Upload folder using huggingface_hub
Browse files- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- tokenizer.json +2 -2
- trainer_state.json +2181 -0
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6720fa1cff5f47b30cd428b79d20ec7e61f6f29f9bf66bfecc7de2b58d2c6c24
|
| 3 |
+
size 347119730
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81ce40ac181b7f7d522f257479605a2af1ccfc321f0c8aa1c1860a291649c3de
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c79c66b60a0abe3dc1f1792ced2b6c99f10b3ada4ba94ee60000ba5931c603a9
|
| 3 |
+
size 1064
|
tokenizer.json
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:038968f16122fee53c1cd8b410a01c21d9cf995cda80be434bb34e4376820027
|
| 3 |
+
size 17209500
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2181 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 6.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 300,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.02,
|
| 13 |
+
"grad_norm": 2.7934532165527344,
|
| 14 |
+
"learning_rate": 0.0002,
|
| 15 |
+
"loss": 2.3404,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.04,
|
| 20 |
+
"grad_norm": 1.5168005228042603,
|
| 21 |
+
"learning_rate": 0.0002,
|
| 22 |
+
"loss": 2.0804,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.06,
|
| 27 |
+
"grad_norm": 1.047807216644287,
|
| 28 |
+
"learning_rate": 0.0002,
|
| 29 |
+
"loss": 1.9184,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.08,
|
| 34 |
+
"grad_norm": 1.041599988937378,
|
| 35 |
+
"learning_rate": 0.0002,
|
| 36 |
+
"loss": 2.0393,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.1,
|
| 41 |
+
"grad_norm": 0.8074644804000854,
|
| 42 |
+
"learning_rate": 0.0002,
|
| 43 |
+
"loss": 2.1779,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12,
|
| 48 |
+
"grad_norm": 0.7784727811813354,
|
| 49 |
+
"learning_rate": 0.0002,
|
| 50 |
+
"loss": 2.1583,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.14,
|
| 55 |
+
"grad_norm": 0.8535248637199402,
|
| 56 |
+
"learning_rate": 0.0002,
|
| 57 |
+
"loss": 2.0766,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.16,
|
| 62 |
+
"grad_norm": 0.8507911562919617,
|
| 63 |
+
"learning_rate": 0.0002,
|
| 64 |
+
"loss": 2.1175,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.18,
|
| 69 |
+
"grad_norm": 0.8497746586799622,
|
| 70 |
+
"learning_rate": 0.0002,
|
| 71 |
+
"loss": 2.1156,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.2,
|
| 76 |
+
"grad_norm": 0.9205197691917419,
|
| 77 |
+
"learning_rate": 0.0002,
|
| 78 |
+
"loss": 2.1782,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.22,
|
| 83 |
+
"grad_norm": 1.2332911491394043,
|
| 84 |
+
"learning_rate": 0.0002,
|
| 85 |
+
"loss": 2.0536,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.24,
|
| 90 |
+
"grad_norm": 1.3457396030426025,
|
| 91 |
+
"learning_rate": 0.0002,
|
| 92 |
+
"loss": 2.2186,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.26,
|
| 97 |
+
"grad_norm": 0.6494730114936829,
|
| 98 |
+
"learning_rate": 0.0002,
|
| 99 |
+
"loss": 1.8011,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.28,
|
| 104 |
+
"grad_norm": 0.6363134980201721,
|
| 105 |
+
"learning_rate": 0.0002,
|
| 106 |
+
"loss": 1.8819,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.3,
|
| 111 |
+
"grad_norm": 0.7927612662315369,
|
| 112 |
+
"learning_rate": 0.0002,
|
| 113 |
+
"loss": 1.8845,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.32,
|
| 118 |
+
"grad_norm": 0.7082176804542542,
|
| 119 |
+
"learning_rate": 0.0002,
|
| 120 |
+
"loss": 1.8748,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.34,
|
| 125 |
+
"grad_norm": 0.861709713935852,
|
| 126 |
+
"learning_rate": 0.0002,
|
| 127 |
+
"loss": 1.8777,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.36,
|
| 132 |
+
"grad_norm": 0.7901681661605835,
|
| 133 |
+
"learning_rate": 0.0002,
|
| 134 |
+
"loss": 1.8811,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.38,
|
| 139 |
+
"grad_norm": 0.7719288468360901,
|
| 140 |
+
"learning_rate": 0.0002,
|
| 141 |
+
"loss": 1.8898,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.4,
|
| 146 |
+
"grad_norm": 1.027469277381897,
|
| 147 |
+
"learning_rate": 0.0002,
|
| 148 |
+
"loss": 2.0089,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.42,
|
| 153 |
+
"grad_norm": 0.9486727714538574,
|
| 154 |
+
"learning_rate": 0.0002,
|
| 155 |
+
"loss": 1.968,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.44,
|
| 160 |
+
"grad_norm": 0.9629890322685242,
|
| 161 |
+
"learning_rate": 0.0002,
|
| 162 |
+
"loss": 2.0568,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.46,
|
| 167 |
+
"grad_norm": 1.033793568611145,
|
| 168 |
+
"learning_rate": 0.0002,
|
| 169 |
+
"loss": 1.8401,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.48,
|
| 174 |
+
"grad_norm": 1.3298218250274658,
|
| 175 |
+
"learning_rate": 0.0002,
|
| 176 |
+
"loss": 1.9165,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5,
|
| 181 |
+
"grad_norm": 0.6936089396476746,
|
| 182 |
+
"learning_rate": 0.0002,
|
| 183 |
+
"loss": 1.6614,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.52,
|
| 188 |
+
"grad_norm": 0.6136096119880676,
|
| 189 |
+
"learning_rate": 0.0002,
|
| 190 |
+
"loss": 1.8167,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.54,
|
| 195 |
+
"grad_norm": 0.6043046712875366,
|
| 196 |
+
"learning_rate": 0.0002,
|
| 197 |
+
"loss": 1.7217,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.56,
|
| 202 |
+
"grad_norm": 0.6395452618598938,
|
| 203 |
+
"learning_rate": 0.0002,
|
| 204 |
+
"loss": 1.7353,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.58,
|
| 209 |
+
"grad_norm": 0.6829009056091309,
|
| 210 |
+
"learning_rate": 0.0002,
|
| 211 |
+
"loss": 1.7708,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6,
|
| 216 |
+
"grad_norm": 0.8561712503433228,
|
| 217 |
+
"learning_rate": 0.0002,
|
| 218 |
+
"loss": 1.774,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.62,
|
| 223 |
+
"grad_norm": 0.7594190239906311,
|
| 224 |
+
"learning_rate": 0.0002,
|
| 225 |
+
"loss": 1.8788,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.64,
|
| 230 |
+
"grad_norm": 0.867341160774231,
|
| 231 |
+
"learning_rate": 0.0002,
|
| 232 |
+
"loss": 1.8708,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.66,
|
| 237 |
+
"grad_norm": 0.9393973350524902,
|
| 238 |
+
"learning_rate": 0.0002,
|
| 239 |
+
"loss": 1.9839,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.68,
|
| 244 |
+
"grad_norm": 1.0540133714675903,
|
| 245 |
+
"learning_rate": 0.0002,
|
| 246 |
+
"loss": 1.7637,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7,
|
| 251 |
+
"grad_norm": 1.2020256519317627,
|
| 252 |
+
"learning_rate": 0.0002,
|
| 253 |
+
"loss": 1.9187,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.72,
|
| 258 |
+
"grad_norm": 1.7588919401168823,
|
| 259 |
+
"learning_rate": 0.0002,
|
| 260 |
+
"loss": 1.5851,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.74,
|
| 265 |
+
"grad_norm": 0.9404975175857544,
|
| 266 |
+
"learning_rate": 0.0002,
|
| 267 |
+
"loss": 1.7903,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.76,
|
| 272 |
+
"grad_norm": 0.7744253873825073,
|
| 273 |
+
"learning_rate": 0.0002,
|
| 274 |
+
"loss": 1.7476,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.78,
|
| 279 |
+
"grad_norm": 0.7260447144508362,
|
| 280 |
+
"learning_rate": 0.0002,
|
| 281 |
+
"loss": 1.5547,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8,
|
| 286 |
+
"grad_norm": 0.9214150905609131,
|
| 287 |
+
"learning_rate": 0.0002,
|
| 288 |
+
"loss": 1.6342,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.82,
|
| 293 |
+
"grad_norm": 0.834932267665863,
|
| 294 |
+
"learning_rate": 0.0002,
|
| 295 |
+
"loss": 1.6577,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.84,
|
| 300 |
+
"grad_norm": 0.8663449883460999,
|
| 301 |
+
"learning_rate": 0.0002,
|
| 302 |
+
"loss": 1.7133,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.86,
|
| 307 |
+
"grad_norm": 0.9534509181976318,
|
| 308 |
+
"learning_rate": 0.0002,
|
| 309 |
+
"loss": 1.7999,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.88,
|
| 314 |
+
"grad_norm": 1.058899164199829,
|
| 315 |
+
"learning_rate": 0.0002,
|
| 316 |
+
"loss": 1.5964,
|
| 317 |
+
"step": 44
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9,
|
| 321 |
+
"grad_norm": 1.1835004091262817,
|
| 322 |
+
"learning_rate": 0.0002,
|
| 323 |
+
"loss": 1.7596,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.92,
|
| 328 |
+
"grad_norm": 1.2041128873825073,
|
| 329 |
+
"learning_rate": 0.0002,
|
| 330 |
+
"loss": 1.6529,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.94,
|
| 335 |
+
"grad_norm": 1.5300588607788086,
|
| 336 |
+
"learning_rate": 0.0002,
|
| 337 |
+
"loss": 1.7852,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.96,
|
| 342 |
+
"grad_norm": 1.6037429571151733,
|
| 343 |
+
"learning_rate": 0.0002,
|
| 344 |
+
"loss": 1.6873,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.98,
|
| 349 |
+
"grad_norm": 0.6437931656837463,
|
| 350 |
+
"learning_rate": 0.0002,
|
| 351 |
+
"loss": 1.5729,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 1.0,
|
| 356 |
+
"grad_norm": 1.1996123790740967,
|
| 357 |
+
"learning_rate": 0.0002,
|
| 358 |
+
"loss": 1.5609,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.0,
|
| 363 |
+
"eval_loss": 1.6186010837554932,
|
| 364 |
+
"eval_runtime": 565.3333,
|
| 365 |
+
"eval_samples_per_second": 0.708,
|
| 366 |
+
"eval_steps_per_second": 0.177,
|
| 367 |
+
"step": 50
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 1.02,
|
| 371 |
+
"grad_norm": 0.6480159163475037,
|
| 372 |
+
"learning_rate": 0.0002,
|
| 373 |
+
"loss": 1.5754,
|
| 374 |
+
"step": 51
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.04,
|
| 378 |
+
"grad_norm": 0.8269457817077637,
|
| 379 |
+
"learning_rate": 0.0002,
|
| 380 |
+
"loss": 1.4743,
|
| 381 |
+
"step": 52
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 1.06,
|
| 385 |
+
"grad_norm": 1.1054280996322632,
|
| 386 |
+
"learning_rate": 0.0002,
|
| 387 |
+
"loss": 1.3447,
|
| 388 |
+
"step": 53
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.08,
|
| 392 |
+
"grad_norm": 0.9144314527511597,
|
| 393 |
+
"learning_rate": 0.0002,
|
| 394 |
+
"loss": 1.3652,
|
| 395 |
+
"step": 54
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.1,
|
| 399 |
+
"grad_norm": 0.8429620862007141,
|
| 400 |
+
"learning_rate": 0.0002,
|
| 401 |
+
"loss": 1.4696,
|
| 402 |
+
"step": 55
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.12,
|
| 406 |
+
"grad_norm": 1.3091776371002197,
|
| 407 |
+
"learning_rate": 0.0002,
|
| 408 |
+
"loss": 1.3109,
|
| 409 |
+
"step": 56
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.1400000000000001,
|
| 413 |
+
"grad_norm": 1.2086460590362549,
|
| 414 |
+
"learning_rate": 0.0002,
|
| 415 |
+
"loss": 1.3424,
|
| 416 |
+
"step": 57
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.16,
|
| 420 |
+
"grad_norm": 1.1823766231536865,
|
| 421 |
+
"learning_rate": 0.0002,
|
| 422 |
+
"loss": 1.213,
|
| 423 |
+
"step": 58
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.18,
|
| 427 |
+
"grad_norm": 1.817803144454956,
|
| 428 |
+
"learning_rate": 0.0002,
|
| 429 |
+
"loss": 1.2911,
|
| 430 |
+
"step": 59
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.2,
|
| 434 |
+
"grad_norm": 1.2870073318481445,
|
| 435 |
+
"learning_rate": 0.0002,
|
| 436 |
+
"loss": 1.2712,
|
| 437 |
+
"step": 60
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.22,
|
| 441 |
+
"grad_norm": 1.2424544095993042,
|
| 442 |
+
"learning_rate": 0.0002,
|
| 443 |
+
"loss": 1.2292,
|
| 444 |
+
"step": 61
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.24,
|
| 448 |
+
"grad_norm": 1.4258471727371216,
|
| 449 |
+
"learning_rate": 0.0002,
|
| 450 |
+
"loss": 1.0814,
|
| 451 |
+
"step": 62
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.26,
|
| 455 |
+
"grad_norm": 1.1297271251678467,
|
| 456 |
+
"learning_rate": 0.0002,
|
| 457 |
+
"loss": 1.5303,
|
| 458 |
+
"step": 63
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.28,
|
| 462 |
+
"grad_norm": 0.8728504776954651,
|
| 463 |
+
"learning_rate": 0.0002,
|
| 464 |
+
"loss": 1.4354,
|
| 465 |
+
"step": 64
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.3,
|
| 469 |
+
"grad_norm": 0.7809789776802063,
|
| 470 |
+
"learning_rate": 0.0002,
|
| 471 |
+
"loss": 1.4398,
|
| 472 |
+
"step": 65
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.32,
|
| 476 |
+
"grad_norm": 0.844166100025177,
|
| 477 |
+
"learning_rate": 0.0002,
|
| 478 |
+
"loss": 1.2171,
|
| 479 |
+
"step": 66
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.34,
|
| 483 |
+
"grad_norm": 0.8636218905448914,
|
| 484 |
+
"learning_rate": 0.0002,
|
| 485 |
+
"loss": 1.23,
|
| 486 |
+
"step": 67
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.3599999999999999,
|
| 490 |
+
"grad_norm": 0.9831591248512268,
|
| 491 |
+
"learning_rate": 0.0002,
|
| 492 |
+
"loss": 1.2496,
|
| 493 |
+
"step": 68
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.38,
|
| 497 |
+
"grad_norm": 1.4268325567245483,
|
| 498 |
+
"learning_rate": 0.0002,
|
| 499 |
+
"loss": 1.2354,
|
| 500 |
+
"step": 69
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.4,
|
| 504 |
+
"grad_norm": 1.6133723258972168,
|
| 505 |
+
"learning_rate": 0.0002,
|
| 506 |
+
"loss": 1.2123,
|
| 507 |
+
"step": 70
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.42,
|
| 511 |
+
"grad_norm": 1.5462720394134521,
|
| 512 |
+
"learning_rate": 0.0002,
|
| 513 |
+
"loss": 1.0258,
|
| 514 |
+
"step": 71
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.44,
|
| 518 |
+
"grad_norm": 1.1962395906448364,
|
| 519 |
+
"learning_rate": 0.0002,
|
| 520 |
+
"loss": 1.0858,
|
| 521 |
+
"step": 72
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.46,
|
| 525 |
+
"grad_norm": 1.413921594619751,
|
| 526 |
+
"learning_rate": 0.0002,
|
| 527 |
+
"loss": 1.0169,
|
| 528 |
+
"step": 73
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.48,
|
| 532 |
+
"grad_norm": 1.442657470703125,
|
| 533 |
+
"learning_rate": 0.0002,
|
| 534 |
+
"loss": 0.9553,
|
| 535 |
+
"step": 74
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.5,
|
| 539 |
+
"grad_norm": 0.9919085502624512,
|
| 540 |
+
"learning_rate": 0.0002,
|
| 541 |
+
"loss": 1.2394,
|
| 542 |
+
"step": 75
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.52,
|
| 546 |
+
"grad_norm": 0.988468587398529,
|
| 547 |
+
"learning_rate": 0.0002,
|
| 548 |
+
"loss": 1.3642,
|
| 549 |
+
"step": 76
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.54,
|
| 553 |
+
"grad_norm": 0.9793186187744141,
|
| 554 |
+
"learning_rate": 0.0002,
|
| 555 |
+
"loss": 1.2818,
|
| 556 |
+
"step": 77
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.56,
|
| 560 |
+
"grad_norm": 0.7799855470657349,
|
| 561 |
+
"learning_rate": 0.0002,
|
| 562 |
+
"loss": 1.1705,
|
| 563 |
+
"step": 78
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.58,
|
| 567 |
+
"grad_norm": 0.8288784027099609,
|
| 568 |
+
"learning_rate": 0.0002,
|
| 569 |
+
"loss": 1.0484,
|
| 570 |
+
"step": 79
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.6,
|
| 574 |
+
"grad_norm": 1.064773440361023,
|
| 575 |
+
"learning_rate": 0.0002,
