Update README.md

README.md

@@ -19,6 +19,7 @@ Loading the raw model in code:
 ```python
 from peft import PeftModel, PeftConfig
 from transformers import AutoModelForTokenClassification
+import torch
 
 id2label = {
     0: "O",
@@ -76,8 +77,10 @@ Results for commas on the wikitext validation set:
 |----------|-----------|--------|------|---------|
 | baseline* | 0.79      | 0.72   | 0.75 | 10079   |
 | ours      | 0.84      | 0.84   | 0.84 | 10079   |
+
 *baseline is the [oliverguhr/fullstop-punctuation-multilang-large](https://huggingface.co/oliverguhr/fullstop-punctuation-multilang-large)
-model evaluated on commas.
+model evaluated on commas, out of domain on wikitext. In-domain, authors report F1 of 0.819 for English political speeches, however, 
+it seems that wikipedia text could be more challenging for comma restoration.
 
 ## Training procedure