.174884437596302,
|
| 1029 |
+
"grad_norm": 6.060078144073486,
|
| 1030 |
+
"learning_rate": 0.00019602310062770596,
|
| 1031 |
+
"loss": 2.9249896240234374,
|
| 1032 |
+
"step": 1525
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 1.174884437596302,
|
| 1036 |
+
"step": 1525,
|
| 1037 |
+
"train/loss_commentary": 2.9415407180786133
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 1.174884437596302,
|
| 1041 |
+
"step": 1525,
|
| 1042 |
+
"train/loss_grounding": 0.3085947036743164
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 1.1941448382126347,
|
| 1046 |
+
"grad_norm": 5.011383533477783,
|
| 1047 |
+
"learning_rate": 0.0001958470566368864,
|
| 1048 |
+
"loss": 2.89454345703125,
|
| 1049 |
+
"step": 1550
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 1.1941448382126347,
|
| 1053 |
+
"step": 1550,
|
| 1054 |
+
"train/loss_commentary": 2.964836359024048
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 1.1941448382126347,
|
| 1058 |
+
"step": 1550,
|
| 1059 |
+
"train/loss_grounding": 0.18082605302333832
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.2134052388289676,
|
| 1063 |
+
"grad_norm": 5.406578540802002,
|
| 1064 |
+
"learning_rate": 0.00019566728267318126,
|
| 1065 |
+
"loss": 2.8610693359375,
|
| 1066 |
+
"step": 1575
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.2134052388289676,
|
| 1070 |
+
"step": 1575,
|
| 1071 |
+
"train/loss_commentary": 3.132497549057007
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.2134052388289676,
|
| 1075 |
+
"step": 1575,
|
| 1076 |
+
"train/loss_grounding": 0.8671011328697205
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.2326656394453004,
|
| 1080 |
+
"grad_norm": 5.859145641326904,
|
| 1081 |
+
"learning_rate": 0.00019548378573265316,
|
| 1082 |
+
"loss": 2.8463009643554686,
|
| 1083 |
+
"step": 1600
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 1.2326656394453004,
|
| 1087 |
+
"step": 1600,
|
| 1088 |
+
"train/loss_commentary": 2.4842326641082764
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 1.2519260400616332,
|
| 1092 |
+
"grad_norm": 5.900818824768066,
|
| 1093 |
+
"learning_rate": 0.00019529657295624765,
|
| 1094 |
+
"loss": 2.863280029296875,
|
| 1095 |
+
"step": 1625
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 1.2519260400616332,
|
| 1099 |
+
"step": 1625,
|
| 1100 |
+
"train/loss_commentary": 3.284883499145508
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.2519260400616332,
|
| 1104 |
+
"step": 1625,
|
| 1105 |
+
"train/loss_grounding": 0.13215823471546173
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 1.271186440677966,
|
| 1109 |
+
"grad_norm": 5.326910018920898,
|
| 1110 |
+
"learning_rate": 0.00019510565162951537,
|
| 1111 |
+
"loss": 2.9212069702148438,
|
| 1112 |
+
"step": 1650
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.271186440677966,
|
| 1116 |
+
"step": 1650,
|
| 1117 |
+
"train/loss_commentary": 3.3257949352264404
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 1.271186440677966,
|
| 1121 |
+
"step": 1650,
|
| 1122 |
+
"train/loss_grounding": 0.3461397886276245
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.2904468412942989,
|
| 1126 |
+
"grad_norm": 5.251513481140137,
|
| 1127 |
+
"learning_rate": 0.00019491102918232835,
|
| 1128 |
+
"loss": 2.873790283203125,
|
| 1129 |
+
"step": 1675
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.2904468412942989,
|
| 1133 |
+
"step": 1675,
|
| 1134 |
+
"train/loss_commentary": 3.392631769180298
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.2904468412942989,
|
| 1138 |
+
"step": 1675,
|
| 1139 |
+
"train/loss_grounding": 0.27927327156066895
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 1.3097072419106317,
|
| 1143 |
+
"grad_norm": 6.155456066131592,
|
| 1144 |
+
"learning_rate": 0.00019471271318859113,
|
| 1145 |
+
"loss": 2.8460769653320312,
|
| 1146 |
+
"step": 1700
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 1.3097072419106317,
|
| 1150 |
+
"step": 1700,
|
| 1151 |
+
"train/loss_commentary": 2.5610594749450684
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 1.3097072419106317,
|
| 1155 |
+
"step": 1700,
|
| 1156 |
+
"train/loss_grounding": 0.4441479742527008
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 1.3289676425269645,
|
| 1160 |
+
"grad_norm": 5.349524974822998,
|
| 1161 |
+
"learning_rate": 0.00019451071136594573,
|
| 1162 |
+
"loss": 2.9001910400390627,
|
| 1163 |
+
"step": 1725
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 1.3289676425269645,
|
| 1167 |
+
"step": 1725,
|
| 1168 |
+
"train/loss_commentary": 3.204929828643799
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 1.3482280431432974,
|
| 1172 |
+
"grad_norm": 5.620206832885742,
|
| 1173 |
+
"learning_rate": 0.00019430503157547157,
|
| 1174 |
+
"loss": 2.882921142578125,
|
| 1175 |
+
"step": 1750
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 1.3482280431432974,
|
| 1179 |
+
"step": 1750,
|
| 1180 |
+
"train/loss_commentary": 3.0682384967803955
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 1.3482280431432974,
|
| 1184 |
+
"step": 1750,
|
| 1185 |
+
"train/loss_grounding": 0.15147441625595093
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.3674884437596302,
|
| 1189 |
+
"grad_norm": 5.115348815917969,
|
| 1190 |
+
"learning_rate": 0.00019409568182137944,
|
| 1191 |
+
"loss": 2.8463238525390624,
|
| 1192 |
+
"step": 1775
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.3674884437596302,
|
| 1196 |
+
"step": 1775,
|
| 1197 |
+
"train/loss_commentary": 3.1464598178863525
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 1.386748844375963,
|
| 1201 |
+
"grad_norm": 4.789853096008301,
|
| 1202 |
+
"learning_rate": 0.00019388267025069995,
|
| 1203 |
+
"loss": 2.882373046875,
|
| 1204 |
+
"step": 1800
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 1.386748844375963,
|
| 1208 |
+
"step": 1800,
|
| 1209 |
+
"train/loss_commentary": 2.999337911605835
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.386748844375963,
|
| 1213 |
+
"step": 1800,
|
| 1214 |
+
"train/loss_grounding": 0.1654149442911148
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 1.4060092449922958,
|
| 1218 |
+
"grad_norm": 5.377254486083984,
|
| 1219 |
+
"learning_rate": 0.00019366600515296657,
|
| 1220 |
+
"loss": 2.837642822265625,
|
| 1221 |
+
"step": 1825
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.4060092449922958,