|
| 576 |
+
"loss": 1.1608,
|
| 577 |
+
"step": 80
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.62,
|
| 581 |
+
"grad_norm": 1.0099600553512573,
|
| 582 |
+
"learning_rate": 0.0002,
|
| 583 |
+
"loss": 1.1873,
|
| 584 |
+
"step": 81
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.6400000000000001,
|
| 588 |
+
"grad_norm": 1.9040124416351318,
|
| 589 |
+
"learning_rate": 0.0002,
|
| 590 |
+
"loss": 0.9739,
|
| 591 |
+
"step": 82
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.6600000000000001,
|
| 595 |
+
"grad_norm": 1.2448644638061523,
|
| 596 |
+
"learning_rate": 0.0002,
|
| 597 |
+
"loss": 0.9418,
|
| 598 |
+
"step": 83
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.6800000000000002,
|
| 602 |
+
"grad_norm": 1.2129086256027222,
|
| 603 |
+
"learning_rate": 0.0002,
|
| 604 |
+
"loss": 0.8821,
|
| 605 |
+
"step": 84
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.7,
|
| 609 |
+
"grad_norm": 1.6727265119552612,
|
| 610 |
+
"learning_rate": 0.0002,
|
| 611 |
+
"loss": 0.9965,
|
| 612 |
+
"step": 85
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.72,
|
| 616 |
+
"grad_norm": 1.6569440364837646,
|
| 617 |
+
"learning_rate": 0.0002,
|
| 618 |
+
"loss": 0.9182,
|
| 619 |
+
"step": 86
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.74,
|
| 623 |
+
"grad_norm": 0.8596146702766418,
|
| 624 |
+
"learning_rate": 0.0002,
|
| 625 |
+
"loss": 1.3188,
|
| 626 |
+
"step": 87
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.76,
|
| 630 |
+
"grad_norm": 0.8928490281105042,
|
| 631 |
+
"learning_rate": 0.0002,
|
| 632 |
+
"loss": 1.3601,
|
| 633 |
+
"step": 88
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.78,
|
| 637 |
+
"grad_norm": 0.7409713268280029,
|
| 638 |
+
"learning_rate": 0.0002,
|
| 639 |
+
"loss": 1.1212,
|
| 640 |
+
"step": 89
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.8,
|
| 644 |
+
"grad_norm": 0.8979334831237793,
|
| 645 |
+
"learning_rate": 0.0002,
|
| 646 |
+
"loss": 1.4162,
|
| 647 |
+
"step": 90
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.8199999999999998,
|
| 651 |
+
"grad_norm": 0.979978621006012,
|
| 652 |
+
"learning_rate": 0.0002,
|
| 653 |
+
"loss": 1.1969,
|
| 654 |
+
"step": 91
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.8399999999999999,
|
| 658 |
+
"grad_norm": 0.9733594059944153,
|
| 659 |
+
"learning_rate": 0.0002,
|
| 660 |
+
"loss": 1.0468,
|
| 661 |
+
"step": 92
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.8599999999999999,
|
| 665 |
+
"grad_norm": 0.9226842522621155,
|
| 666 |
+
"learning_rate": 0.0002,
|
| 667 |
+
"loss": 1.1807,
|
| 668 |
+
"step": 93
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.88,
|
| 672 |
+
"grad_norm": 1.1638745069503784,
|
| 673 |
+
"learning_rate": 0.0002,
|
| 674 |
+
"loss": 1.139,
|
| 675 |
+
"step": 94
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.9,
|
| 679 |
+
"grad_norm": 1.5604937076568604,
|
| 680 |
+
"learning_rate": 0.0002,
|
| 681 |
+
"loss": 1.1872,
|
| 682 |
+
"step": 95
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.92,
|
| 686 |
+
"grad_norm": 1.3674428462982178,
|
| 687 |
+
"learning_rate": 0.0002,
|
| 688 |
+
"loss": 1.1865,
|
| 689 |
+
"step": 96
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.94,
|
| 693 |
+
"grad_norm": 1.8469598293304443,
|
| 694 |
+
"learning_rate": 0.0002,
|
| 695 |
+
"loss": 1.0469,
|
| 696 |
+
"step": 97
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.96,
|
| 700 |
+
"grad_norm": 1.3148952722549438,
|
| 701 |
+
"learning_rate": 0.0002,
|
| 702 |
+
"loss": 0.9915,
|
| 703 |
+
"step": 98
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.98,
|
| 707 |
+
"grad_norm": 1.599141001701355,
|
| 708 |
+
"learning_rate": 0.0002,
|
| 709 |
+
"loss": 1.2296,
|
| 710 |
+
"step": 99
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 2.0,
|
| 714 |
+
"grad_norm": 1.3382114171981812,
|
| 715 |
+
"learning_rate": 0.0002,
|
| 716 |
+
"loss": 1.1813,
|
| 717 |
+
"step": 100
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 2.0,
|
| 721 |
+
"eval_loss": 1.409305453300476,
|
| 722 |
+
"eval_runtime": 565.9517,
|
| 723 |
+
"eval_samples_per_second": 0.707,
|
| 724 |
+
"eval_steps_per_second": 0.177,
|
| 725 |
+
"step": 100
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 2.02,
|
| 729 |
+
"grad_norm": 1.0162380933761597,
|
| 730 |
+
"learning_rate": 0.0002,
|
| 731 |
+
"loss": 1.1481,
|
| 732 |
+
"step": 101
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 2.04,
|
| 736 |
+
"grad_norm": 0.7402092814445496,
|
| 737 |
+
"learning_rate": 0.0002,
|
| 738 |
+
"loss": 1.0086,
|
| 739 |
+
"step": 102
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 2.06,
|
| 743 |
+
"grad_norm": 0.8824872970581055,
|
| 744 |
+
"learning_rate": 0.0002,
|
| 745 |
+
"loss": 1.0588,
|
| 746 |
+
"step": 103
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 2.08,
|
| 750 |
+
"grad_norm": 0.7582442760467529,
|
| 751 |
+
"learning_rate": 0.0002,
|
| 752 |
+
"loss": 0.8181,
|
| 753 |
+
"step": 104
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 2.1,
|
| 757 |
+
"grad_norm": 1.0200812816619873,
|
| 758 |
+
"learning_rate": 0.0002,
|
| 759 |
+
"loss": 0.9041,
|
| 760 |
+
"step": 105
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 2.12,
|
| 764 |
+
"grad_norm": 1.08174467086792,
|
| 765 |
+
"learning_rate": 0.0002,
|
| 766 |
+
"loss": 0.8479,
|
| 767 |
+
"step": 106
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 2.14,
|
| 771 |
+
"grad_norm": 1.01225745677948,
|
| 772 |
+
"learning_rate": 0.0002,
|
| 773 |
+
"loss": 0.7659,
|
| 774 |
+
"step": 107
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 2.16,
|
| 778 |
+
"grad_norm": 1.2194840908050537,
|
| 779 |
+
"learning_rate": 0.0002,
|
| 780 |
+
"loss": 0.7926,
|
| 781 |
+
"step": 108
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 2.18,
|
| 785 |
+
"grad_norm": 1.0519524812698364,
|
| 786 |
+
"learning_rate": 0.0002,
|
| 787 |
+
"loss": 0.6604,
|
| 788 |
+
"step": 109
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.2,
|
| 792 |
+
"grad_norm": 1.2860150337219238,
|
| 793 |
+
"learning_rate": 0.0002,
|
| 794 |
+
"loss": 0.663,
|
| 795 |
+
"step": 110
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 2.22,
|
| 799 |
+
"grad_norm": 1.5521994829177856,
|
| 800 |
+
"learning_rate": 0.0002,
|
| 801 |
+
"loss": 0.7791,
|
| 802 |
+
"step": 111
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.24,
|
| 806 |
+
"grad_norm": 1.455283284187317,
|
| 807 |
+
"learning_rate": 0.0002,
|
| 808 |
+
"loss": 0.5112,
|
| 809 |
+
"step": 112
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 2.26,
|
| 813 |
+
"grad_norm": 1.7097278833389282,
|
| 814 |
+
"learning_rate": 0.0002,
|
| 815 |
+
"loss": 1.2219,
|
| 816 |
+
"step": 113
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 2.2800000000000002,
|
| 820 |
+
"grad_norm": 1.5385531187057495,
|
| 821 |
+
"learning_rate": 0.0002,
|
| 822 |
+
"loss": 1.1261,
|
| 823 |
+
"step": 114
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 2.3,
|
| 827 |
+
"grad_norm": 1.0525436401367188,
|
| 828 |
+
"learning_rate": 0.0002,
|
| 829 |
+
"loss": 0.86,
|
| 830 |
+
"step": 115
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 2.32,
|
| 834 |
+
"grad_norm": 1.0388120412826538,
|
| 835 |
+
"learning_rate": 0.0002,
|
| 836 |
+
"loss": 0.9022,
|
| 837 |
+
"step": 116
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.34,
|
| 841 |
+
"grad_norm": 1.060497760772705,
|
| 842 |
+
"learning_rate": 0.0002,
|
| 843 |
+
"loss": 0.9265,
|
| 844 |
+
"step": 117
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 2.36,
|
| 848 |
+
