Add 'auto' MODEL_IMPL_TYPE that resolves based on architecture

- Add 'auto' as default value for MODEL_IMPL_TYPE env var
- For GptOssForCausalLM, 'auto' resolves to 'vllm' for better performance
- For all other architectures, 'auto' resolves to 'flax_nnx'
- Add _VLLM_REQUIRED_ARCHITECTURES frozenset in model_loader.py
- Use match/case pattern in get_model() for implementation selection
- Add tests for 'auto' resolution behavior

Signed-off-by: Xing Liu <xingliu14@gmail.com>

Author: xingliu14

tests/models/common/test_model_loader.py

@@ -381,3 +381,43 @@ def test_get_model_not_implemented(self, mock_get_flax, mock_get_vllm,
 
         mock_get_flax.assert_not_called()
         mock_get_vllm.assert_not_called()
+
+    @patch.dict(os.environ, {"MODEL_IMPL_TYPE": "auto"}, clear=True)
+    @patch("tpu_inference.models.common.model_loader.get_vllm_model")
+    @patch("tpu_inference.models.common.model_loader.get_flax_model")
+    def test_get_model_auto_resolves_to_flax_nnx(self, mock_get_flax,
+                                                 mock_get_vllm, vllm_config,
+                                                 rng, mesh):
+        """
+        Tests that 'auto' resolves to 'flax_nnx' for standard architectures
+        (not in _VLLM_REQUIRED_ARCHITECTURES).
+        """
+        # vllm_config uses Qwen3 which is NOT in _VLLM_REQUIRED_ARCHITECTURES
+        mock_get_flax.return_value = "flax_model_sentinel"
+
+        result = model_loader.get_model(vllm_config, rng, mesh)
+
+        mock_get_flax.assert_called_once_with(vllm_config, rng, mesh, False)
+        mock_get_vllm.assert_not_called()
+        assert result == "flax_model_sentinel"
+
+    @patch.dict(os.environ, {"MODEL_IMPL_TYPE": "auto"}, clear=True)
+    @patch("tpu_inference.models.common.model_loader.get_vllm_model")
+    @patch("tpu_inference.models.common.model_loader.get_flax_model")
+    def test_get_model_auto_resolves_to_vllm_for_gpt_oss(
+            self, mock_get_flax, mock_get_vllm, vllm_config, rng, mesh):
+        """
+        Tests that 'auto' resolves to 'vllm' for architectures in
+        _VLLM_REQUIRED_ARCHITECTURES (e.g., GptOssForCausalLM).
+        """
+        # Mock the architecture to be GptOssForCausalLM
+        vllm_config.model_config.hf_config.architectures = [
+            "GptOssForCausalLM"
+        ]
+        mock_get_vllm.return_value = "vllm_model_sentinel"
+
+        result = model_loader.get_model(vllm_config, rng, mesh)
+
+        mock_get_flax.assert_not_called()
+        mock_get_vllm.assert_called_once_with(vllm_config, rng, mesh)
+        assert result == "vllm_model_sentinel"

tests/test_envs.py

@@ -256,7 +256,7 @@ def test_disaggregated_serving_env_vars(monkeypatch: pytest.MonkeyPatch):
 
 def test_model_impl_type_default(monkeypatch: pytest.MonkeyPatch):
     monkeypatch.delenv("MODEL_IMPL_TYPE", raising=False)
-    assert envs.MODEL_IMPL_TYPE == "flax_nnx"
+    assert envs.MODEL_IMPL_TYPE == "auto"
 
 
 def test_cache_preserves_values_across_env_changes(

tpu_inference/envs.py

@@ -16,7 +16,7 @@
     DECODE_SLICES: str = ""
     SKIP_JAX_PRECOMPILE: bool = False
     VLLM_XLA_CHECK_RECOMPILATION: bool = False
-    MODEL_IMPL_TYPE: str = "flax_nnx"
+    MODEL_IMPL_TYPE: str = "auto"
     NEW_MODEL_DESIGN: bool = False
     PHASED_PROFILING_DIR: str = ""
     PYTHON_TRACER_LEVEL: int = 1
@@ -128,8 +128,8 @@ def _get_bool_env() -> bool:
     env_bool("VLLM_XLA_CHECK_RECOMPILATION", default=False),
     # Model implementation type (e.g., "flax_nnx")
     "MODEL_IMPL_TYPE":
-    env_with_choices("MODEL_IMPL_TYPE", "flax_nnx",
-                     ["vllm", "flax_nnx", "jetpack"]),
+    env_with_choices("MODEL_IMPL_TYPE", "auto",
+                     ["auto", "vllm", "flax_nnx", "jetpack"]),
     # Enable new experimental model design
     "NEW_MODEL_DESIGN":
     env_bool("NEW_MODEL_DESIGN", default=False),

