[examples] improve block_size warning message (#21463)

examples/pytorch/language-modeling/run_clm.py

@@ -459,8 +459,9 @@ def main():
         block_size = tokenizer.model_max_length
         if block_size > 1024:
             logger.warning(
-                f"The tokenizer picked seems to have a very large `model_max_length` ({tokenizer.model_max_length}). "
+                "The chosen tokenizer supports a `model_max_length` that is longer than the default `block_size` value"
-                "Picking 1024 instead. You can change that default value by passing --block_size xxx."
+                " of 1024. If you would like to use a longer `block_size` up to `tokenizer.model_max_length` you can"
+                " override this default with `--block_size xxx`."
             )
             block_size = 1024
     else:

examples/pytorch/language-modeling/run_clm_no_trainer.py

@@ -407,8 +407,9 @@ def main():
         block_size = tokenizer.model_max_length
         if block_size > 1024:
             logger.warning(
-                f"The tokenizer picked seems to have a very large `model_max_length` ({tokenizer.model_max_length}). "
+                "The chosen tokenizer supports a `model_max_length` that is longer than the default `block_size` value"
-                "Picking 1024 instead. You can change that default value by passing --block_size xxx."
+                " of 1024. If you would like to use a longer `block_size` up to `tokenizer.model_max_length` you can"
+                " override this default with `--block_size xxx`."
             )
         block_size = 1024
     else:

examples/pytorch/language-modeling/run_mlm.py

@@ -414,8 +414,9 @@ def main():
         max_seq_length = tokenizer.model_max_length
         if max_seq_length > 1024:
             logger.warning(
-                f"The tokenizer picked seems to have a very large `model_max_length` ({tokenizer.model_max_length}). "
+                "The chosen tokenizer supports a `model_max_length` that is longer than the default `block_size` value"
-                "Picking 1024 instead. You can change that default value by passing --max_seq_length xxx."
+                " of 1024. If you would like to use a longer `block_size` up to `tokenizer.model_max_length` you can"
+                " override this default with `--block_size xxx`."
             )
             max_seq_length = 1024
     else:

examples/pytorch/language-modeling/run_mlm_no_trainer.py

@@ -399,8 +399,9 @@ def main():
         max_seq_length = tokenizer.model_max_length
         if max_seq_length > 1024:
             logger.warning(
-                f"The tokenizer picked seems to have a very large `model_max_length` ({tokenizer.model_max_length}). "
+                "The chosen tokenizer supports a `model_max_length` that is longer than the default `block_size` value"
-                "Picking 1024 instead. You can change that default value by passing --max_seq_length xxx."
+                " of 1024. If you would like to use a longer `block_size` up to `tokenizer.model_max_length` you can"
+                " override this default with `--block_size xxx`."
             )
             max_seq_length = 1024
     else:

examples/pytorch/multiple-choice/run_swag.py

@@ -336,8 +336,9 @@ def main():
         max_seq_length = tokenizer.model_max_length
         if max_seq_length > 1024:
             logger.warning(
-                f"The tokenizer picked seems to have a very large `model_max_length` ({tokenizer.model_max_length}). "
+                "The chosen tokenizer supports a `model_max_length` that is longer than the default `block_size` value"
-                "Picking 1024 instead. You can change that default value by passing --max_seq_length xxx."
+                " of 1024. If you would like to use a longer `block_size` up to `tokenizer.model_max_length` you can"
+                " override this default with `--block_size xxx`."
             )
             max_seq_length = 1024
     else: