enable odml-torch as default

PiperOrigin-RevId: 705931084

Author: chunnienc

.github/workflows/model_coverage.yml

@@ -19,7 +19,7 @@ jobs:
   test-model-coverage:
     strategy:
       matrix:
-        python-version: ["3.9", "3.10", "3.11"]
+        python-version: ["3.10", "3.11"]
     runs-on:
       labels: Linux_runner_8_core
     steps:

.github/workflows/nightly_release.yml

@@ -64,7 +64,7 @@ jobs:
     needs: build-and-release-nightly
     strategy:
       matrix:
-        python-version: ["3.9", "3.10", "3.11"]
+        python-version: ["3.10", "3.11"]
 
     name: Test Install and Import ai-edge-torch
     runs-on: ubuntu-latest

.github/workflows/unittests_python.yml

@@ -19,7 +19,7 @@ jobs:
   test:
     strategy:
       matrix:
-        python-version: ["3.9", "3.10", "3.11"]
+        python-version: ["3.10", "3.11"]
     runs-on:
       labels: Linux_runner_8_core
     steps:

README.md

@@ -77,7 +77,7 @@ Nightly Release    | [![](https://github.com/google-ai-edge/ai-edge-torch/action
 
 ### Requirements and Dependencies
 
- * Python versions:  3.9, 3.10, 3.11
+ * Python versions: >=3.10
  * Operating system: Linux
  * PyTorch: [![torch](https://img.shields.io/badge/torch->=2.4.0-blue)](https://pypi.org/project/torch/)
  * TensorFlow: [![tf-nightly](https://img.shields.io/badge/tf--nightly-latest-blue)](https://pypi.org/project/tf-nightly/)

ai_edge_torch/__init__.py

@@ -13,13 +13,13 @@
 # limitations under the License.
 # ==============================================================================
 
+from ai_edge_torch._config import config
 from ai_edge_torch._convert.converter import convert
 from ai_edge_torch._convert.converter import signature
 from ai_edge_torch._convert.to_channel_last_io import to_channel_last_io
 from ai_edge_torch.model import Model
 from ai_edge_torch.version import __version__
 
-
 def load(path: str) -> Model:
   """Imports an ai_edge_torch model from disk.
 

ai_edge_torch/_config.py

@@ -0,0 +1,52 @@
+# Copyright 2024 The AI Edge Torch Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+"""Provides a configuration for the ai-edge-torch."""
+
+import functools
+import logging
+import os
+
+__all__ = ["config"]
+
+
+class _Config:
+  """ai-edge-torch global configs."""
+
+  @property
+  @functools.cache  # pylint: disable=method-cache-max-size-none
+  def use_torch_xla(self) -> bool:
+    """True if using torch_xla to lower torch ops to StableHLO.
+
+    To use torch_xla as the lowering backend, set environment variable
+    `USE_TORCH_XLA` to "true".
+    """
+    var = os.environ.get("USE_TORCH_XLA", "false")
+    var = var.lower().strip()
+    if var in ("y", "yes", "t", "true", "on", "1"):
+      return True
+    elif var in ("n", "no", "f", "false", "off", "0"):
+      return False
+    else:
+      logging.warning("Invalid USE_TORCH_XLA value is ignored: %s.", var)
+      return False
+
+  @property
+  def in_oss(self) -> bool:
+    """True if the code is not running in google internal environment."""
+    return True
+
+
+config = _Config()

ai_edge_torch/_convert/test/test_convert.py

@@ -19,7 +19,6 @@
 from typing import Tuple
 
 import ai_edge_torch
-from ai_edge_torch import config
 from ai_edge_torch._convert import conversion_utils
 from ai_edge_torch.quantize import pt2e_quantizer
 from ai_edge_torch.testing import model_coverage
@@ -292,7 +291,7 @@ def test_convert_conv_transpose_batch_norm(self):
     self.assertTrue(result)
 
   @googletest.skipIf(
-      not config.Config.use_torch_xla,
+      not ai_edge_torch.config.use_torch_xla,
       reason="Shape polymorphism is not yet support with odml_torch.",
   )
   def test_convert_model_with_dynamic_batch(self):

ai_edge_torch/generative/fx_passes/test/test_remove_sdpa_zero_mask_pass.py

@@ -15,7 +15,7 @@
 import re
 from typing import Callable, Union
 
-from ai_edge_torch import config
+import ai_edge_torch
 from ai_edge_torch import fx_pass_base
 from ai_edge_torch import lowertools
 from ai_edge_torch.generative.fx_passes import CanonicalizePass
@@ -112,7 +112,7 @@ def get_model_config() -> unet_cfg.AttentionBlock2DConfig:
         (torch.rand(1, 512, 64, 64),),
     )
 
-    if config.Config.use_torch_xla:
+    if ai_edge_torch.config.use_torch_xla:
       self.assertTrue(
           re.search(
               'stablehlo\.composite "odml\.scaled_dot_product_attention" %\d+,'

ai_edge_torch/generative/test/test_model_conversion.py

@@ -16,7 +16,6 @@
 """Testing model conversion for a few gen-ai models."""
 
