Bump min torch version (#1641)

* Update min torch version * add sentencepiece * remove install from source
huggingface · Jan 19, 2024 · d6fe4e3 · d6fe4e3
1 parent 6b88276
commit d6fe4e3
Show file tree

Hide file tree

Showing 2 changed files with 50 additions and 56 deletions.
diff --git a/optimum/exporters/onnx/convert.py b/optimum/exporters/onnx/convert.py
@@ -46,7 +46,6 @@
     import torch
     import torch.nn as nn
     from transformers.modeling_utils import PreTrainedModel
-    from transformers.pytorch_utils import is_torch_less_than_1_11
 
 if is_diffusers_available():
     from diffusers import ModelMixin
@@ -556,62 +555,57 @@ def remap(value):
 
         dummy_inputs = config.rename_ambiguous_inputs(dummy_inputs)
 
-        # PyTorch deprecated the `enable_onnx_checker` and `use_external_data_format` arguments in v1.11,
-        # so we check the torch version for backwards compatibility
-        if is_torch_less_than_1_11:
-            raise RuntimeError("The ONNX export using the PyTorch framework is only supported for v1.11+")
-        else:
-            with config.patch_model_for_export(model, model_kwargs=model_kwargs):
-                check_dummy_inputs_are_allowed(model, dummy_inputs)
-
-                inputs = config.ordered_inputs(model)
-                input_names = list(inputs.keys())
-                output_names = list(config.outputs.keys())
-
-                # Export can work with named args but the dict containing named args has to be the last element of the args
-                # tuple.
-                onnx_export(
-                    model,
-                    (dummy_inputs,),
-                    f=output.as_posix(),
-                    input_names=input_names,
-                    output_names=output_names,
-                    dynamic_axes=dict(chain(inputs.items(), config.outputs.items())),
-                    do_constant_folding=True,
-                    opset_version=opset,
-                )
+        with config.patch_model_for_export(model, model_kwargs=model_kwargs):
+            check_dummy_inputs_are_allowed(model, dummy_inputs)
+
+            inputs = config.ordered_inputs(model)
+            input_names = list(inputs.keys())
+            output_names = list(config.outputs.keys())
+
+            # Export can work with named args but the dict containing named args has to be the last element of the args
+            # tuple.
+            onnx_export(
+                model,
+                (dummy_inputs,),
+                f=output.as_posix(),
+                input_names=input_names,
+                output_names=output_names,
+                dynamic_axes=dict(chain(inputs.items(), config.outputs.items())),
+                do_constant_folding=True,
+                opset_version=opset,
+            )
 
-            # check if external data was exported
-            # TODO: this is quite inefficient as we load in memory if models are <2GB without external data
-            onnx_model = onnx.load(str(output), load_external_data=False)
-            model_uses_external_data = check_model_uses_external_data(onnx_model)
-
-            if model_uses_external_data or FORCE_ONNX_EXTERNAL_DATA:
-                tensors_paths = _get_onnx_external_data_tensors(onnx_model)
-                logger.info("Saving external data to one file...")
-
-                # try free model memory
-                del model
-                del onnx_model
-                gc.collect()
-                if device.type == "cuda" and torch.cuda.is_available():
-                    torch.cuda.empty_cache()
-
-                onnx_model = onnx.load(
-                    str(output), load_external_data=True
-                )  # this will probably be too memory heavy for large models
-                onnx.save(
-                    onnx_model,
-                    str(output),
-                    save_as_external_data=True,
-                    all_tensors_to_one_file=True,
-                    location=output.name + "_data",
-                    size_threshold=1024 if not FORCE_ONNX_EXTERNAL_DATA else 0,
-                )
+        # check if external data was exported
+        # TODO: this is quite inefficient as we load in memory if models are <2GB without external data
+        onnx_model = onnx.load(str(output), load_external_data=False)
+        model_uses_external_data = check_model_uses_external_data(onnx_model)
+
+        if model_uses_external_data or FORCE_ONNX_EXTERNAL_DATA:
+            tensors_paths = _get_onnx_external_data_tensors(onnx_model)
+            logger.info("Saving external data to one file...")
+
+            # try free model memory
+            del model
+            del onnx_model
+            gc.collect()
+            if device.type == "cuda" and torch.cuda.is_available():
+                torch.cuda.empty_cache()
+
+            onnx_model = onnx.load(
+                str(output), load_external_data=True
+            )  # this will probably be too memory heavy for large models
+            onnx.save(
+                onnx_model,
+                str(output),
+                save_as_external_data=True,
+                all_tensors_to_one_file=True,
+                location=output.name + "_data",
+                size_threshold=1024 if not FORCE_ONNX_EXTERNAL_DATA else 0,
+            )
 
-                # delete previous external data
-                for tensor in tensors_paths:
-                    os.remove(output.parent / tensor)
+            # delete previous external data
+            for tensor in tensors_paths:
+                os.remove(output.parent / tensor)
 
     return input_names, output_names
 

diff --git a/setup.py b/setup.py
@@ -16,7 +16,7 @@
     "coloredlogs",
     "sympy",
     "transformers[sentencepiece]>=4.26.0",
-    "torch>=1.9",
+    "torch>=1.11",
     "packaging",
     "numpy",
     "huggingface_hub>=0.8.0",