Fix typo in documentation. (#21632)

src/transformers/generation/configuration_utils.py

@@ -99,7 +99,7 @@ class GenerationConfig(PushToHubMixin):
         > Parameters for manipulation of the model output logits
 
         temperature (`float`, *optional*, defaults to 1.0):
-            The value used to module the next token probabilities.
+            The value used to modulate the next token probabilities.
         top_k (`int`, *optional*, defaults to 50):
             The number of highest probability vocabulary tokens to keep for top-k-filtering.
         top_p (`float`, *optional*, defaults to 1.0):