"grad_norm": 1.0629950761795044,
|
| 849 |
+
"learning_rate": 0.0002,
|
| 850 |
+
"loss": 0.7222,
|
| 851 |
+
"step": 118
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 2.38,
|
| 855 |
+
"grad_norm": 1.2574018239974976,
|
| 856 |
+
"learning_rate": 0.0002,
|
| 857 |
+
"loss": 0.7952,
|
| 858 |
+
"step": 119
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 2.4,
|
| 862 |
+
"grad_norm": 1.0951610803604126,
|
| 863 |
+
"learning_rate": 0.0002,
|
| 864 |
+
"loss": 0.6647,
|
| 865 |
+
"step": 120
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 2.42,
|
| 869 |
+
"grad_norm": 1.46285879611969,
|
| 870 |
+
"learning_rate": 0.0002,
|
| 871 |
+
"loss": 0.7845,
|
| 872 |
+
"step": 121
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 2.44,
|
| 876 |
+
"grad_norm": 1.3611388206481934,
|
| 877 |
+
"learning_rate": 0.0002,
|
| 878 |
+
"loss": 0.7084,
|
| 879 |
+
"step": 122
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 2.46,
|
| 883 |
+
"grad_norm": 1.6670907735824585,
|
| 884 |
+
"learning_rate": 0.0002,
|
| 885 |
+
"loss": 0.6594,
|
| 886 |
+
"step": 123
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 2.48,
|
| 890 |
+
"grad_norm": 2.1525955200195312,
|
| 891 |
+
"learning_rate": 0.0002,
|
| 892 |
+
"loss": 0.6401,
|
| 893 |
+
"step": 124
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 2.5,
|
| 897 |
+
"grad_norm": 2.5126793384552,
|
| 898 |
+
"learning_rate": 0.0002,
|
| 899 |
+
"loss": 0.926,
|
| 900 |
+
"step": 125
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 2.52,
|
| 904 |
+
"grad_norm": 1.800521969795227,
|
| 905 |
+
"learning_rate": 0.0002,
|
| 906 |
+
"loss": 1.0936,
|
| 907 |
+
"step": 126
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 2.54,
|
| 911 |
+
"grad_norm": 1.0617576837539673,
|
| 912 |
+
"learning_rate": 0.0002,
|
| 913 |
+
"loss": 0.8052,
|
| 914 |
+
"step": 127
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 2.56,
|
| 918 |
+
"grad_norm": 1.0823312997817993,
|
| 919 |
+
"learning_rate": 0.0002,
|
| 920 |
+
"loss": 0.9443,
|
| 921 |
+
"step": 128
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 2.58,
|
| 925 |
+
"grad_norm": 1.2193264961242676,
|
| 926 |
+
"learning_rate": 0.0002,
|
| 927 |
+
"loss": 0.7955,
|
| 928 |
+
"step": 129
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 2.6,
|
| 932 |
+
"grad_norm": 1.0502954721450806,
|
| 933 |
+
"learning_rate": 0.0002,
|
| 934 |
+
"loss": 0.8365,
|
| 935 |
+
"step": 130
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 2.62,
|
| 939 |
+
"grad_norm": 1.1898560523986816,
|
| 940 |
+
"learning_rate": 0.0002,
|
| 941 |
+
"loss": 0.9706,
|
| 942 |
+
"step": 131
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 2.64,
|
| 946 |
+
"grad_norm": 1.1076680421829224,
|
| 947 |
+
"learning_rate": 0.0002,
|
| 948 |
+
"loss": 0.7529,
|
| 949 |
+
"step": 132
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 2.66,
|
| 953 |
+
"grad_norm": 1.3826709985733032,
|
| 954 |
+
"learning_rate": 0.0002,
|
| 955 |
+
"loss": 0.7474,
|
| 956 |
+
"step": 133
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 2.68,
|
| 960 |
+
"grad_norm": 1.2504832744598389,
|
| 961 |
+
"learning_rate": 0.0002,
|
| 962 |
+
"loss": 0.7086,
|
| 963 |
+
"step": 134
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.7,
|
| 967 |
+
"grad_norm": 1.6292765140533447,
|
| 968 |
+
"learning_rate": 0.0002,
|
| 969 |
+
"loss": 0.6305,
|
| 970 |
+
"step": 135
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 2.7199999999999998,
|
| 974 |
+
"grad_norm": 1.9603074789047241,
|
| 975 |
+
"learning_rate": 0.0002,
|
| 976 |
+
"loss": 0.6834,
|
| 977 |
+
"step": 136
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 2.74,
|
| 981 |
+
"grad_norm": 2.202030897140503,
|
| 982 |
+
"learning_rate": 0.0002,
|
| 983 |
+
"loss": 1.1712,
|
| 984 |
+
"step": 137
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.76,
|
| 988 |
+
"grad_norm": 1.6344685554504395,
|
| 989 |
+
"learning_rate": 0.0002,
|
| 990 |
+
"loss": 1.0772,
|
| 991 |
+
"step": 138
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.7800000000000002,
|
| 995 |
+
"grad_norm": 1.3579537868499756,
|
| 996 |
+
"learning_rate": 0.0002,
|
| 997 |
+
"loss": 0.8803,
|
| 998 |
+
"step": 139
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.8,
|
| 1002 |
+
"grad_norm": 1.0554553270339966,
|
| 1003 |
+
"learning_rate": 0.0002,
|
| 1004 |
+
"loss": 0.9222,
|
| 1005 |
+
"step": 140
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 2.82,
|
| 1009 |
+
"grad_norm": 0.9431642889976501,
|
| 1010 |
+
"learning_rate": 0.0002,
|
| 1011 |
+
"loss": 0.8031,
|
| 1012 |
+
"step": 141
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 2.84,
|
| 1016 |
+
"grad_norm": 1.0826098918914795,
|
| 1017 |
+
"learning_rate": 0.0002,
|
| 1018 |
+
"loss": 0.8259,
|
| 1019 |
+
"step": 142
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 2.86,
|
| 1023 |
+
"grad_norm": 1.24959135055542,
|
| 1024 |
+
"learning_rate": 0.0002,
|
| 1025 |
+
"loss": 0.7957,
|
| 1026 |
+
"step": 143
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 2.88,
|
| 1030 |
+
"grad_norm": 1.1057368516921997,
|
| 1031 |
+
"learning_rate": 0.0002,
|
| 1032 |
+
"loss": 0.7079,
|
| 1033 |
+
"step": 144
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 2.9,
|
| 1037 |
+
"grad_norm": 1.144061803817749,
|
| 1038 |
+
"learning_rate": 0.0002,
|
| 1039 |
+
"loss": 0.7165,
|
| 1040 |
+
"step": 145
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 2.92,
|
| 1044 |
+
"grad_norm": 1.0690631866455078,
|
| 1045 |
+
"learning_rate": 0.0002,
|
| 1046 |
+
"loss": 0.601,
|
| 1047 |
+
"step": 146
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 2.94,
|
| 1051 |
+
"grad_norm": 1.292758584022522,
|
| 1052 |
+
"learning_rate": 0.0002,
|
| 1053 |
+
"loss": 0.7191,
|
| 1054 |
+
"step": 147
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 2.96,
|
| 1058 |
+
"grad_norm": 1.729408860206604,
|
| 1059 |
+
"learning_rate": 0.0002,
|
| 1060 |
+
"loss": 0.5851,
|
| 1061 |
+
"step": 148
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 2.98,
|
| 1065 |
+
"grad_norm": 2.078197717666626,
|
| 1066 |
+
"learning_rate": 0.0002,
|
| 1067 |
+
"loss": 0.8942,
|
| 1068 |
+
"step": 149
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 3.0,
|
| 1072 |
+
"grad_norm": 2.0007128715515137,
|
| 1073 |
+
"learning_rate": 0.0002,
|
| 1074 |
+
"loss": 0.7139,
|
| 1075 |
+
"step": 150
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 3.0,
|
| 1079 |
+
"eval_loss": 1.4874461889266968,
|
| 1080 |
+
"eval_runtime": 566.1506,
|
| 1081 |
+
"eval_samples_per_second": 0.707,
|
| 1082 |
+
"eval_steps_per_second": 0.177,
|
| 1083 |
+
"step": 150
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 3.02,
|
| 1087 |
+
"grad_norm": 1.024670958518982,
|
| 1088 |
+
"learning_rate": 0.0002,
|
| 1089 |
+
"loss": 0.8969,
|
| 1090 |
+
"step": 151
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 3.04,
|
| 1094 |
+
"grad_norm": 0.9738882184028625,
|
| 1095 |
+
"learning_rate": 0.0002,
|
| 1096 |
+
"loss": 0.9056,
|
| 1097 |
+
"step": 152
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 3.06,
|
| 1101 |
+
"grad_norm": 0.9969688653945923,
|
| 1102 |
+
"learning_rate": 0.0002,
|
| 1103 |
+
"loss": 0.6676,
|
| 1104 |
+
"step": 153
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 3.08,
|
| 1108 |
+
"grad_norm": 1.12136971950531,
|
| 1109 |
+
"learning_rate": 0.0002,
|
| 1110 |
+
"loss": 0.5916,
|
| 1111 |
+
"step": 154
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 3.1,
|
| 1115 |
+
"grad_norm": 1.3517699241638184,
|
| 1116 |
+
"learning_rate": 0.0002,
|
| 1117 |