tpu_inference/models/common/model_loader.py

@@ -333,6 +333,12 @@ def get_vllm_model(
     return jit_model, compute_logits_fn, combine_hidden_states_fn, None, params, lora_manager, model
 
 
+# Architectures that require "vllm" implementation type when MODEL_IMPL_TYPE is "auto".
+# These architectures are listed here because they have better performance with the
+# vLLM PyTorch backend compared to the flax_nnx JAX backend for now.
+_VLLM_REQUIRED_ARCHITECTURES: frozenset[str] = frozenset({"GptOssForCausalLM"})
+
+
 def get_model(
     vllm_config: VllmConfig,
     rng: jax.Array,
@@ -342,24 +348,37 @@ def get_model(
     impl = envs.MODEL_IMPL_TYPE
     logger.info(f"Loading model with MODEL_IMPL_TYPE={impl}")
 
-    if impl == "flax_nnx":
-        try:
-            # Try to load the flax model first
-            return get_flax_model(vllm_config, rng, mesh, is_draft_model)
-        except UnsupportedArchitectureError as e:
-            # Convert the error message to a string to check its contents
-            error_msg = str(e)
-
-            logger.warning(error_msg)
-
-            # Fall back to the vLLM model and updating the dtype accordingly
-            vllm_config.model_config.dtype = j2t_dtype(
-                vllm_config.model_config.dtype.dtype)
+    if impl == "auto":
+        # Resolve "auto" based on architecture
+        architectures = getattr(vllm_config.model_config.hf_config,
+                                "architectures", [])
+        for arch in architectures:
+            if arch in _VLLM_REQUIRED_ARCHITECTURES:
+                impl = "vllm"
+                break
+        else:
+            impl = "flax_nnx"
+        logger.info(f"Resolved MODEL_IMPL_TYPE 'auto' to '{impl}'")
+
+    match impl:
+        case "flax_nnx":
+            try:
+                # Try to load the flax model first
+                return get_flax_model(vllm_config, rng, mesh, is_draft_model)
+            except UnsupportedArchitectureError as e:
+                # Convert the error message to a string to check its contents
+                error_msg = str(e)
+
+                logger.warning(error_msg)
+
+                # Fall back to the vLLM model and updating the dtype accordingly
+                vllm_config.model_config.dtype = j2t_dtype(
+                    vllm_config.model_config.dtype.dtype)
+                return get_vllm_model(vllm_config, rng, mesh)
+        case "vllm":
             return get_vllm_model(vllm_config, rng, mesh)
-    elif impl == "vllm":
-        return get_vllm_model(vllm_config, rng, mesh)
-    else:
-        raise NotImplementedError("Unsupported MODEL_IMPL_TYPE")
+        case _:
+            raise NotImplementedError(f"Unsupported MODEL_IMPL_TYPE: {impl}")
 
 
 def _validate_model_interface(model: Any) -> None:

tpu_inference/runner/tpu_runner.py

@@ -1,6 +1,5 @@
 import copy
 import functools
-import os
 import random
 from contextlib import nullcontext
 from dataclasses import dataclass
@@ -1719,7 +1718,7 @@ def _sync_weights(
             shard=shard)
 
     def get_intermediate_tensor_spec(self, num_tokens: int):
-        impl = os.getenv("MODEL_IMPL_TYPE", "flax_nnx").lower()
+        impl = envs.MODEL_IMPL_TYPE
         jax_dtype = t2j_dtype(self.dtype) if impl == "vllm" else self.dtype
         num_padded_tokens = runner_utils.get_padded_token_len(
             self.num_tokens_paddings, num_tokens)