1-800-BAD-CODE
commited on
Commit
•
8d75e5b
1
Parent(s):
932cc97
Update README.md
Browse files
README.md
CHANGED
@@ -285,11 +285,8 @@ This model predicts the following set of punctuation tokens before each subword:
|
|
285 |
|
286 |
|
287 |
# Training Details
|
288 |
-
This model was trained in the NeMo framework.
|
289 |
|
290 |
-
This model was trained on an A100 for approximately 9 hours.
|
291 |
-
|
292 |
-
## Training Data
|
293 |
This model was trained with News Crawl data from WMT.
|
294 |
1M lines of text for each language was used, except for a few low-resource languages which may have used less.
|
295 |
Languages were chosen based on whether the News Crawl corpus contained enough reliable-quality data as judged by the author.
|
@@ -308,6 +305,8 @@ by selecting more of these sentences from additional training data that was not
|
|
308 |
|
309 |
The model may also over-predict commas.
|
310 |
|
|
|
|
|
311 |
|
312 |
# Evaluation
|
313 |
In these metrics, keep in mind that
|
|
|
285 |
|
286 |
|
287 |
# Training Details
|
288 |
+
This model was trained in the NeMo framework on an A100 for approximately 9 hours.
|
289 |
|
|
|
|
|
|
|
290 |
This model was trained with News Crawl data from WMT.
|
291 |
1M lines of text for each language was used, except for a few low-resource languages which may have used less.
|
292 |
Languages were chosen based on whether the News Crawl corpus contained enough reliable-quality data as judged by the author.
|
|
|
305 |
|
306 |
The model may also over-predict commas.
|
307 |
|
308 |
+
If you find any general limitations not mentioned here, let me know so all limitations can be addressed in the
|
309 |
+
next fine-tuning.
|
310 |
|
311 |
# Evaluation
|
312 |
In these metrics, keep in mind that
|