Update logs/README.md
Browse files- logs/README.md +8 -8
logs/README.md
CHANGED
|
@@ -15,7 +15,7 @@
|
|
| 15 |
- precision: bfloat16
|
| 16 |
- embeddings: RoPE
|
| 17 |
- attention: flash
|
| 18 |
-
- num_epochs:
|
| 19 |
|
| 20 |
***
|
| 21 |
|
|
@@ -23,14 +23,14 @@
|
|
| 23 |
|
| 24 |
### Cross-entropy losses/accs (averaged last 100 values)
|
| 25 |
|
| 26 |
-
- train_loss: 0.
|
| 27 |
-
- train_acc: 0.
|
| 28 |
-
-
|
| 29 |
-
-
|
| 30 |
|
| 31 |
-
- epochs:
|
| 32 |
-
- num_steps:
|
| 33 |
-
- training_time:
|
| 34 |
|
| 35 |
***
|
| 36 |
|
|
|
|
| 15 |
- precision: bfloat16
|
| 16 |
- embeddings: RoPE
|
| 17 |
- attention: flash
|
| 18 |
+
- num_epochs: 4
|
| 19 |
|
| 20 |
***
|
| 21 |
|
|
|
|
| 23 |
|
| 24 |
### Cross-entropy losses/accs (averaged last 100 values)
|
| 25 |
|
| 26 |
+
- train_loss: 0.6657808522880078
|
| 27 |
+
- train_acc: 0.8036309605836869
|
| 28 |
+
- val_loss: 0.685554896891117
|
| 29 |
+
- val_acc: 0.7972136563062668
|
| 30 |
|
| 31 |
+
- epochs: 4
|
| 32 |
+
- num_steps: 128497
|
| 33 |
+
- training_time: 256 hours / 10.66 days
|
| 34 |
|
| 35 |
***
|
| 36 |
|