+
"loss": 0.5908,
|
| 1118 |
+
"step": 155
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 3.12,
|
| 1122 |
+
"grad_norm": 1.5965360403060913,
|
| 1123 |
+
"learning_rate": 0.0002,
|
| 1124 |
+
"loss": 0.6148,
|
| 1125 |
+
"step": 156
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 3.14,
|
| 1129 |
+
"grad_norm": 1.3009252548217773,
|
| 1130 |
+
"learning_rate": 0.0002,
|
| 1131 |
+
"loss": 0.4902,
|
| 1132 |
+
"step": 157
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 3.16,
|
| 1136 |
+
"grad_norm": 1.2742400169372559,
|
| 1137 |
+
"learning_rate": 0.0002,
|
| 1138 |
+
"loss": 0.4515,
|
| 1139 |
+
"step": 158
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 3.18,
|
| 1143 |
+
"grad_norm": 1.2994771003723145,
|
| 1144 |
+
"learning_rate": 0.0002,
|
| 1145 |
+
"loss": 0.416,
|
| 1146 |
+
"step": 159
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 3.2,
|
| 1150 |
+
"grad_norm": 1.3306324481964111,
|
| 1151 |
+
"learning_rate": 0.0002,
|
| 1152 |
+
"loss": 0.4144,
|
| 1153 |
+
"step": 160
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 3.22,
|
| 1157 |
+
"grad_norm": 1.5406475067138672,
|
| 1158 |
+
"learning_rate": 0.0002,
|
| 1159 |
+
"loss": 0.4256,
|
| 1160 |
+
"step": 161
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 3.24,
|
| 1164 |
+
"grad_norm": 1.584506630897522,
|
| 1165 |
+
"learning_rate": 0.0002,
|
| 1166 |
+
"loss": 0.4511,
|
| 1167 |
+
"step": 162
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 3.26,
|
| 1171 |
+
"grad_norm": 1.6618622541427612,
|
| 1172 |
+
"learning_rate": 0.0002,
|
| 1173 |
+
"loss": 0.8865,
|
| 1174 |
+
"step": 163
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 3.2800000000000002,
|
| 1178 |
+
"grad_norm": 1.6019847393035889,
|
| 1179 |
+
"learning_rate": 0.0002,
|
| 1180 |
+
"loss": 0.7339,
|
| 1181 |
+
"step": 164
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 3.3,
|
| 1185 |
+
"grad_norm": 1.1740251779556274,
|
| 1186 |
+
"learning_rate": 0.0002,
|
| 1187 |
+
"loss": 0.6945,
|
| 1188 |
+
"step": 165
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 3.32,
|
| 1192 |
+
"grad_norm": 1.1268410682678223,
|
| 1193 |
+
"learning_rate": 0.0002,
|
| 1194 |
+
"loss": 0.579,
|
| 1195 |
+
"step": 166
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 3.34,
|
| 1199 |
+
"grad_norm": 1.3038002252578735,
|
| 1200 |
+
"learning_rate": 0.0002,
|
| 1201 |
+
"loss": 0.5217,
|
| 1202 |
+
"step": 167
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 3.36,
|
| 1206 |
+
"grad_norm": 1.112185001373291,
|
| 1207 |
+
"learning_rate": 0.0002,
|
| 1208 |
+
"loss": 0.4766,
|
| 1209 |
+
"step": 168
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 3.38,
|
| 1213 |
+
"grad_norm": 1.3828542232513428,
|
| 1214 |
+
"learning_rate": 0.0002,
|
| 1215 |
+
"loss": 0.4781,
|
| 1216 |
+
"step": 169
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 3.4,
|
| 1220 |
+
"grad_norm": 1.1456600427627563,
|
| 1221 |
+
"learning_rate": 0.0002,
|
| 1222 |
+
"loss": 0.4056,
|
| 1223 |
+
"step": 170
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 3.42,
|
| 1227 |
+
"grad_norm": 1.2479093074798584,
|
| 1228 |
+
"learning_rate": 0.0002,
|
| 1229 |
+
"loss": 0.4447,
|
| 1230 |
+
"step": 171
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 3.44,
|
| 1234 |
+
"grad_norm": 1.4044010639190674,
|
| 1235 |
+
"learning_rate": 0.0002,
|
| 1236 |
+
"loss": 0.3814,
|
| 1237 |
+
"step": 172
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 3.46,
|
| 1241 |
+
"grad_norm": 1.565138339996338,
|
| 1242 |
+
"learning_rate": 0.0002,
|
| 1243 |
+
"loss": 0.3982,
|
| 1244 |
+
"step": 173
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 3.48,
|
| 1248 |
+
"grad_norm": 1.4442418813705444,
|
| 1249 |
+
"learning_rate": 0.0002,
|
| 1250 |
+
"loss": 0.4262,
|
| 1251 |
+
"step": 174
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 3.5,
|
| 1255 |
+
"grad_norm": 1.1203701496124268,
|
| 1256 |
+
"learning_rate": 0.0002,
|
| 1257 |
+
"loss": 0.8025,
|
| 1258 |
+
"step": 175
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 3.52,
|
| 1262 |
+
"grad_norm": 1.3620504140853882,
|
| 1263 |
+
"learning_rate": 0.0002,
|
| 1264 |
+
"loss": 0.9045,
|
| 1265 |
+
"step": 176
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 3.54,
|
| 1269 |
+
"grad_norm": 1.5145343542099,
|
| 1270 |
+
"learning_rate": 0.0002,
|
| 1271 |
+
"loss": 0.703,
|
| 1272 |
+
"step": 177
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 3.56,
|
| 1276 |
+
"grad_norm": 1.333682656288147,
|
| 1277 |
+
"learning_rate": 0.0002,
|
| 1278 |
+
"loss": 0.6285,
|
| 1279 |
+
"step": 178
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 3.58,
|
| 1283 |
+
"grad_norm": 1.4228661060333252,
|
| 1284 |
+
"learning_rate": 0.0002,
|
| 1285 |
+
"loss": 0.6004,
|
| 1286 |
+
"step": 179
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 3.6,
|
| 1290 |
+
"grad_norm": 1.2111386060714722,
|
| 1291 |
+
"learning_rate": 0.0002,
|
| 1292 |
+
"loss": 0.4754,
|
| 1293 |
+
"step": 180
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 3.62,
|
| 1297 |
+
"grad_norm": 1.410719394683838,
|
| 1298 |
+
"learning_rate": 0.0002,
|
| 1299 |
+
"loss": 0.5324,
|
| 1300 |
+
"step": 181
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 3.64,
|
| 1304 |
+
"grad_norm": 1.4157259464263916,
|
| 1305 |
+
"learning_rate": 0.0002,
|
| 1306 |
+
"loss": 0.4556,
|
| 1307 |
+
"step": 182
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 3.66,
|
| 1311 |
+
"grad_norm": 1.3982216119766235,
|
| 1312 |
+
"learning_rate": 0.0002,
|
| 1313 |
+
"loss": 0.4465,
|
| 1314 |
+
"step": 183
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 3.68,
|
| 1318 |
+
"grad_norm": 1.4364334344863892,
|
| 1319 |
+
"learning_rate": 0.0002,
|
| 1320 |
+
"loss": 0.4313,
|
| 1321 |
+
"step": 184
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 3.7,
|
| 1325 |
+
"grad_norm": 1.5408861637115479,
|
| 1326 |
+
"learning_rate": 0.0002,
|
| 1327 |
+
"loss": 0.4108,
|
| 1328 |
+
"step": 185
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 3.7199999999999998,
|
| 1332 |
+
"grad_norm": 1.5500551462173462,
|
| 1333 |
+
"learning_rate": 0.0002,
|
| 1334 |
+
"loss": 0.4664,
|
| 1335 |
+
"step": 186
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 3.74,
|
| 1339 |
+
"grad_norm": 1.1150060892105103,
|
| 1340 |
+
"learning_rate": 0.0002,
|
| 1341 |
+
"loss": 0.8951,
|
| 1342 |
+
"step": 187
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 3.76,
|
| 1346 |
+
"grad_norm": 1.0168464183807373,
|
| 1347 |
+
"learning_rate": 0.0002,
|
| 1348 |
+
"loss": 0.8619,
|
| 1349 |
+
"step": 188
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 3.7800000000000002,
|
| 1353 |
+
"grad_norm": 1.2093026638031006,
|
| 1354 |
+
"learning_rate": 0.0002,
|
| 1355 |
+
"loss": 0.5782,
|
| 1356 |
+
"step": 189
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 3.8,
|
| 1360 |
+
"grad_norm": 1.3905984163284302,
|
| 1361 |
+
"learning_rate": 0.0002,
|
| 1362 |
+
"loss": 0.6929,
|
| 1363 |
+
"step": 190
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 3.82,
|
| 1367 |
+
"grad_norm": 1.3665902614593506,
|
| 1368 |
+
"learning_rate": 0.0002,
|
| 1369 |
+
"loss": 0.5701,
|
| 1370 |
+
"step": 191
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 3.84,
|
| 1374 |
+
"grad_norm": 1.1478445529937744,
|
| 1375 |
+
"learning_rate": 0.0002,
|
| 1376 |
+
"loss": 0.4515,
|
| 1377 |
+
"step": 192
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 3.86,
|
| 1381 |
+
"grad_norm": 1.2758458852767944,
|
| 1382 |
+
"learning_rate": 0.0002,
|
| 1383 |
+
"loss": 0.5819,
|