 import ai_edge_torch
-from ai_edge_torch import config as ai_edge_config
 from ai_edge_torch.generative.examples.test_models import toy_model_with_kv_cache
 from ai_edge_torch.generative.examples.tiny_llama import tiny_llama
 from ai_edge_torch.generative.layers import kv_cache
@@ -83,22 +82,22 @@ def _test_model_with_kv_cache(self, enable_hlfb: bool):
     )
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_toy_model_with_kv_cache(self):
     self._test_model_with_kv_cache(enable_hlfb=False)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_toy_model_with_kv_cache_with_hlfb(self):
     self._test_model_with_kv_cache(enable_hlfb=True)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_toy_model_has_dus_op(self):
     """Tests that the model has the dynamic update slice op."""
@@ -179,8 +178,8 @@ def _test_multisig_model(self, config, pytorch_model, atol, rtol):
     )
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_tiny_llama_multisig(self):
     config = tiny_llama.get_fake_model_config()

ai_edge_torch/generative/test/test_model_conversion_large.py

@@ -16,7 +16,6 @@
 """Testing model conversion for a few gen-ai models."""
 
 import ai_edge_torch
-from ai_edge_torch import config as ai_edge_config
 from ai_edge_torch.generative.examples.amd_llama_135m import amd_llama_135m
 from ai_edge_torch.generative.examples.gemma import gemma1
 from ai_edge_torch.generative.examples.gemma import gemma2
@@ -91,35 +90,35 @@ def _test_model(self, config, model, signature_name, atol, rtol):
     )
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_gemma1(self):
     config = gemma1.get_fake_model_config()
     pytorch_model = gemma1.Gemma1(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-3, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_gemma2(self):
     config = gemma2.get_fake_model_config()
     pytorch_model = gemma2.Gemma2(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-4, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_llama(self):
     config = llama.get_fake_model_config()
     pytorch_model = llama.Llama(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-3, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_phi2(self):
     config = phi2.get_fake_model_config()
@@ -128,53 +127,53 @@ def test_phi2(self):
     self._test_model(config, pytorch_model, "prefill", atol=1e-3, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_phi3(self):
     config = phi3.get_fake_model_config()
     pytorch_model = phi3.Phi3_5Mini(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-5, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_smollm(self):
     config = smollm.get_fake_model_config()
     pytorch_model = smollm.SmolLM(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-4, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_openelm(self):
     config = openelm.get_fake_model_config()
     pytorch_model = openelm.OpenELM(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-4, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_qwen(self):
     config = qwen.get_fake_model_config()
     pytorch_model = qwen.Qwen(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-3, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_amd_llama_135m(self):
     config = amd_llama_135m.get_fake_model_config()
     pytorch_model = amd_llama_135m.AmdLlama(config).eval()
     self._test_model(config, pytorch_model, "prefill", atol=1e-5, rtol=1e-5)
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def disabled_test_paligemma(self):
     config = paligemma.get_fake_model_config()
@@ -222,8 +221,8 @@ def disabled_test_paligemma(self):
     )
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_stable_diffusion_clip(self):
     config = sd_clip.get_fake_model_config()
@@ -254,8 +253,8 @@ def test_stable_diffusion_clip(self):
     )
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_stable_diffusion_diffusion(self):
     config = sd_diffusion.get_fake_model_config(2)
@@ -296,8 +295,8 @@ def test_stable_diffusion_diffusion(self):
     )
 
   @googletest.skipIf(
-      ai_edge_config.Config.use_torch_xla,
-      reason="tests with custom ops are not supported on oss",
+      ai_edge_torch.config.in_oss,
+      reason="tests with custom ops are not supported in oss",
   )
   def test_stable_diffusion_decoder(self):
     config = sd_decoder.get_fake_model_config()

ai_edge_torch/hlfb/test/test_stablehlo_composite_builder.py

@@ -16,7 +16,8 @@
 
 import math
 
-from ai_edge_torch import config
+import ai_edge_torch
+from ai_edge_torch import hlfb
 from ai_edge_torch import lowertools
 from ai_edge_torch.hlfb import StableHLOCompositeBuilder
 import torch
@@ -29,9 +30,11 @@ def _export_stablehlo_mlir(model, args):
   ep = torch.export.export(model, args)
   return lowertools.exported_program_to_mlir_text(ep)
 
+StableHLOCompositeBuilder = hlfb.StableHLOCompositeBuilder
+
 
 @googletest.skipIf(
-    not config.Config.use_torch_xla,
+    not ai_edge_torch.config.use_torch_xla,
     reason="The odml_torch counter part is in odml_torch.",
 )
 class TestStableHLOCompositeBuilder(googletest.TestCase):

ai_edge_torch/lowertools/_shim.py

@@ -15,13 +15,15 @@
 
 from typing import Any, Optional
 
-from ai_edge_torch import config
+from ai_edge_torch import _config
 from ai_edge_torch._convert import signature
 from ai_edge_torch.quantize import quant_config as qcfg
 import torch
 
+config = _config.config
+
 # isort: off
-if config.Config.use_torch_xla:
+if config.use_torch_xla:
   from ai_edge_torch.lowertools import torch_xla_utils as utils
   from ai_edge_torch.lowertools.torch_xla_utils import exported_program_to_mlir_text
   from torch_xla.experimental.mark_pattern_utils import StableHLOCompositeBuilder