|
| 1225 |
+
"step": 1825,
|
| 1226 |
+
"train/loss_commentary": 2.7034339904785156
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 1.4060092449922958,
|
| 1230 |
+
"step": 1825,
|
| 1231 |
+
"train/loss_grounding": 0.8760122656822205
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 1.4252696456086287,
|
| 1235 |
+
"grad_norm": 5.0359272956848145,
|
| 1236 |
+
"learning_rate": 0.00019344569495989293,
|
| 1237 |
+
"loss": 2.8039517211914062,
|
| 1238 |
+
"step": 1850
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 1.4252696456086287,
|
| 1242 |
+
"step": 1850,
|
| 1243 |
+
"train/loss_commentary": 2.4528393745422363
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 1.4252696456086287,
|
| 1247 |
+
"step": 1850,
|
| 1248 |
+
"train/loss_grounding": 0.4055929481983185
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.4445300462249615,
|
| 1252 |
+
"grad_norm": 5.267066955566406,
|
| 1253 |
+
"learning_rate": 0.0001932217482450449,
|
| 1254 |
+
"loss": 2.8433935546875,
|
| 1255 |
+
"step": 1875
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.4445300462249615,
|
| 1259 |
+
"step": 1875,
|
| 1260 |
+
"train/loss_commentary": 3.1104671955108643
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 1.4445300462249615,
|
| 1264 |
+
"step": 1875,
|
| 1265 |
+
"train/loss_grounding": 0.16951119899749756
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 1.4637904468412943,
|
| 1269 |
+
"grad_norm": 5.126423358917236,
|
| 1270 |
+
"learning_rate": 0.00019299417372350673,
|
| 1271 |
+
"loss": 2.810785827636719,
|
| 1272 |
+
"step": 1900
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 1.4637904468412943,
|
| 1276 |
+
"step": 1900,
|
| 1277 |
+
"train/loss_commentary": 3.0781702995300293
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 1.4637904468412943,
|
| 1281 |
+
"step": 1900,
|
| 1282 |
+
"train/loss_grounding": 0.23617996275424957
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.4830508474576272,
|
| 1286 |
+
"grad_norm": 4.853616237640381,
|
| 1287 |
+
"learning_rate": 0.0001927629802515419,
|
| 1288 |
+
"loss": 2.88142578125,
|
| 1289 |
+
"step": 1925
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.4830508474576272,
|
| 1293 |
+
"step": 1925,
|
| 1294 |
+
"train/loss_commentary": 3.6762850284576416
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 1.4830508474576272,
|
| 1298 |
+
"step": 1925,
|
| 1299 |
+
"train/loss_grounding": 0.25680238008499146
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.50231124807396,
|
| 1303 |
+
"grad_norm": 5.107408046722412,
|
| 1304 |
+
"learning_rate": 0.00019252817682624876,
|
| 1305 |
+
"loss": 2.872176513671875,
|
| 1306 |
+
"step": 1950
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 1.50231124807396,
|
| 1310 |
+
"step": 1950,
|
| 1311 |
+
"train/loss_commentary": 3.264618158340454
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.50231124807396,
|
| 1315 |
+
"step": 1950,
|
| 1316 |
+
"train/loss_grounding": 0.23078688979148865
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 1.5215716486902928,
|
| 1320 |
+
"grad_norm": 5.371690273284912,
|
| 1321 |
+
"learning_rate": 0.0001922897725852099,
|
| 1322 |
+
"loss": 2.8697796630859376,
|
| 1323 |
+
"step": 1975
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 1.5215716486902928,
|
| 1327 |
+
"step": 1975,
|
| 1328 |
+
"train/loss_commentary": 2.8013217449188232
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 1.5215716486902928,
|
| 1332 |
+
"step": 1975,
|
| 1333 |
+
"train/loss_grounding": 0.5688762664794922
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 1.5408320493066254,
|
| 1337 |
+
"grad_norm": 5.919620990753174,
|
| 1338 |
+
"learning_rate": 0.00019204777680613706,
|
| 1339 |
+
"loss": 2.855506591796875,
|
| 1340 |
+
"step": 2000
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 1.5408320493066254,
|
| 1344 |
+
"step": 2000,
|
| 1345 |
+
"train/loss_commentary": 3.0764052867889404
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.5408320493066254,
|
| 1349 |
+
"step": 2000,
|
| 1350 |
+
"train/loss_grounding": 2.5531704425811768
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 1.5600924499229585,
|
| 1354 |
+
"grad_norm": 5.038231372833252,
|
| 1355 |
+
"learning_rate": 0.00019180219890650976,
|
| 1356 |
+
"loss": 2.8509075927734373,
|
| 1357 |
+
"step": 2025
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 1.5600924499229585,
|
| 1361 |
+
"step": 2025,
|
| 1362 |
+
"train/loss_commentary": 2.815100908279419
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 1.579352850539291,
|
| 1366 |
+
"grad_norm": 5.328333854675293,
|
| 1367 |
+
"learning_rate": 0.00019155304844320888,
|
| 1368 |
+
"loss": 2.853030090332031,
|
| 1369 |
+
"step": 2050
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 1.579352850539291,
|
| 1373 |
+
"step": 2050,
|
| 1374 |
+
"train/loss_commentary": 4.636526107788086
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.579352850539291,
|
| 1378 |
+
"step": 2050,
|
| 1379 |
+
"train/loss_grounding": 1.0121015310287476
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 1.5986132511556241,
|
| 1383 |
+
"grad_norm": 5.38020133972168,
|
| 1384 |
+
"learning_rate": 0.0001913003351121449,
|
| 1385 |
+
"loss": 2.8369183349609375,
|
| 1386 |
+
"step": 2075
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 1.5986132511556241,
|
| 1390 |
+
"step": 2075,
|
| 1391 |
+
"train/loss_commentary": 2.957808256149292
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 1.5986132511556241,
|
| 1395 |
+
"step": 2075,
|
| 1396 |
+
"train/loss_grounding": 0.0730256512761116
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.6178736517719567,
|
| 1400 |
+
"grad_norm": 4.666707992553711,
|
| 1401 |
+
"learning_rate": 0.00019104406874788027,
|
| 1402 |
+
"loss": 2.8393048095703124,
|
| 1403 |
+
"step": 2100
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 1.6178736517719567,
|
| 1407 |
+
"step": 2100,
|
| 1408 |
+
"train/loss_commentary": 3.086062431335449
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.6178736517719567,
|
| 1412 |
+
"step": 2100,
|
| 1413 |
+
"train/loss_grounding": 0.5748575329780579
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.6371340523882898,
|
| 1417 |
+
"grad_norm": 5.214188575744629,
|
| 1418 |
+
"learning_rate": 0.000190784259323247,
|
| 1419 |
+
"loss": 2.817139892578125,
