Remove legacy export (#57)

- Remove legacy export of decoders
- Remove deprecated arguments: use_auth_token, fp16 vs dtype, for_ort,
some optimization arguments
- Add some missing optimization arguments (added clip optimization to
catch them) following [FusionOptions
args](https://github.com/microsoft/onnxruntime/blob/88f2652dee465faa485cf68cb39c4489f28945bb/onnxruntime/python/tools/transformers/fusion_options.py#L136)
- Fixed CI by adding a Hub token with access to optimum-internal-testing
and optimum orgs
- removed direct imports from namespaces packages (optimum.exporters,
optimum.commands, etc) for compatibility with native namespace packages
(see huggingface/optimum#2361).

Didn't remove support for loading and inference with legacy models
entirely as it would break many tests that still use them.

Author: IlyasMoutawwakil

.github/workflows/test_exporters_common.yml

@@ -14,7 +14,7 @@ env:
   UV_SYSTEM_PYTHON: true
   UV_TORCH_BACKEND: auto
   TRANSFORMERS_IS_CI: true
-  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
+  HF_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
 
 jobs:
   build:

.github/workflows/test_exporters_onnx.yml

@@ -14,7 +14,7 @@ env:
   UV_SYSTEM_PYTHON: true
   UV_TORCH_BACKEND: auto
   TRANSFORMERS_IS_CI: true
-  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
+  HF_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
 
 jobs:
   build:

.github/workflows/test_exporters_onnx_cli.yml

@@ -14,7 +14,7 @@ env:
   UV_SYSTEM_PYTHON: true
   UV_TORCH_BACKEND: auto
   TRANSFORMERS_IS_CI: true
-  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
+  HF_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
 
 jobs:
   build:

.github/workflows/test_onnxruntime.yml

@@ -15,7 +15,7 @@ env:
   UV_SYSTEM_PYTHON: true
   UV_TORCH_BACKEND: auto
   TRANSFORMERS_IS_CI: true
-  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
+  HF_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
 
 jobs:
   build:

.github/workflows/test_onnxruntime_slow.yml

@@ -18,7 +18,9 @@ concurrency:
   cancel-in-progress: true
 
 env:
+  RUN_SLOW: true
   TRANSFORMERS_IS_CI: true
+  HF_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
 
 jobs:
   build:
@@ -69,6 +71,3 @@ jobs:
       - name: Test with pytest
         run: |
           pytest tests/onnxruntime --durations=0 -vvvv
-        env:
-          HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
-          RUN_SLOW: 1

docs/source/onnx/usage_guides/contribute.mdx

@@ -175,7 +175,7 @@ This function expects the ONNX configuration, along with the base model, and the
 
 ```python
 >>> from pathlib import Path
->>> from optimum.exporters import TasksManager
+>>> from optimum.exporters.tasks import TasksManager
 >>> from optimum.exporters.onnx import export
 >>> from transformers import AutoModel
 

docs/source/onnx/usage_guides/export_a_model.mdx

@@ -391,18 +391,9 @@ class CustomMPTOnnxConfig(TextDecoderOnnxConfig):
 
 model_id = "fxmarty/tiny-mpt-random-remote-code"
 config = AutoConfig.from_pretrained(model_id, trust_remote_code=True)
-
-onnx_config = CustomMPTOnnxConfig(
-    config=config,
-    task="text-generation",
-    use_past_in_inputs=False,
-)
 onnx_config_with_past = CustomMPTOnnxConfig(config, task="text-generation", use_past=True)
 
-custom_onnx_configs = {
-    "decoder_model": onnx_config,
-    "decoder_with_past_model": onnx_config_with_past,
-}
+custom_onnx_configs = {"model": onnx_config_with_past}
 
 main_export(
     model_id,
@@ -411,7 +402,6 @@ main_export(
     trust_remote_code=True,
     custom_onnx_configs=custom_onnx_configs,
     no_post_process=True,
-    legacy=True,
     opset=14
 )
 ```

examples/onnxruntime/optimization/multiple-choice/run_swag.py

@@ -134,13 +134,6 @@ class OptimizationArguments:
             "99 will enable all available optimizations including layout optimizations."
         },
     )
-    optimize_with_onnxruntime_only: bool = field(
-        default=False,
-        metadata={
-            "help": "Whether to only use ONNX Runtime to optimize the model and no graph fusion in Python."
-            "Graph fusion might require offline, Python scripts, to be run."
-        },
-    )
     optimize_for_gpu: bool = field(
         default=False,
         metadata={
@@ -232,7 +225,6 @@ def main():
     # Create the optimization configuration containing all the optimization parameters
     optimization_config = OptimizationConfig(
         optimization_level=optim_args.optimization_level,
-        optimize_with_onnxruntime_only=optim_args.optimize_with_onnxruntime_only,
         optimize_for_gpu=optim_args.optimize_for_gpu,
     )
 

examples/onnxruntime/optimization/question-answering/run_qa.py

@@ -202,13 +202,7 @@ class OptimizationArguments:
             "99 will enable all available optimizations including layout optimizations."
         },
     )
-    optimize_with_onnxruntime_only: bool = field(
-        default=False,
-        metadata={
-            "help": "Whether to only use ONNX Runtime to optimize the model and no graph fusion in Python."
-            "Graph fusion might require offline, Python scripts, to be run."
-        },
-    )
+
     optimize_for_gpu: bool = field(
         default=False,
         metadata={
@@ -301,7 +295,6 @@ def main():
     # Create the optimization configuration containing all the optimization parameters
     optimization_config = OptimizationConfig(
         optimization_level=optim_args.optimization_level,
-        optimize_with_onnxruntime_only=optim_args.optimize_with_onnxruntime_only,
         optimize_for_gpu=optim_args.optimize_for_gpu,
     )
 

examples/onnxruntime/optimization/text-classification/run_glue.py

@@ -150,13 +150,6 @@ class OptimizationArguments:
             "99 will enable all available optimizations including layout optimizations."
         },
     )
-    optimize_with_onnxruntime_only: bool = field(
-        default=False,
-        metadata={
-            "help": "Whether to only use ONNX Runtime to optimize the model and no graph fusion in Python."
-            "Graph fusion might require offline, Python scripts, to be run."
-        },
-    )
     optimize_for_gpu: bool = field(
         default=False,
         metadata={
@@ -249,7 +242,6 @@ def main():
     # Create the optimization configuration containing all the optimization parameters
     optimization_config = OptimizationConfig(
         optimization_level=optim_args.optimization_level,
-        optimize_with_onnxruntime_only=optim_args.optimize_with_onnxruntime_only,
         optimize_for_gpu=optim_args.optimize_for_gpu,
     )