| 1384 |
+
"step": 193
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 3.88,
|
| 1388 |
+
"grad_norm": 1.0731329917907715,
|
| 1389 |
+
"learning_rate": 0.0002,
|
| 1390 |
+
"loss": 0.4448,
|
| 1391 |
+
"step": 194
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 3.9,
|
| 1395 |
+
"grad_norm": 1.20659339427948,
|
| 1396 |
+
"learning_rate": 0.0002,
|
| 1397 |
+
"loss": 0.4295,
|
| 1398 |
+
"step": 195
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 3.92,
|
| 1402 |
+
"grad_norm": 1.3976835012435913,
|
| 1403 |
+
"learning_rate": 0.0002,
|
| 1404 |
+
"loss": 0.4566,
|
| 1405 |
+
"step": 196
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 3.94,
|
| 1409 |
+
"grad_norm": 1.617711067199707,
|
| 1410 |
+
"learning_rate": 0.0002,
|
| 1411 |
+
"loss": 0.4303,
|
| 1412 |
+
"step": 197
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 3.96,
|
| 1416 |
+
"grad_norm": 1.707471489906311,
|
| 1417 |
+
"learning_rate": 0.0002,
|
| 1418 |
+
"loss": 0.4539,
|
| 1419 |
+
"step": 198
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 3.98,
|
| 1423 |
+
"grad_norm": 1.2962028980255127,
|
| 1424 |
+
"learning_rate": 0.0002,
|
| 1425 |
+
"loss": 0.5971,
|
| 1426 |
+
"step": 199
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 4.0,
|
| 1430 |
+
"grad_norm": 1.8809109926223755,
|
| 1431 |
+
"learning_rate": 0.0002,
|
| 1432 |
+
"loss": 0.4973,
|
| 1433 |
+
"step": 200
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 4.0,
|
| 1437 |
+
"eval_loss": 1.5414550304412842,
|
| 1438 |
+
"eval_runtime": 565.7272,
|
| 1439 |
+
"eval_samples_per_second": 0.707,
|
| 1440 |
+
"eval_steps_per_second": 0.177,
|
| 1441 |
+
"step": 200
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 4.02,
|
| 1445 |
+
"grad_norm": 0.9540490508079529,
|
| 1446 |
+
"learning_rate": 0.0002,
|
| 1447 |
+
"loss": 0.56,
|
| 1448 |
+
"step": 201
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 4.04,
|
| 1452 |
+
"grad_norm": 1.0443426370620728,
|
| 1453 |
+
"learning_rate": 0.0002,
|
| 1454 |
+
"loss": 0.6562,
|
| 1455 |
+
"step": 202
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 4.06,
|
| 1459 |
+
"grad_norm": 1.020203948020935,
|
| 1460 |
+
"learning_rate": 0.0002,
|
| 1461 |
+
"loss": 0.4533,
|
| 1462 |
+
"step": 203
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 4.08,
|
| 1466 |
+
"grad_norm": 1.5309128761291504,
|
| 1467 |
+
"learning_rate": 0.0002,
|
| 1468 |
+
"loss": 0.3575,
|
| 1469 |
+
"step": 204
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 4.1,
|
| 1473 |
+
"grad_norm": 1.7135676145553589,
|
| 1474 |
+
"learning_rate": 0.0002,
|
| 1475 |
+
"loss": 0.3286,
|
| 1476 |
+
"step": 205
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 4.12,
|
| 1480 |
+
"grad_norm": 1.602728247642517,
|
| 1481 |
+
"learning_rate": 0.0002,
|
| 1482 |
+
"loss": 0.2556,
|
| 1483 |
+
"step": 206
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 4.14,
|
| 1487 |
+
"grad_norm": 1.8623350858688354,
|
| 1488 |
+
"learning_rate": 0.0002,
|
| 1489 |
+
"loss": 0.314,
|
| 1490 |
+
"step": 207
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 4.16,
|
| 1494 |
+
"grad_norm": 1.5630223751068115,
|
| 1495 |
+
"learning_rate": 0.0002,
|
| 1496 |
+
"loss": 0.2716,
|
| 1497 |
+
"step": 208
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 4.18,
|
| 1501 |
+
"grad_norm": 1.3671077489852905,
|
| 1502 |
+
"learning_rate": 0.0002,
|
| 1503 |
+
"loss": 0.2506,
|
| 1504 |
+
"step": 209
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 4.2,
|
| 1508 |
+
"grad_norm": 1.0884723663330078,
|
| 1509 |
+
"learning_rate": 0.0002,
|
| 1510 |
+
"loss": 0.2473,
|
| 1511 |
+
"step": 210
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 4.22,
|
| 1515 |
+
"grad_norm": 1.193832516670227,
|
| 1516 |
+
"learning_rate": 0.0002,
|
| 1517 |
+
"loss": 0.2836,
|
| 1518 |
+
"step": 211
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 4.24,
|
| 1522 |
+
"grad_norm": 1.0041422843933105,
|
| 1523 |
+
"learning_rate": 0.0002,
|
| 1524 |
+
"loss": 0.391,
|
| 1525 |
+
"step": 212
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 4.26,
|
| 1529 |
+
"grad_norm": 1.013597846031189,
|
| 1530 |
+
"learning_rate": 0.0002,
|
| 1531 |
+
"loss": 0.628,
|
| 1532 |
+
"step": 213
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 4.28,
|
| 1536 |
+
"grad_norm": 0.9650751948356628,
|
| 1537 |
+
"learning_rate": 0.0002,
|
| 1538 |
+
"loss": 0.5202,
|
| 1539 |
+
"step": 214
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 4.3,
|
| 1543 |
+
"grad_norm": 1.0781069993972778,
|
| 1544 |
+
"learning_rate": 0.0002,
|
| 1545 |
+
"loss": 0.4967,
|
| 1546 |
+
"step": 215
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 4.32,
|
| 1550 |
+
"grad_norm": 1.1297317743301392,
|
| 1551 |
+
"learning_rate": 0.0002,
|
| 1552 |
+
"loss": 0.4154,
|
| 1553 |
+
"step": 216
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 4.34,
|
| 1557 |
+
"grad_norm": 1.2913479804992676,
|
| 1558 |
+
"learning_rate": 0.0002,
|
| 1559 |
+
"loss": 0.3014,
|
| 1560 |
+
"step": 217
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 4.36,
|
| 1564 |
+
"grad_norm": 1.4399878978729248,
|
| 1565 |
+
"learning_rate": 0.0002,
|
| 1566 |
+
"loss": 0.3344,
|
| 1567 |
+
"step": 218
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 4.38,
|
| 1571 |
+
"grad_norm": 1.4960243701934814,
|
| 1572 |
+
"learning_rate": 0.0002,
|
| 1573 |
+
"loss": 0.2894,
|
| 1574 |
+
"step": 219
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 4.4,
|
| 1578 |
+
"grad_norm": 1.925826072692871,
|
| 1579 |
+
"learning_rate": 0.0002,
|
| 1580 |
+
"loss": 0.281,
|
| 1581 |
+
"step": 220
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 4.42,
|
| 1585 |
+
"grad_norm": 1.6930102109909058,
|
| 1586 |
+
"learning_rate": 0.0002,
|
| 1587 |
+
"loss": 0.2512,
|
| 1588 |
+
"step": 221
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 4.44,
|
| 1592 |
+
"grad_norm": 1.6776522397994995,
|
| 1593 |
+
"learning_rate": 0.0002,
|
| 1594 |
+
"loss": 0.2744,
|
| 1595 |
+
"step": 222
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 4.46,
|
| 1599 |
+
"grad_norm": 1.3323974609375,
|
| 1600 |
+
"learning_rate": 0.0002,
|
| 1601 |
+
"loss": 0.2951,
|
| 1602 |
+
"step": 223
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 4.48,
|
| 1606 |
+
"grad_norm": 1.2120009660720825,
|
| 1607 |
+
"learning_rate": 0.0002,
|
| 1608 |
+
"loss": 0.3707,
|
| 1609 |
+
"step": 224
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 4.5,
|
| 1613 |
+
"grad_norm": 1.2817238569259644,
|
| 1614 |
+
"learning_rate": 0.0002,
|
| 1615 |
+
"loss": 0.6035,
|
| 1616 |
+
"step": 225
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 4.52,
|
| 1620 |
+
"grad_norm": 1.1797271966934204,
|
| 1621 |
+
"learning_rate": 0.0002,
|
| 1622 |
+
"loss": 0.5878,
|
| 1623 |
+
"step": 226
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 4.54,
|
| 1627 |
+
"grad_norm": 0.9533390402793884,
|
| 1628 |
+
"learning_rate": 0.0002,
|
| 1629 |
+
"loss": 0.3458,
|
| 1630 |
+
"step": 227
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 4.5600000000000005,
|
| 1634 |
+
"grad_norm": 1.0915648937225342,
|
| 1635 |
+
"learning_rate": 0.0002,
|
| 1636 |
+
"loss": 0.3617,
|
| 1637 |
+
"step": 228
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 4.58,
|
| 1641 |
+
"grad_norm": 1.3463889360427856,
|
| 1642 |
+
"learning_rate": 0.0002,
|
| 1643 |
+
"loss": 0.3853,
|
| 1644 |
+
"step": 229
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 4.6,
|
| 1648 |
+
"grad_norm": 1.457556128501892,
|
| 1649 |
+
"learning_rate": 0.0002,
|
| 1650 |
+
"loss": 0.3343,
|