|
| 1420 |
+
"step": 2125
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 1.6371340523882898,
|
| 1424 |
+
"step": 2125,
|
| 1425 |
+
"train/loss_commentary": 3.0936436653137207
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 1.6563944530046224,
|
| 1429 |
+
"grad_norm": 5.7970170974731445,
|
| 1430 |
+
"learning_rate": 0.0001905209169489584,
|
| 1431 |
+
"loss": 2.8522073364257814,
|
| 1432 |
+
"step": 2150
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 1.6563944530046224,
|
| 1436 |
+
"step": 2150,
|
| 1437 |
+
"train/loss_commentary": 3.106658935546875
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.6563944530046224,
|
| 1441 |
+
"step": 2150,
|
| 1442 |
+
"train/loss_grounding": 0.33706483244895935
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 1.6756548536209555,
|
| 1446 |
+
"grad_norm": 5.0848469734191895,
|
| 1447 |
+
"learning_rate": 0.00019025405187321557,
|
| 1448 |
+
"loss": 2.8525799560546874,
|
| 1449 |
+
"step": 2175
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.6756548536209555,
|
| 1453 |
+
"step": 2175,
|
| 1454 |
+
"train/loss_commentary": 2.871026039123535
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 1.6756548536209555,
|
| 1458 |
+
"step": 2175,
|
| 1459 |
+
"train/loss_grounding": 0.3417157232761383
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 1.694915254237288,
|
| 1463 |
+
"grad_norm": 5.509637355804443,
|
| 1464 |
+
"learning_rate": 0.0001899836744813087,
|
| 1465 |
+
"loss": 2.818135986328125,
|
| 1466 |
+
"step": 2200
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 1.694915254237288,
|
| 1470 |
+
"step": 2200,
|
| 1471 |
+
"train/loss_commentary": 2.323927879333496
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.694915254237288,
|
| 1475 |
+
"step": 2200,
|
| 1476 |
+
"train/loss_grounding": 0.13981002569198608
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 1.7141756548536211,
|
| 1480 |
+
"grad_norm": 5.253782272338867,
|
| 1481 |
+
"learning_rate": 0.00018970979529521291,
|
| 1482 |
+
"loss": 2.8384332275390625,
|
| 1483 |
+
"step": 2225
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.7141756548536211,
|
| 1487 |
+
"step": 2225,
|
| 1488 |
+
"train/loss_commentary": 3.267014503479004
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 1.7141756548536211,
|
| 1492 |
+
"step": 2225,
|
| 1493 |
+
"train/loss_grounding": 0.3427298069000244
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 1.7334360554699537,
|
| 1497 |
+
"grad_norm": 5.7060112953186035,
|
| 1498 |
+
"learning_rate": 0.00018943242497317865,
|
| 1499 |
+
"loss": 2.872622375488281,
|
| 1500 |
+
"step": 2250
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 1.7334360554699537,
|
| 1504 |
+
"step": 2250,
|
| 1505 |
+
"train/loss_commentary": 2.5845558643341064
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 1.7526964560862865,
|
| 1509 |
+
"grad_norm": 5.4669389724731445,
|
| 1510 |
+
"learning_rate": 0.00018915157430931701,
|
| 1511 |
+
"loss": 2.8570907592773436,
|
| 1512 |
+
"step": 2275
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 1.7526964560862865,
|
| 1516 |
+
"step": 2275,
|
| 1517 |
+
"train/loss_commentary": 3.4804110527038574
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 1.7526964560862865,
|
| 1521 |
+
"step": 2275,
|
| 1522 |
+
"train/loss_grounding": 1.3138675689697266
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 1.7719568567026194,
|
| 1526 |
+
"grad_norm": 5.46798849105835,
|
| 1527 |
+
"learning_rate": 0.0001888672542331798,
|
| 1528 |
+
"loss": 2.8511175537109374,
|
| 1529 |
+
"step": 2300
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 1.7719568567026194,
|
| 1533 |
+
"step": 2300,
|
| 1534 |
+
"train/loss_commentary": 2.359330654144287
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.7719568567026194,
|
| 1538 |
+
"step": 2300,
|
| 1539 |
+
"train/loss_grounding": 0.03408310189843178
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 1.7912172573189522,
|
| 1543 |
+
"grad_norm": 5.422307014465332,
|
| 1544 |
+
"learning_rate": 0.00018857947580933385,
|
| 1545 |
+
"loss": 2.8429791259765627,
|
| 1546 |
+
"step": 2325
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 1.7912172573189522,
|
| 1550 |
+
"step": 2325,
|
| 1551 |
+
"train/loss_commentary": 3.1596031188964844
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 1.810477657935285,
|
| 1555 |
+
"grad_norm": 4.941929817199707,
|
| 1556 |
+
"learning_rate": 0.0001882882502369308,
|
| 1557 |
+
"loss": 2.8539962768554688,
|
| 1558 |
+
"step": 2350
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 1.810477657935285,
|
| 1562 |
+
"step": 2350,
|
| 1563 |
+
"train/loss_commentary": 2.6755995750427246
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.810477657935285,
|
| 1567 |
+
"step": 2350,
|
| 1568 |
+
"train/loss_grounding": 0.36783620715141296
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 1.8297380585516179,
|
| 1572 |
+
"grad_norm": 4.98409366607666,
|
| 1573 |
+
"learning_rate": 0.00018799358884927117,
|
| 1574 |
+
"loss": 2.88785888671875,
|
| 1575 |
+
"step": 2375
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 1.8297380585516179,
|
| 1579 |
+
"step": 2375,
|
| 1580 |
+
"train/loss_commentary": 2.7453420162200928
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 1.8297380585516179,
|
| 1584 |
+
"step": 2375,
|
| 1585 |
+
"train/loss_grounding": 0.38387295603752136
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 1.8489984591679507,
|
| 1589 |
+
"grad_norm": 5.071564674377441,
|
| 1590 |
+
"learning_rate": 0.00018769550311336305,
|
| 1591 |
+
"loss": 2.8275177001953127,
|
| 1592 |
+
"step": 2400
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 1.8489984591679507,
|
| 1596 |
+
"step": 2400,
|
| 1597 |
+
"train/loss_commentary": 2.960709571838379
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.8489984591679507,
|
| 1601 |
+
"step": 2400,
|
| 1602 |
+
"train/loss_grounding": 0.5020418763160706
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 1.8682588597842835,
|
| 1606 |
+
"grad_norm": 5.50569486618042,
|
| 1607 |
+
"learning_rate": 0.00018739400462947626,
|
| 1608 |
+
"loss": 2.8057931518554686,
|
| 1609 |
+
"step": 2425
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 1.8682588597842835,
|
| 1613 |
+
"step": 2425,
|
| 1614 |
+
"train/loss_commentary": 2.8025293350219727
