Update README.md

README.md

@@ -18,7 +18,7 @@ pipeline_tag: text-generation
 
 The model is intended to be used on segments of **250** characters. Anything else will compromise performance. 
 
-In initial testing, using **0.25** as a probability threshold typically proved to produce the best F1 score.
+In initial testing, using **0.25** as a probability threshold proved to typically produce the best F1 score across all degrees of corruption.