diff --git a/.github/workflows/test_onnx.yml b/.github/workflows/test_onnx.yml index 5a21f12d01..5524ca67ea 100644 --- a/.github/workflows/test_onnx.yml +++ b/.github/workflows/test_onnx.yml @@ -27,7 +27,7 @@ jobs: python-version: ${{ matrix.python-version }} - name: Install dependencies run: | - pip install .[tests,onnxruntime,exporters-tf] + pip install .[tests,onnxruntime,exporters] - name: Test with unittest working-directory: tests run: | diff --git a/optimum/onnxruntime/modeling_decoder.py b/optimum/onnxruntime/modeling_decoder.py index fd7e741d7c..6a0dcbba2f 100644 --- a/optimum/onnxruntime/modeling_decoder.py +++ b/optimum/onnxruntime/modeling_decoder.py @@ -121,6 +121,7 @@ class ORTModelForCausalLM(ORTModel, GenerationMixin): auto_model_class = AutoModelForCausalLM main_input_name = "input_ids" + _supports_cache_class = False def __init__( self, diff --git a/optimum/onnxruntime/modeling_seq2seq.py b/optimum/onnxruntime/modeling_seq2seq.py index 89a0ae44d5..3b1af05d0f 100644 --- a/optimum/onnxruntime/modeling_seq2seq.py +++ b/optimum/onnxruntime/modeling_seq2seq.py @@ -570,6 +570,7 @@ class ORTModelForConditionalGeneration(ORTModel, ABC): # Used in from_transformers to export model to onnxORTEncoder base_model_prefix = "onnx_model" + _supports_cache_class = False def __init__( self,