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 1.8682588597842835,
|
| 1618 |
+
"step": 2425,
|
| 1619 |
+
"train/loss_grounding": 0.13795208930969238
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 1.8875192604006163,
|
| 1623 |
+
"grad_norm": 5.285870552062988,
|
| 1624 |
+
"learning_rate": 0.00018708910513069068,
|
| 1625 |
+
"loss": 2.837681884765625,
|
| 1626 |
+
"step": 2450
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 1.8875192604006163,
|
| 1630 |
+
"step": 2450,
|
| 1631 |
+
"train/loss_commentary": 2.7294793128967285
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 1.8875192604006163,
|
| 1635 |
+
"step": 2450,
|
| 1636 |
+
"train/loss_grounding": 1.5065491199493408
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 1.9067796610169492,
|
| 1640 |
+
"grad_norm": 5.168023586273193,
|
| 1641 |
+
"learning_rate": 0.00018678081648243963,
|
| 1642 |
+
"loss": 2.8178057861328125,
|
| 1643 |
+
"step": 2475
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 1.9067796610169492,
|
| 1647 |
+
"step": 2475,
|
| 1648 |
+
"train/loss_commentary": 2.875864028930664
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 1.9067796610169492,
|
| 1652 |
+
"step": 2475,
|
| 1653 |
+
"train/loss_grounding": 0.10482349991798401
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.926040061633282,
|
| 1657 |
+
"grad_norm": 5.817725658416748,
|
| 1658 |
+
"learning_rate": 0.00018646915068204825,
|
| 1659 |
+
"loss": 2.790821533203125,
|
| 1660 |
+
"step": 2500
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.926040061633282,
|
| 1664 |
+
"step": 2500,
|
| 1665 |
+
"train/loss_commentary": 2.16098690032959
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 1.926040061633282,
|
| 1669 |
+
"step": 2500,
|
| 1670 |
+
"train/loss_grounding": 0.8708534240722656
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 1.9453004622496148,
|
| 1674 |
+
"grad_norm": 4.8011040687561035,
|
| 1675 |
+
"learning_rate": 0.00018615411985826647,
|
| 1676 |
+
"loss": 2.79728271484375,
|
| 1677 |
+
"step": 2525
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 1.9453004622496148,
|
| 1681 |
+
"step": 2525,
|
| 1682 |
+
"train/loss_commentary": 2.6063613891601562
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 1.9453004622496148,
|
| 1686 |
+
"step": 2525,
|
| 1687 |
+
"train/loss_grounding": 0.17927438020706177
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 1.9645608628659477,
|
| 1691 |
+
"grad_norm": 4.666601657867432,
|
| 1692 |
+
"learning_rate": 0.00018583573627079715,
|
| 1693 |
+
"loss": 2.858675842285156,
|
| 1694 |
+
"step": 2550
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 1.9645608628659477,
|
| 1698 |
+
"step": 2550,
|
| 1699 |
+
"train/loss_commentary": 2.5858216285705566
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 1.9838212634822803,
|
| 1703 |
+
"grad_norm": 5.368419170379639,
|
| 1704 |
+
"learning_rate": 0.00018551401230981884,
|
| 1705 |
+
"loss": 2.8355035400390625,
|
| 1706 |
+
"step": 2575
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 1.9838212634822803,
|
| 1710 |
+
"step": 2575,
|
| 1711 |
+
"train/loss_commentary": 2.751040458679199
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 1.9838212634822803,
|
| 1715 |
+
"step": 2575,
|
| 1716 |
+
"train/loss_grounding": 0.5289782285690308
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 2.0,
|
| 1720 |
+
"eval_loss": 2.9402315616607666,
|
| 1721 |
+
"eval_runtime": 14.9129,
|
| 1722 |
+
"eval_samples_per_second": 125.73,
|
| 1723 |
+
"eval_steps_per_second": 2.012,
|
| 1724 |
+
"step": 2596
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 2.0030816640986133,
|
| 1728 |
+
"grad_norm": 6.034282684326172,
|
| 1729 |
+
"learning_rate": 0.00018518896049550375,
|
| 1730 |
+
"loss": 2.791551513671875,
|
| 1731 |
+
"step": 2600
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 2.0030816640986133,
|
| 1735 |
+
"step": 2600,
|
| 1736 |
+
"train/loss_commentary": 2.2208616733551025
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 2.0030816640986133,
|
| 1740 |
+
"step": 2600,
|
| 1741 |
+
"train/loss_grounding": 0.2887810170650482
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 2.022342064714946,
|
| 1745 |
+
"grad_norm": 5.422306537628174,
|
| 1746 |
+
"learning_rate": 0.00018486059347753035,
|
| 1747 |
+
"loss": 2.6534307861328124,
|
| 1748 |
+
"step": 2625
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 2.022342064714946,
|
| 1752 |
+
"step": 2625,
|
| 1753 |
+
"train/loss_commentary": 2.4502615928649902
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 2.022342064714946,
|
| 1757 |
+
"step": 2625,
|
| 1758 |
+
"train/loss_grounding": 0.7137656211853027
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 2.041602465331279,
|
| 1762 |
+
"grad_norm": 5.733297348022461,
|
| 1763 |
+
"learning_rate": 0.00018452892403459127,
|
| 1764 |
+
"loss": 2.6396722412109375,
|
| 1765 |
+
"step": 2650
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 2.041602465331279,
|
| 1769 |
+
"step": 2650,
|
| 1770 |
+
"train/loss_commentary": 3.033895492553711
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.041602465331279,
|
| 1774 |
+
"step": 2650,
|
| 1775 |
+
"train/loss_grounding": 1.2966415882110596
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 2.0608628659476116,
|
| 1779 |
+
"grad_norm": 5.566621780395508,
|
| 1780 |
+
"learning_rate": 0.00018419396507389594,
|
| 1781 |
+
"loss": 2.681654052734375,
|
| 1782 |
+
"step": 2675
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 2.0608628659476116,
|
| 1786 |
+
"step": 2675,
|
| 1787 |
+
"train/loss_commentary": 3.039780855178833
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 2.0608628659476116,
|
| 1791 |
+
"step": 2675,
|
| 1792 |
+
"train/loss_grounding": 0.26366662979125977
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 2.0801232665639446,
|
| 1796 |
+
"grad_norm": 5.401014804840088,
|
| 1797 |
+
"learning_rate": 0.00018385572963066813,
|
| 1798 |
+
"loss": 2.669170227050781,
|
| 1799 |
+
"step": 2700
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 2.0801232665639446,
|
| 1803 |
+
"step": 2700,
|
| 1804 |
+
"train/loss_commentary": 1.9797307252883911
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 2.0993836671802772,
|
| 1808 |
+
"grad_norm": 5.241693019866943,
|
| 1809 |
+
"learning_rate": 0.00018351423086763904,
|
| 1810 |
+
"loss": 2.649722900390625,
|
