Add can_generate method to fix compatibility with transformers v4.39.0

optimum/onnxruntime/modeling_decoder.py

@@ -676,10 +676,6 @@ def _reorder_cache(past: Tuple[Tuple[torch.Tensor]], beam_idx: torch.Tensor) ->
             for layer_past in past
         )
 
-    def can_generate(self):
-        """Returns True to validate the check that the model using `GenerationMixin.generate()` can indeed generate."""
-        return True
-
 
 class ORTGPTBigCodeForCausalLM(ORTModelForCausalLM):
     # Adapted from transformers.models.gpt_bigcode.modeling_gpt_bigcode.GPTBigCodeForCausalLM.prepare_inputs_for_generation

optimum/onnxruntime/modeling_ort.py

@@ -52,6 +52,7 @@
     SequenceClassifierOutput,
     TokenClassifierOutput,
     XVectorOutput,
+    GenerationMixin,
 )
 
 import onnxruntime as ort
@@ -894,6 +895,12 @@ def _cached_file(
 
         return model_cache_path, preprocessors
 
+    def can_generate(self) -> bool:
+        """
+        Returns whether this model can generate sequences with `.generate()`.
+        """
+        return isinstance(self, GenerationMixin)
+
 
 FEATURE_EXTRACTION_EXAMPLE = r"""
     Example of feature extraction:

optimum/onnxruntime/modeling_seq2seq.py

@@ -1111,12 +1111,6 @@ def to(self, device: Union[torch.device, str, int]):
 
         return self
 
-    def can_generate(self):
-        logger.warning(
-            "ORTModelForConditionalGeneration is an abstract class and is not meant to be used for generation. Please use ORTModelForSeq2SeqLM or ORTModelForSpeechSeq2Seq."
-        )
-        return False
-
 
 @add_end_docstrings(ONNX_MODEL_END_DOCSTRING)
 class ORTModelForSeq2SeqLM(ORTModelForConditionalGeneration, GenerationMixin):
@@ -1262,10 +1256,6 @@ def _reorder_cache(past, beam_idx) -> Tuple[Tuple[torch.FloatTensor]]:
             )
         return reordered_past
 
-    def can_generate(self):
-        """Returns True to validate the check that the model using `GenerationMixin.generate()` can indeed generate."""
-        return True
-
 
 @add_end_docstrings(ONNX_MODEL_END_DOCSTRING)
 class ORTModelForSpeechSeq2Seq(ORTModelForConditionalGeneration, GenerationMixin):
@@ -1397,10 +1387,6 @@ def _reorder_cache(past, beam_idx) -> Tuple[Tuple[torch.FloatTensor]]:
             )
         return reordered_past
 
-    def can_generate(self):
-        """Returns True to validate the check that the model using `GenerationMixin.generate()` can indeed generate."""
-        return True
-
     @classmethod
     def _from_pretrained(
         cls,
@@ -1986,10 +1972,6 @@ def _reorder_cache(past, beam_idx) -> Tuple[Tuple[torch.FloatTensor]]:
             )
         return reordered_past
 
-    def can_generate(self):
-        """Returns True to validate the check that the model using `GenerationMixin.generate()` can indeed generate."""
-        return True
-
 
 @add_end_docstrings(ONNX_MODEL_END_DOCSTRING)
 class ORTModelForPix2Struct(ORTModelForConditionalGeneration, GenerationMixin):
@@ -2105,7 +2087,3 @@ def get_encoder(self) -> ORTEncoder:
     @staticmethod
     def _reorder_cache(past, beam_idx) -> Tuple[Tuple[torch.FloatTensor]]:
         ORTModelForSeq2SeqLM._reorder_cache(past, beam_idx)
-
-    def can_generate(self):
-        """Returns True to validate the check that the model using `GenerationMixin.generate()` can indeed generate."""
-        return True