| 1651 |
+
"step": 230
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 4.62,
|
| 1655 |
+
"grad_norm": 1.681526780128479,
|
| 1656 |
+
"learning_rate": 0.0002,
|
| 1657 |
+
"loss": 0.3132,
|
| 1658 |
+
"step": 231
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 4.64,
|
| 1662 |
+
"grad_norm": 1.7101032733917236,
|
| 1663 |
+
"learning_rate": 0.0002,
|
| 1664 |
+
"loss": 0.2913,
|
| 1665 |
+
"step": 232
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 4.66,
|
| 1669 |
+
"grad_norm": 2.1125667095184326,
|
| 1670 |
+
"learning_rate": 0.0002,
|
| 1671 |
+
"loss": 0.304,
|
| 1672 |
+
"step": 233
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 4.68,
|
| 1676 |
+
"grad_norm": 1.5824134349822998,
|
| 1677 |
+
"learning_rate": 0.0002,
|
| 1678 |
+
"loss": 0.2878,
|
| 1679 |
+
"step": 234
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 4.7,
|
| 1683 |
+
"grad_norm": 1.5257948637008667,
|
| 1684 |
+
"learning_rate": 0.0002,
|
| 1685 |
+
"loss": 0.3049,
|
| 1686 |
+
"step": 235
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 4.72,
|
| 1690 |
+
"grad_norm": 1.1413626670837402,
|
| 1691 |
+
"learning_rate": 0.0002,
|
| 1692 |
+
"loss": 0.3875,
|
| 1693 |
+
"step": 236
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 4.74,
|
| 1697 |
+
"grad_norm": 1.4785950183868408,
|
| 1698 |
+
"learning_rate": 0.0002,
|
| 1699 |
+
"loss": 0.7915,
|
| 1700 |
+
"step": 237
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 4.76,
|
| 1704 |
+
"grad_norm": 1.0445829629898071,
|
| 1705 |
+
"learning_rate": 0.0002,
|
| 1706 |
+
"loss": 0.4765,
|
| 1707 |
+
"step": 238
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 4.78,
|
| 1711 |
+
"grad_norm": 1.0932363271713257,
|
| 1712 |
+
"learning_rate": 0.0002,
|
| 1713 |
+
"loss": 0.4081,
|
| 1714 |
+
"step": 239
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 4.8,
|
| 1718 |
+
"grad_norm": 1.313068151473999,
|
| 1719 |
+
"learning_rate": 0.0002,
|
| 1720 |
+
"loss": 0.4278,
|
| 1721 |
+
"step": 240
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 4.82,
|
| 1725 |
+
"grad_norm": 1.2771199941635132,
|
| 1726 |
+
"learning_rate": 0.0002,
|
| 1727 |
+
"loss": 0.3481,
|
| 1728 |
+
"step": 241
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 4.84,
|
| 1732 |
+
"grad_norm": 1.3306118249893188,
|
| 1733 |
+
"learning_rate": 0.0002,
|
| 1734 |
+
"loss": 0.3301,
|
| 1735 |
+
"step": 242
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 4.86,
|
| 1739 |
+
"grad_norm": 1.2204334735870361,
|
| 1740 |
+
"learning_rate": 0.0002,
|
| 1741 |
+
"loss": 0.2974,
|
| 1742 |
+
"step": 243
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 4.88,
|
| 1746 |
+
"grad_norm": 1.1585593223571777,
|
| 1747 |
+
"learning_rate": 0.0002,
|
| 1748 |
+
"loss": 0.2404,
|
| 1749 |
+
"step": 244
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 4.9,
|
| 1753 |
+
"grad_norm": 1.6888794898986816,
|
| 1754 |
+
"learning_rate": 0.0002,
|
| 1755 |
+
"loss": 0.2817,
|
| 1756 |
+
"step": 245
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 4.92,
|
| 1760 |
+
"grad_norm": 1.4956034421920776,
|
| 1761 |
+
"learning_rate": 0.0002,
|
| 1762 |
+
"loss": 0.2583,
|
| 1763 |
+
"step": 246
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 4.9399999999999995,
|
| 1767 |
+
"grad_norm": 1.6130638122558594,
|
| 1768 |
+
"learning_rate": 0.0002,
|
| 1769 |
+
"loss": 0.2888,
|
| 1770 |
+
"step": 247
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 4.96,
|
| 1774 |
+
"grad_norm": 2.280722141265869,
|
| 1775 |
+
"learning_rate": 0.0002,
|
| 1776 |
+
"loss": 0.3732,
|
| 1777 |
+
"step": 248
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 4.98,
|
| 1781 |
+
"grad_norm": 1.8177040815353394,
|
| 1782 |
+
"learning_rate": 0.0002,
|
| 1783 |
+
"loss": 0.6086,
|
| 1784 |
+
"step": 249
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 5.0,
|
| 1788 |
+
"grad_norm": 1.674232840538025,
|
| 1789 |
+
"learning_rate": 0.0002,
|
| 1790 |
+
"loss": 0.3232,
|
| 1791 |
+
"step": 250
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 5.0,
|
| 1795 |
+
"eval_loss": 1.6614739894866943,
|
| 1796 |
+
"eval_runtime": 565.5774,
|
| 1797 |
+
"eval_samples_per_second": 0.707,
|
| 1798 |
+
"eval_steps_per_second": 0.177,
|
| 1799 |
+
"step": 250
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 5.02,
|
| 1803 |
+
"grad_norm": 1.1646928787231445,
|
| 1804 |
+
"learning_rate": 0.0002,
|
| 1805 |
+
"loss": 0.4812,
|
| 1806 |
+
"step": 251
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 5.04,
|
| 1810 |
+
"grad_norm": 1.0134564638137817,
|
| 1811 |
+
"learning_rate": 0.0002,
|
| 1812 |
+
"loss": 0.3754,
|
| 1813 |
+
"step": 252
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 5.06,
|
| 1817 |
+
"grad_norm": 1.0580028295516968,
|
| 1818 |
+
"learning_rate": 0.0002,
|
| 1819 |
+
"loss": 0.3408,
|
| 1820 |
+
"step": 253
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 5.08,
|
| 1824 |
+
"grad_norm": 0.9608296751976013,
|
| 1825 |
+
"learning_rate": 0.0002,
|
| 1826 |
+
"loss": 0.1999,
|
| 1827 |
+
"step": 254
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 5.1,
|
| 1831 |
+
"grad_norm": 1.2543963193893433,
|
| 1832 |
+
"learning_rate": 0.0002,
|
| 1833 |
+
"loss": 0.208,
|
| 1834 |
+
"step": 255
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 5.12,
|
| 1838 |
+
"grad_norm": 1.2029474973678589,
|
| 1839 |
+
"learning_rate": 0.0002,
|
| 1840 |
+
"loss": 0.1838,
|
| 1841 |
+
"step": 256
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 5.14,
|
| 1845 |
+
"grad_norm": 1.3406902551651,
|
| 1846 |
+
"learning_rate": 0.0002,
|
| 1847 |
+
"loss": 0.1828,
|
| 1848 |
+
"step": 257
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 5.16,
|
| 1852 |
+
"grad_norm": 1.3970423936843872,
|
| 1853 |
+
"learning_rate": 0.0002,
|
| 1854 |
+
"loss": 0.1882,
|
| 1855 |
+
"step": 258
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 5.18,
|
| 1859 |
+
"grad_norm": 1.5742741823196411,
|
| 1860 |
+
"learning_rate": 0.0002,
|
| 1861 |
+
"loss": 0.2117,
|
| 1862 |
+
"step": 259
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 5.2,
|
| 1866 |
+
"grad_norm": 1.4690401554107666,
|
| 1867 |
+
"learning_rate": 0.0002,
|
| 1868 |
+
"loss": 0.2133,
|
| 1869 |
+
"step": 260
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 5.22,
|
| 1873 |
+
"grad_norm": 1.337886095046997,
|
| 1874 |
+
"learning_rate": 0.0002,
|
| 1875 |
+
"loss": 0.2375,
|
| 1876 |
+
"step": 261
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 5.24,
|
| 1880 |
+
"grad_norm": 1.3524601459503174,
|
| 1881 |
+
"learning_rate": 0.0002,
|
| 1882 |
+
"loss": 0.3295,
|
| 1883 |
+
"step": 262
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 5.26,
|
| 1887 |
+
"grad_norm": 1.533918023109436,
|
| 1888 |
+
"learning_rate": 0.0002,
|
| 1889 |
+
"loss": 0.4466,
|
| 1890 |
+
"step": 263
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 5.28,
|
| 1894 |
+
"grad_norm": 1.3888379335403442,
|
| 1895 |
+
"learning_rate": 0.0002,
|
| 1896 |
+
"loss": 0.3435,
|
| 1897 |
+
"step": 264
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 5.3,
|
| 1901 |
+
"grad_norm": 1.0731428861618042,
|
| 1902 |
+
"learning_rate": 0.0002,
|
| 1903 |
+
"loss": 0.2129,
|
| 1904 |
+
"step": 265
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 5.32,
|
| 1908 |
+
"grad_norm": 1.0117309093475342,
|
| 1909 |
+
"learning_rate": 0.0002,
|
| 1910 |
+
"loss": 0.2522,
|
| 1911 |
+
"step": 266
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 5.34,
|
| 1915 |
+
"grad_norm": 1.1278893947601318,
|
| 1916 |
+
"learning_rate": 0.0002,
|
| 1917 |
+
"loss": 0.2326,