| 1811 |
+
"step": 2725
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 2.0993836671802772,
|
| 1815 |
+
"step": 2725,
|
| 1816 |
+
"train/loss_commentary": 3.0994417667388916
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 2.1186440677966103,
|
| 1820 |
+
"grad_norm": 5.540595531463623,
|
| 1821 |
+
"learning_rate": 0.00018316948207453465,
|
| 1822 |
+
"loss": 2.6695376586914064,
|
| 1823 |
+
"step": 2750
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 2.1186440677966103,
|
| 1827 |
+
"step": 2750,
|
| 1828 |
+
"train/loss_commentary": 2.4479124546051025
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 2.1186440677966103,
|
| 1832 |
+
"step": 2750,
|
| 1833 |
+
"train/loss_grounding": 0.16718991100788116
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 2.137904468412943,
|
| 1837 |
+
"grad_norm": 5.459566593170166,
|
| 1838 |
+
"learning_rate": 0.0001828214966675589,
|
| 1839 |
+
"loss": 2.6481033325195313,
|
| 1840 |
+
"step": 2775
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 2.137904468412943,
|
| 1844 |
+
"step": 2775,
|
| 1845 |
+
"train/loss_commentary": 2.8178234100341797
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 2.137904468412943,
|
| 1849 |
+
"step": 2775,
|
| 1850 |
+
"train/loss_grounding": 0.5014618635177612
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 2.157164869029276,
|
| 1854 |
+
"grad_norm": 5.371711254119873,
|
| 1855 |
+
"learning_rate": 0.0001824702881888712,
|
| 1856 |
+
"loss": 2.687843933105469,
|
| 1857 |
+
"step": 2800
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 2.157164869029276,
|
| 1861 |
+
"step": 2800,
|
| 1862 |
+
"train/loss_commentary": 2.713888168334961
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 2.157164869029276,
|
| 1866 |
+
"step": 2800,
|
| 1867 |
+
"train/loss_grounding": 0.055346786975860596
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 2.1764252696456086,
|
| 1871 |
+
"grad_norm": 4.793285369873047,
|
| 1872 |
+
"learning_rate": 0.0001821158703060599,
|
| 1873 |
+
"loss": 2.7188433837890624,
|
| 1874 |
+
"step": 2825
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 2.1764252696456086,
|
| 1878 |
+
"step": 2825,
|
| 1879 |
+
"train/loss_commentary": 3.3175947666168213
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 2.1764252696456086,
|
| 1883 |
+
"step": 2825,
|
| 1884 |
+
"train/loss_grounding": 0.28087353706359863
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 2.1956856702619416,
|
| 1888 |
+
"grad_norm": 5.136083602905273,
|
| 1889 |
+
"learning_rate": 0.00018175825681161003,
|
| 1890 |
+
"loss": 2.706739501953125,
|
| 1891 |
+
"step": 2850
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 2.1956856702619416,
|
| 1895 |
+
"step": 2850,
|
| 1896 |
+
"train/loss_commentary": 2.600324869155884
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 2.214946070878274,
|
| 1900 |
+
"grad_norm": 5.682553291320801,
|
| 1901 |
+
"learning_rate": 0.00018139746162236655,
|
| 1902 |
+
"loss": 2.667236633300781,
|
| 1903 |
+
"step": 2875
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 2.214946070878274,
|
| 1907 |
+
"step": 2875,
|
| 1908 |
+
"train/loss_commentary": 2.577211380004883
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 2.214946070878274,
|
| 1912 |
+
"step": 2875,
|
| 1913 |
+
"train/loss_grounding": 0.5466868281364441
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 2.2342064714946073,
|
| 1917 |
+
"grad_norm": 5.907873630523682,
|
| 1918 |
+
"learning_rate": 0.00018103349877899308,
|
| 1919 |
+
"loss": 2.7007156372070313,
|
| 1920 |
+
"step": 2900
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 2.2342064714946073,
|
| 1924 |
+
"step": 2900,
|
| 1925 |
+
"train/loss_commentary": 3.142120122909546
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 2.2342064714946073,
|
| 1929 |
+
"step": 2900,
|
| 1930 |
+
"train/loss_grounding": 0.8227649927139282
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 2.25346687211094,
|
| 1934 |
+
"grad_norm": 5.540514945983887,
|
| 1935 |
+
"learning_rate": 0.00018066638244542516,
|
| 1936 |
+
"loss": 2.6651422119140626,
|
| 1937 |
+
"step": 2925
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 2.25346687211094,
|
| 1941 |
+
"step": 2925,
|
| 1942 |
+
"train/loss_commentary": 3.031633138656616
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 2.25346687211094,
|
| 1946 |
+
"step": 2925,
|
| 1947 |
+
"train/loss_grounding": 0.9412539601325989
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 2.2727272727272725,
|
| 1951 |
+
"grad_norm": 5.641888618469238,
|
| 1952 |
+
"learning_rate": 0.00018029612690831929,
|
| 1953 |
+
"loss": 2.695490417480469,
|
| 1954 |
+
"step": 2950
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 2.2727272727272725,
|
| 1958 |
+
"step": 2950,
|
| 1959 |
+
"train/loss_commentary": 2.5322117805480957
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 2.2919876733436055,
|
| 1963 |
+
"grad_norm": 4.826931953430176,
|
| 1964 |
+
"learning_rate": 0.00017992274657649678,
|
| 1965 |
+
"loss": 2.697246398925781,
|
| 1966 |
+
"step": 2975
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 2.2919876733436055,
|
| 1970 |
+
"step": 2975,
|
| 1971 |
+
"train/loss_commentary": 2.8198845386505127
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 2.2919876733436055,
|
| 1975 |
+
"step": 2975,
|
| 1976 |
+
"train/loss_grounding": 0.3170523941516876
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 2.3112480739599386,
|
| 1980 |
+
"grad_norm": 5.923715591430664,
|
| 1981 |
+
"learning_rate": 0.0001795462559803831,
|
| 1982 |
+
"loss": 2.667850341796875,
|
| 1983 |
+
"step": 3000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 2.3112480739599386,
|
| 1987 |
+
"step": 3000,
|
| 1988 |
+
"train/loss_commentary": 2.817070484161377
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 2.3112480739599386,
|
| 1992 |
+
"step": 3000,
|
| 1993 |
+
"train/loss_grounding": 0.04586929455399513
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 2.330508474576271,
|
| 1997 |
+
"grad_norm": 6.09520149230957,
|
| 1998 |
+
"learning_rate": 0.00017916666977144243,
|
| 1999 |
+
"loss": 2.707633361816406,
|
| 2000 |
+
"step": 3025
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 2.330508474576271,
|
| 2004 |
+
"step": 3025,
|
| 2005 |
+
"train/loss_commentary": 2.6274843215942383
