Update notes
Browse files
README.md
CHANGED
|
@@ -39,6 +39,7 @@ This table tracks the performance of our model on various tasks over time. The m
|
|
| 39 |
- All accuracy values are presented as percentages
|
| 40 |
- Empty cells indicate that the task was not evaluated on that date or for that metric
|
| 41 |
- Standard errors are also converted to percentages for consistency
|
|
|
|
| 42 |
|
| 43 |
# Tokenizer
|
| 44 |
Our tokenizer was trained from scratch on 500,000 samples from the Openwebtext dataset.
|
|
|
|
| 39 |
- All accuracy values are presented as percentages
|
| 40 |
- Empty cells indicate that the task was not evaluated on that date or for that metric
|
| 41 |
- Standard errors are also converted to percentages for consistency
|
| 42 |
+
- Averages do not include preplexity scores.
|
| 43 |
|
| 44 |
# Tokenizer
|
| 45 |
Our tokenizer was trained from scratch on 500,000 samples from the Openwebtext dataset.
|