add the ability to run eager runner via buck

Differential Revision: D64730344

Author: Lunwen He

examples/models/llama/runner/TARGETS

@@ -1,8 +1,41 @@
 # Any targets that should be shared between fbcode and xplat must be defined in
 # targets.bzl. This file can contain fbcode-only targets.
 
+load("@fbsource//xplat/executorch/build:runtime_wrapper.bzl", "runtime")
 load(":targets.bzl", "define_common_targets")
 
 oncall("executorch")
 
 define_common_targets()
+
+runtime.python_library(
+    name = "eager_runner_library",
+    srcs = [
+        "eager.py",
+        "generation.py"
+    ],
+    _is_external_target = True,
+    base_module = "executorch.examples.models.llama.runner",
+    visibility = [
+        "//bento/...",
+        "//bento_kernels/...",
+        "//executorch/examples/...",
+        "@EXECUTORCH_CLIENTS",
+    ],
+    deps = [
+        "//executorch/examples/models/llama:export_library",
+    ],
+)
+
+runtime.python_binary(
+    name = "eager",
+    main_function = "executorch.examples.models.llama.runner.eager.main",
+    preload_deps = [
+        "//executorch/extension/llm/custom_ops:custom_ops_aot_lib",
+        "//executorch/kernels/quantized:aot_lib",
+    ],
+    deps = [
+        ":eager_runner_library",
+        "//caffe2:torch",
+    ],
+)

examples/models/llama/runner/eager.py

@@ -9,14 +9,13 @@
 from typing import Optional
 
 import torch
-
-from examples.models.llama.llama_transformer import ModelArgs
 from executorch.examples.models.llama.export_llama_lib import (
     _prepare_for_llama_export,
     build_args_parser as _build_args_parser,
 )
+from executorch.examples.models.llama.llama_transformer import ModelArgs
 from executorch.examples.models.llama.runner.generation import LlamaRunner
-from executorch.extension.llm.export import LLMEdgeManager
+from executorch.extension.llm.export.builder import LLMEdgeManager
 
 
 class EagerLlamaRunner(LlamaRunner):