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 2.349768875192604,
|
| 2009 |
+
"grad_norm": 6.1906256675720215,
|
| 2010 |
+
"learning_rate": 0.0001787840027216074,
|
| 2011 |
+
"loss": 2.6906646728515624,
|
| 2012 |
+
"step": 3050
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 2.349768875192604,
|
| 2016 |
+
"step": 3050,
|
| 2017 |
+
"train/loss_commentary": 2.075770378112793
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 2.349768875192604,
|
| 2021 |
+
"step": 3050,
|
| 2022 |
+
"train/loss_grounding": 0.7645119428634644
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 2.369029275808937,
|
| 2026 |
+
"grad_norm": 4.972816467285156,
|
| 2027 |
+
"learning_rate": 0.00017839826972270437,
|
| 2028 |
+
"loss": 2.66882080078125,
|
| 2029 |
+
"step": 3075
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 2.369029275808937,
|
| 2033 |
+
"step": 3075,
|
| 2034 |
+
"train/loss_commentary": 2.8061487674713135
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 2.369029275808937,
|
| 2038 |
+
"step": 3075,
|
| 2039 |
+
"train/loss_grounding": 0.332036554813385
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 2.3882896764252695,
|
| 2043 |
+
"grad_norm": 5.463335990905762,
|
| 2044 |
+
"learning_rate": 0.00017800948578587376,
|
| 2045 |
+
"loss": 2.6757513427734376,
|
| 2046 |
+
"step": 3100
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 2.3882896764252695,
|
| 2050 |
+
"step": 3100,
|
| 2051 |
+
"train/loss_commentary": 2.4306581020355225
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 2.3882896764252695,
|
| 2055 |
+
"step": 3100,
|
| 2056 |
+
"train/loss_grounding": 0.09657474607229233
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 2.4075500770416025,
|
| 2060 |
+
"grad_norm": 5.794223308563232,
|
| 2061 |
+
"learning_rate": 0.00017761766604098593,
|
| 2062 |
+
"loss": 2.6851986694335936,
|
| 2063 |
+
"step": 3125
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 2.4075500770416025,
|
| 2067 |
+
"step": 3125,
|
| 2068 |
+
"train/loss_commentary": 2.9277796745300293
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 2.4075500770416025,
|
| 2072 |
+
"step": 3125,
|
| 2073 |
+
"train/loss_grounding": 0.4521333575248718
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 2.426810477657935,
|
| 2077 |
+
"grad_norm": 5.309329986572266,
|
| 2078 |
+
"learning_rate": 0.00017722282573605242,
|
| 2079 |
+
"loss": 2.7272637939453124,
|
| 2080 |
+
"step": 3150
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 2.426810477657935,
|
| 2084 |
+
"step": 3150,
|
| 2085 |
+
"train/loss_commentary": 3.041442632675171
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 2.426810477657935,
|
| 2089 |
+
"step": 3150,
|
| 2090 |
+
"train/loss_grounding": 0.019025040790438652
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 2.446070878274268,
|
| 2094 |
+
"grad_norm": 4.8955159187316895,
|
| 2095 |
+
"learning_rate": 0.0001768249802366325,
|
| 2096 |
+
"loss": 2.7085125732421873,
|
| 2097 |
+
"step": 3175
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 2.446070878274268,
|
| 2101 |
+
"step": 3175,
|
| 2102 |
+
"train/loss_commentary": 2.912773847579956
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 2.446070878274268,
|
| 2106 |
+
"step": 3175,
|
| 2107 |
+
"train/loss_grounding": 0.7067486047744751
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 2.4653312788906008,
|
| 2111 |
+
"grad_norm": 5.2584309577941895,
|
| 2112 |
+
"learning_rate": 0.00017642414502523526,
|
| 2113 |
+
"loss": 2.6713641357421873,
|
| 2114 |
+
"step": 3200
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 2.4653312788906008,
|
| 2118 |
+
"step": 3200,
|
| 2119 |
+
"train/loss_commentary": 3.1406283378601074
|
| 2120 |
+
},
|
| 2121 |
+
{
|
| 2122 |
+
"epoch": 2.4653312788906008,
|
| 2123 |
+
"step": 3200,
|
| 2124 |
+
"train/loss_grounding": 0.040799036622047424
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 2.484591679506934,
|
| 2128 |
+
"grad_norm": 5.2439866065979,
|
| 2129 |
+
"learning_rate": 0.0001760203357007171,
|
| 2130 |
+
"loss": 2.7007220458984373,
|
| 2131 |
+
"step": 3225
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 2.484591679506934,
|
| 2135 |
+
"step": 3225,
|
| 2136 |
+
"train/loss_commentary": 2.4081380367279053
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 2.484591679506934,
|
| 2140 |
+
"step": 3225,
|
| 2141 |
+
"train/loss_grounding": 0.5926041603088379
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 2.5038520801232664,
|
| 2145 |
+
"grad_norm": 4.918389797210693,
|
| 2146 |
+
"learning_rate": 0.0001756135679776746,
|
| 2147 |
+
"loss": 2.6809231567382814,
|
| 2148 |
+
"step": 3250
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 2.5038520801232664,
|
| 2152 |
+
"step": 3250,
|
| 2153 |
+
"train/loss_commentary": 2.8008952140808105
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 2.5038520801232664,
|
| 2157 |
+
"step": 3250,
|
| 2158 |
+
"train/loss_grounding": 0.8448731899261475
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 2.5231124807395995,
|
| 2162 |
+
"grad_norm": 5.497406005859375,
|
| 2163 |
+
"learning_rate": 0.0001752038576858331,
|
| 2164 |
+
"loss": 2.7189471435546877,
|
| 2165 |
+
"step": 3275
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 2.5231124807395995,
|
| 2169 |
+
"step": 3275,
|
| 2170 |
+
"train/loss_commentary": 2.29134202003479
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 2.5231124807395995,
|
| 2174 |
+
"step": 3275,
|
| 2175 |
+
"train/loss_grounding": 0.01784053072333336
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 2.542372881355932,
|
| 2179 |
+
"grad_norm": 4.876806735992432,
|
| 2180 |
+
"learning_rate": 0.00017479122076943053,
|
| 2181 |
+
"loss": 2.6589410400390623,
|
| 2182 |
+
"step": 3300
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"epoch": 2.542372881355932,
|
| 2186 |
+
"step": 3300,
|
| 2187 |
+
"train/loss_commentary": 3.289525270462036
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 2.542372881355932,
|
| 2191 |
+
"step": 3300,
|
| 2192 |
+
"train/loss_grounding": 0.6817333102226257
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 2.561633281972265,
|
| 2196 |
+
"grad_norm": 5.2532196044921875,
|
| 2197 |
+
"learning_rate": 0.00017437567328659704,
|
| 2198 |
+
"loss": 2.6795159912109376,
|
| 2199 |
+
"step": 3325
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 2.561633281972265,
|
| 2203 |
