Fix condition for emitting warning when generation exceeds max model length (#40775)

correct warning when generation exceeds max model length

Signed-off-by: Yannick Schnider <yannick.schnider1@ibm.com>

Author: yannicks1

src/transformers/generation/stopping_criteria.py

@@ -76,9 +76,9 @@ def __init__(self, max_length: int, max_position_embeddings: Optional[int] = Non
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> torch.BoolTensor:
         cur_len = input_ids.shape[1]
         is_done = cur_len >= self.max_length
-        if self.max_position_embeddings is not None and not is_done and cur_len >= self.max_position_embeddings:
+        if self.max_position_embeddings is not None and not is_done and cur_len > self.max_position_embeddings:
             logger.warning_once(
-                "This is a friendly reminder - the current text generation call will exceed the model's predefined "
+                "This is a friendly reminder - the current text generation call has exceeded the model's predefined "
                 f"maximum length ({self.max_position_embeddings}). Depending on the model, you may observe "
                 "exceptions, performance degradation, or nothing at all."
             )