Commit
•
023197d
1
Parent(s):
a3433d1
Update README.md
Browse files
README.md
CHANGED
|
@@ -67,7 +67,7 @@ According to our evaluations, Zyda is the most performant per-token open dataset
|
|
| 67 |
<!--  -->
|
| 68 |
|
| 69 |
<center>
|
| 70 |
-
<img src="https://cdn-uploads.huggingface.co/production/uploads/65c05e75c084467acab2f84a/fXaQAOBDJpoaAr1clfTel.png" width="
|
| 71 |
</center>
|
| 72 |
|
| 73 |
These results are an aggregate scores of classic language modelling evaluations (piqa, winogrande, openbookqa, arc-easy, arc-challenge) across time for a 1.4B model trained on 50B tokens of each dataset.
|
|
|
|
| 67 |
<!--  -->
|
| 68 |
|
| 69 |
<center>
|
| 70 |
+
<img src="https://cdn-uploads.huggingface.co/production/uploads/65c05e75c084467acab2f84a/fXaQAOBDJpoaAr1clfTel.png" width="900" alt="Zyda performance across steps.">
|
| 71 |
</center>
|
| 72 |
|
| 73 |
These results are an aggregate scores of classic language modelling evaluations (piqa, winogrande, openbookqa, arc-easy, arc-challenge) across time for a 1.4B model trained on 50B tokens of each dataset.
|