+
"step": 3325,
|
| 2204 |
+
"train/loss_commentary": 2.975726366043091
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 2.561633281972265,
|
| 2208 |
+
"step": 3325,
|
| 2209 |
+
"train/loss_grounding": 0.11424746364355087
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 2.5808936825885977,
|
| 2213 |
+
"grad_norm": 5.171988010406494,
|
| 2214 |
+
"learning_rate": 0.00017395723140873007,
|
| 2215 |
+
"loss": 2.6947360229492188,
|
| 2216 |
+
"step": 3350
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 2.5808936825885977,
|
| 2220 |
+
"step": 3350,
|
| 2221 |
+
"train/loss_commentary": 2.782975912094116
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 2.5808936825885977,
|
| 2225 |
+
"step": 3350,
|
| 2226 |
+
"train/loss_grounding": 0.4522864818572998
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 2.600154083204931,
|
| 2230 |
+
"grad_norm": 5.794159412384033,
|
| 2231 |
+
"learning_rate": 0.00017353591141986495,
|
| 2232 |
+
"loss": 2.7043685913085938,
|
| 2233 |
+
"step": 3375
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 2.600154083204931,
|
| 2237 |
+
"step": 3375,
|
| 2238 |
+
"train/loss_commentary": 3.2270755767822266
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 2.6194144838212634,
|
| 2242 |
+
"grad_norm": 5.0811872482299805,
|
| 2243 |
+
"learning_rate": 0.00017311172971604127,
|
| 2244 |
+
"loss": 2.6681253051757814,
|
| 2245 |
+
"step": 3400
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 2.6194144838212634,
|
| 2249 |
+
"step": 3400,
|
| 2250 |
+
"train/loss_commentary": 2.5207998752593994
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 2.6194144838212634,
|
| 2254 |
+
"step": 3400,
|
| 2255 |
+
"train/loss_grounding": 1.0898680686950684
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 2.6386748844375965,
|
| 2259 |
+
"grad_norm": 5.428940773010254,
|
| 2260 |
+
"learning_rate": 0.00017268470280466477,
|
| 2261 |
+
"loss": 2.6442642211914062,
|
| 2262 |
+
"step": 3425
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 2.6386748844375965,
|
| 2266 |
+
"step": 3425,
|
| 2267 |
+
"train/loss_commentary": 3.0041699409484863
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 2.6386748844375965,
|
| 2271 |
+
"step": 3425,
|
| 2272 |
+
"train/loss_grounding": 0.3483811318874359
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 2.657935285053929,
|
| 2276 |
+
"grad_norm": 5.569027423858643,
|
| 2277 |
+
"learning_rate": 0.00017225484730386498,
|
| 2278 |
+
"loss": 2.7196646118164063,
|
| 2279 |
+
"step": 3450
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"epoch": 2.657935285053929,
|
| 2283 |
+
"step": 3450,
|
| 2284 |
+
"train/loss_commentary": 3.171079635620117
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 2.657935285053929,
|
| 2288 |
+
"step": 3450,
|
| 2289 |
+
"train/loss_grounding": 0.06897285580635071
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 2.677195685670262,
|
| 2293 |
+
"grad_norm": 5.05317497253418,
|
| 2294 |
+
"learning_rate": 0.0001718221799418485,
|
| 2295 |
+
"loss": 2.70868896484375,
|
| 2296 |
+
"step": 3475
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 2.677195685670262,
|
| 2300 |
+
"step": 3475,
|
| 2301 |
+
"train/loss_commentary": 2.217299699783325
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 2.677195685670262,
|
| 2305 |
+
"step": 3475,
|
| 2306 |
+
"train/loss_grounding": 0.012759953737258911
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 2.6964560862865947,
|
| 2310 |
+
"grad_norm": 5.0866875648498535,
|
| 2311 |
+
"learning_rate": 0.0001713867175562479,
|
| 2312 |
+
"loss": 2.7193145751953125,
|
| 2313 |
+
"step": 3500
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 2.6964560862865947,
|
| 2317 |
+
"step": 3500,
|
| 2318 |
+
"train/loss_commentary": 2.718380928039551
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 2.6964560862865947,
|
| 2322 |
+
"step": 3500,
|
| 2323 |
+
"train/loss_grounding": 0.8789747357368469
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 2.7157164869029273,
|
| 2327 |
+
"grad_norm": 5.8160810470581055,
|
| 2328 |
+
"learning_rate": 0.00017094847709346665,
|
| 2329 |
+
"loss": 2.6793386840820315,
|
| 2330 |
+
"step": 3525
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 2.7157164869029273,
|
| 2334 |
+
"step": 3525,
|
| 2335 |
+
"train/loss_commentary": 2.405125141143799
|
| 2336 |
+
},
|
| 2337 |
+
{
|
| 2338 |
+
"epoch": 2.7157164869029273,
|
| 2339 |
+
"step": 3525,
|
| 2340 |
+
"train/loss_grounding": 0.03566470369696617
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 2.7349768875192604,
|
| 2344 |
+
"grad_norm": 5.485836505889893,
|
| 2345 |
+
"learning_rate": 0.00017050747560801955,
|
| 2346 |
+
"loss": 2.649622802734375,
|
| 2347 |
+
"step": 3550
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 2.7349768875192604,
|
| 2351 |
+
"step": 3550,
|
| 2352 |
+
"train/loss_commentary": 2.4564809799194336
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 2.7349768875192604,
|
| 2356 |
+
"step": 3550,
|
| 2357 |
+
"train/loss_grounding": 0.4436793029308319
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 2.7542372881355934,
|
| 2361 |
+
"grad_norm": 5.137184143066406,
|
| 2362 |
+
"learning_rate": 0.00017006373026186896,
|
| 2363 |
+
"loss": 2.69102294921875,
|
| 2364 |
+
"step": 3575
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 2.7542372881355934,
|
| 2368 |
+
"step": 3575,
|
| 2369 |
+
"train/loss_commentary": 3.349208354949951
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 2.7542372881355934,
|
| 2373 |
+
"step": 3575,
|
| 2374 |
+
"train/loss_grounding": 0.3201281726360321
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 2.773497688751926,
|
| 2378 |
+
"grad_norm": 5.047833442687988,
|
| 2379 |
+
"learning_rate": 0.00016961725832375713,
|
| 2380 |
+
"loss": 2.700284423828125,
|
| 2381 |
+
"step": 3600
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 2.773497688751926,
|
| 2385 |
+
"step": 3600,
|
| 2386 |
+
"train/loss_commentary": 3.3294405937194824
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 2.773497688751926,
|
| 2390 |
+
"step": 3600,
|
| 2391 |
+
"train/loss_grounding": 0.16992099583148956
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 2.7927580893682586,
|
| 2395 |
+
"grad_norm": 5.728899955749512,
|
| 2396 |
+
"learning_rate": 0.0001691680771685339,
|
| 2397 |
+
"loss": 2.6935711669921876,