|
| 1918 |
+
"step": 267
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 5.36,
|
| 1922 |
+
"grad_norm": 1.1024526357650757,
|
| 1923 |
+
"learning_rate": 0.0002,
|
| 1924 |
+
"loss": 0.1924,
|
| 1925 |
+
"step": 268
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 5.38,
|
| 1929 |
+
"grad_norm": 1.3139152526855469,
|
| 1930 |
+
"learning_rate": 0.0002,
|
| 1931 |
+
"loss": 0.2193,
|
| 1932 |
+
"step": 269
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 5.4,
|
| 1936 |
+
"grad_norm": 1.2390962839126587,
|
| 1937 |
+
"learning_rate": 0.0002,
|
| 1938 |
+
"loss": 0.2005,
|
| 1939 |
+
"step": 270
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 5.42,
|
| 1943 |
+
"grad_norm": 1.5270419120788574,
|
| 1944 |
+
"learning_rate": 0.0002,
|
| 1945 |
+
"loss": 0.2245,
|
| 1946 |
+
"step": 271
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 5.44,
|
| 1950 |
+
"grad_norm": 1.48665452003479,
|
| 1951 |
+
"learning_rate": 0.0002,
|
| 1952 |
+
"loss": 0.2317,
|
| 1953 |
+
"step": 272
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 5.46,
|
| 1957 |
+
"grad_norm": 1.4966135025024414,
|
| 1958 |
+
"learning_rate": 0.0002,
|
| 1959 |
+
"loss": 0.2785,
|
| 1960 |
+
"step": 273
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 5.48,
|
| 1964 |
+
"grad_norm": 1.2391105890274048,
|
| 1965 |
+
"learning_rate": 0.0002,
|
| 1966 |
+
"loss": 0.3812,
|
| 1967 |
+
"step": 274
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 5.5,
|
| 1971 |
+
"grad_norm": 1.5869184732437134,
|
| 1972 |
+
"learning_rate": 0.0002,
|
| 1973 |
+
"loss": 0.4248,
|
| 1974 |
+
"step": 275
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 5.52,
|
| 1978 |
+
"grad_norm": 1.3854241371154785,
|
| 1979 |
+
"learning_rate": 0.0002,
|
| 1980 |
+
"loss": 0.3517,
|
| 1981 |
+
"step": 276
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 5.54,
|
| 1985 |
+
"grad_norm": 1.2200820446014404,
|
| 1986 |
+
"learning_rate": 0.0002,
|
| 1987 |
+
"loss": 0.2956,
|
| 1988 |
+
"step": 277
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 5.5600000000000005,
|
| 1992 |
+
"grad_norm": 1.0377618074417114,
|
| 1993 |
+
"learning_rate": 0.0002,
|
| 1994 |
+
"loss": 0.2488,
|
| 1995 |
+
"step": 278
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 5.58,
|
| 1999 |
+
"grad_norm": 1.1551989316940308,
|
| 2000 |
+
"learning_rate": 0.0002,
|
| 2001 |
+
"loss": 0.2139,
|
| 2002 |
+
"step": 279
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 5.6,
|
| 2006 |
+
"grad_norm": 1.1790027618408203,
|
| 2007 |
+
"learning_rate": 0.0002,
|
| 2008 |
+
"loss": 0.2154,
|
| 2009 |
+
"step": 280
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 5.62,
|
| 2013 |
+
"grad_norm": 1.4331631660461426,
|
| 2014 |
+
"learning_rate": 0.0002,
|
| 2015 |
+
"loss": 0.2332,
|
| 2016 |
+
"step": 281
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 5.64,
|
| 2020 |
+
"grad_norm": 1.3983947038650513,
|
| 2021 |
+
"learning_rate": 0.0002,
|
| 2022 |
+
"loss": 0.235,
|
| 2023 |
+
"step": 282
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 5.66,
|
| 2027 |
+
"grad_norm": 1.4608796834945679,
|
| 2028 |
+
"learning_rate": 0.0002,
|
| 2029 |
+
"loss": 0.2068,
|
| 2030 |
+
"step": 283
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 5.68,
|
| 2034 |
+
"grad_norm": 1.3449772596359253,
|
| 2035 |
+
"learning_rate": 0.0002,
|
| 2036 |
+
"loss": 0.2145,
|
| 2037 |
+
"step": 284
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 5.7,
|
| 2041 |
+
"grad_norm": 1.5349823236465454,
|
| 2042 |
+
"learning_rate": 0.0002,
|
| 2043 |
+
"loss": 0.2537,
|
| 2044 |
+
"step": 285
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 5.72,
|
| 2048 |
+
"grad_norm": 1.5507280826568604,
|
| 2049 |
+
"learning_rate": 0.0002,
|
| 2050 |
+
"loss": 0.3617,
|
| 2051 |
+
"step": 286
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 5.74,
|
| 2055 |
+
"grad_norm": 1.4181163311004639,
|
| 2056 |
+
"learning_rate": 0.0002,
|
| 2057 |
+
"loss": 0.428,
|
| 2058 |
+
"step": 287
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 5.76,
|
| 2062 |
+
"grad_norm": 1.1474294662475586,
|
| 2063 |
+
"learning_rate": 0.0002,
|
| 2064 |
+
"loss": 0.3593,
|
| 2065 |
+
"step": 288
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 5.78,
|
| 2069 |
+
"grad_norm": 1.0448319911956787,
|
| 2070 |
+
"learning_rate": 0.0002,
|
| 2071 |
+
"loss": 0.2488,
|
| 2072 |
+
"step": 289
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 5.8,
|
| 2076 |
+
"grad_norm": 1.2155684232711792,
|
| 2077 |
+
"learning_rate": 0.0002,
|
| 2078 |
+
"loss": 0.2429,
|
| 2079 |
+
"step": 290
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 5.82,
|
| 2083 |
+
"grad_norm": 1.037227749824524,
|
| 2084 |
+
"learning_rate": 0.0002,
|
| 2085 |
+
"loss": 0.1951,
|
| 2086 |
+
"step": 291
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 5.84,
|
| 2090 |
+
"grad_norm": 1.0879547595977783,
|
| 2091 |
+
"learning_rate": 0.0002,
|
| 2092 |
+
"loss": 0.2003,
|
| 2093 |
+
"step": 292
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 5.86,
|
| 2097 |
+
"grad_norm": 1.2885050773620605,
|
| 2098 |
+
"learning_rate": 0.0002,
|
| 2099 |
+
"loss": 0.2133,
|
| 2100 |
+
"step": 293
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 5.88,
|
| 2104 |
+
"grad_norm": 1.335684061050415,
|
| 2105 |
+
"learning_rate": 0.0002,
|
| 2106 |
+
"loss": 0.2037,
|
| 2107 |
+
"step": 294
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 5.9,
|
| 2111 |
+
"grad_norm": 1.3747302293777466,
|
| 2112 |
+
"learning_rate": 0.0002,
|
| 2113 |
+
"loss": 0.2276,
|
| 2114 |
+
"step": 295
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 5.92,
|
| 2118 |
+
"grad_norm": 1.3095324039459229,
|
| 2119 |
+
"learning_rate": 0.0002,
|
| 2120 |
+
"loss": 0.2255,
|
| 2121 |
+
"step": 296
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 5.9399999999999995,
|
| 2125 |
+
"grad_norm": 1.3633639812469482,
|
| 2126 |
+
"learning_rate": 0.0002,
|
| 2127 |
+
"loss": 0.2837,
|
| 2128 |
+
"step": 297
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 5.96,
|
| 2132 |
+
"grad_norm": 1.3497192859649658,
|
| 2133 |
+
"learning_rate": 0.0002,
|
| 2134 |
+
"loss": 0.3738,
|
| 2135 |
+
"step": 298
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 5.98,
|
| 2139 |
+
"grad_norm": 1.4940162897109985,
|
| 2140 |
+
"learning_rate": 0.0002,
|
| 2141 |
+
"loss": 0.3705,
|
| 2142 |
+
"step": 299
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 6.0,
|
| 2146 |
+
"grad_norm": 1.4202619791030884,
|
| 2147 |
+
"learning_rate": 0.0002,
|
| 2148 |
+
"loss": 0.2504,
|
| 2149 |
+
"step": 300
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 6.0,
|
| 2153 |
+
"eval_loss": 1.7531368732452393,
|
| 2154 |
+
"eval_runtime": 566.1878,
|
| 2155 |
+
"eval_samples_per_second": 0.706,
|
| 2156 |
+
"eval_steps_per_second": 0.177,
|
| 2157 |
+
"step": 300
|
| 2158 |
+
}
|
| 2159 |
+
],
|
| 2160 |
+
"logging_steps": 1,
|
| 2161 |
+
"max_steps": 300,
|
| 2162 |
+
"num_input_tokens_seen": 0,
|
| 2163 |
+
"num_train_epochs": 6,
|
| 2164 |
+
"save_steps": 500,
|
| 2165 |
+
"stateful_callbacks": {
|
| 2166 |
+
"TrainerControl": {
|
| 2167 |
+
"args": {
|
| 2168 |
+
"should_epoch_stop": false,
|
| 2169 |
+
"should_evaluate": false,
|
| 2170 |
+
"should_log": false,
|
| 2171 |
+
"should_save": true,
|
| 2172 |
+
"should_training_stop": true
|
| 2173 |
+
},
|
| 2174 |
+
"attributes": {}
|
| 2175 |
+
}
|
| 2176 |
+
},
|
| 2177 |
+
"total_flos": 1.1321894941576397e+17,
|
| 2178 |
+
"train_batch_size": 8,
|
| 2179 |
+
"trial_name": null,
|
| 2180 |
+
"trial_params": null
|
| 2181 |
+
}
|