|
| 2398 |
+
"step": 3625
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 2.7927580893682586,
|
| 2402 |
+
"step": 3625,
|
| 2403 |
+
"train/loss_commentary": 2.841179847717285
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 2.7927580893682586,
|
| 2407 |
+
"step": 3625,
|
| 2408 |
+
"train/loss_grounding": 0.8771936893463135
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 2.8120184899845917,
|
| 2412 |
+
"grad_norm": 4.6923699378967285,
|
| 2413 |
+
"learning_rate": 0.0001687162042764808,
|
| 2414 |
+
"loss": 2.6620306396484374,
|
| 2415 |
+
"step": 3650
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 2.8120184899845917,
|
| 2419 |
+
"step": 3650,
|
| 2420 |
+
"train/loss_commentary": 2.829336643218994
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 2.8120184899845917,
|
| 2424 |
+
"step": 3650,
|
| 2425 |
+
"train/loss_grounding": 0.6903871297836304
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 2.8312788906009247,
|
| 2429 |
+
"grad_norm": 5.806859016418457,
|
| 2430 |
+
"learning_rate": 0.0001682616572326306,
|
| 2431 |
+
"loss": 2.706590576171875,
|
| 2432 |
+
"step": 3675
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 2.8312788906009247,
|
| 2436 |
+
"step": 3675,
|
| 2437 |
+
"train/loss_commentary": 2.7308096885681152
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 2.8505392912172574,
|
| 2441 |
+
"grad_norm": 5.348742961883545,
|
| 2442 |
+
"learning_rate": 0.00016780445372608312,
|
| 2443 |
+
"loss": 2.709290771484375,
|
| 2444 |
+
"step": 3700
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 2.8505392912172574,
|
| 2448 |
+
"step": 3700,
|
| 2449 |
+
"train/loss_commentary": 2.8261361122131348
|
| 2450 |
+
},
|
| 2451 |
+
{
|
| 2452 |
+
"epoch": 2.86979969183359,
|
| 2453 |
+
"grad_norm": 5.463099479675293,
|
| 2454 |
+
"learning_rate": 0.00016734461154931674,
|
| 2455 |
+
"loss": 2.635542297363281,
|
| 2456 |
+
"step": 3725
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 2.86979969183359,
|
| 2460 |
+
"step": 3725,
|
| 2461 |
+
"train/loss_commentary": 2.431018590927124
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 2.86979969183359,
|
| 2465 |
+
"step": 3725,
|
| 2466 |
+
"train/loss_grounding": 0.03573445603251457
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 2.889060092449923,
|
| 2470 |
+
"grad_norm": 4.804049968719482,
|
| 2471 |
+
"learning_rate": 0.00016688214859749594,
|
| 2472 |
+
"loss": 2.669443359375,
|
| 2473 |
+
"step": 3750
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 2.889060092449923,
|
| 2477 |
+
"step": 3750,
|
| 2478 |
+
"train/loss_commentary": 2.9765515327453613
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 2.9083204930662556,
|
| 2482 |
+
"grad_norm": 5.0031585693359375,
|
| 2483 |
+
"learning_rate": 0.00016641708286777514,
|
| 2484 |
+
"loss": 2.703349914550781,
|
| 2485 |
+
"step": 3775
|
| 2486 |
+
},
|
| 2487 |
+
{
|
| 2488 |
+
"epoch": 2.9083204930662556,
|
| 2489 |
+
"step": 3775,
|
| 2490 |
+
"train/loss_commentary": 2.6072275638580322
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 2.9083204930662556,
|
| 2494 |
+
"step": 3775,
|
| 2495 |
+
"train/loss_grounding": 0.21700447797775269
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 2.9275808936825887,
|
| 2499 |
+
"grad_norm": 5.817076683044434,
|
| 2500 |
+
"learning_rate": 0.00016594943245859805,
|
| 2501 |
+
"loss": 2.7027001953125,
|
| 2502 |
+
"step": 3800
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 2.9275808936825887,
|
| 2506 |
+
"step": 3800,
|
| 2507 |
+
"train/loss_commentary": 2.9400923252105713
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 2.9275808936825887,
|
| 2511 |
+
"step": 3800,
|
| 2512 |
+
"train/loss_grounding": 0.924263596534729
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 2.9468412942989213,
|
| 2516 |
+
"grad_norm": 5.80646276473999,
|
| 2517 |
+
"learning_rate": 0.0001654792155689935,
|
| 2518 |
+
"loss": 2.69057373046875,
|
| 2519 |
+
"step": 3825
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 2.9468412942989213,
|
| 2523 |
+
"step": 3825,
|
| 2524 |
+
"train/loss_commentary": 2.7612967491149902
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 2.9468412942989213,
|
| 2528 |
+
"step": 3825,
|
| 2529 |
+
"train/loss_grounding": 0.2775757312774658
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 2.9661016949152543,
|
| 2533 |
+
"grad_norm": 5.335476875305176,
|
| 2534 |
+
"learning_rate": 0.0001650064504978671,
|
| 2535 |
+
"loss": 2.682196044921875,
|
| 2536 |
+
"step": 3850
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 2.9661016949152543,
|
| 2540 |
+
"step": 3850,
|
| 2541 |
+
"train/loss_commentary": 3.0881881713867188
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 2.9661016949152543,
|
| 2545 |
+
"step": 3850,
|
| 2546 |
+
"train/loss_grounding": 0.15756088495254517
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 2.985362095531587,
|
| 2550 |
+
"grad_norm": 6.374530792236328,
|
| 2551 |
+
"learning_rate": 0.00016453115564328937,
|
| 2552 |
+
"loss": 2.6626876831054687,
|
| 2553 |
+
"step": 3875
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 2.985362095531587,
|
| 2557 |
+
"step": 3875,
|
| 2558 |
+
"train/loss_commentary": 2.18045711517334
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 2.985362095531587,
|
| 2562 |
+
"step": 3875,
|
| 2563 |
+
"train/loss_grounding": 0.31779760122299194
|
| 2564 |
+
},
|
| 2565 |
+
{
|
| 2566 |
+
"epoch": 3.0,
|
| 2567 |
+
"eval_loss": 2.92698335647583,
|
| 2568 |
+
"eval_runtime": 14.8562,
|
| 2569 |
+
"eval_samples_per_second": 126.21,
|
| 2570 |
+
"eval_steps_per_second": 2.019,
|
| 2571 |
+
"step": 3894
|
| 2572 |
+
}
|
| 2573 |
+
],
|
| 2574 |
+
"logging_steps": 25,
|
| 2575 |
+
"max_steps": 12980,
|
| 2576 |
+
"num_input_tokens_seen": 0,
|
| 2577 |
+
"num_train_epochs": 10,
|
| 2578 |
+
"save_steps": 500,
|
| 2579 |
+
"stateful_callbacks": {
|
| 2580 |
+
"TrainerControl": {
|
| 2581 |
+
"args": {
|
| 2582 |
+
"should_epoch_stop": false,
|
| 2583 |
+
"should_evaluate": false,
|
| 2584 |
+
"should_log": false,
|
| 2585 |
+
"should_save": true,
|
| 2586 |
+
"should_training_stop": false
|
| 2587 |
+
},
|
| 2588 |
+
"attributes": {}
|
| 2589 |
+
}
|
| 2590 |
+
},
|
| 2591 |
+
"total_flos": 0.0,
|
| 2592 |
+
"train_batch_size": 8,
|
| 2593 |
+
"trial_name": null,
|
| 2594 |
+
"trial_params": null
|
| 2595 |
+
}
|