feat: Yi models (#651)

Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>

openllm-python/src/openllm/_llm.py

@@ -36,9 +36,9 @@ from openllm_core.exceptions import MissingDependencyError
 from openllm_core.prompts import PromptTemplate
 from openllm_core.utils import (
   DEBUG,
+  ENV_VARS_TRUE_VALUES,
   ReprMixin,
   apply,
-  check_bool_env,
   codegen,
   converter,
   first_not_none,
@@ -326,7 +326,10 @@ class LLM(t.Generic[M, T], ReprMixin):
 
   @property
   def trust_remote_code(self) -> bool:
-    return first_not_none(check_bool_env('TRUST_REMOTE_CODE', False), default=self.__llm_trust_remote_code__)
+    env = os.getenv('TRUST_REMOTE_CODE')
+    if env is not None:
+      return str(env).upper() in ENV_VARS_TRUE_VALUES
+    return self.__llm_trust_remote_code__
 
   @property
   def runner_name(self) -> str:

openllm-python/src/openllm/bundle/_package.py

@@ -134,6 +134,7 @@ def construct_docker_options(
     'BENTOML_DEBUG': str(True),
     'BENTOML_QUIET': str(False),
     'BENTOML_CONFIG_OPTIONS': f"'{environ['BENTOML_CONFIG_OPTIONS']}'",
+    'TRUST_REMOTE_CODE': str(llm.trust_remote_code),
   }
   if adapter_map:
     env_dict['BITSANDBYTES_NOWELCOME'] = os.environ.get('BITSANDBYTES_NOWELCOME', '1')

openllm-python/src/openllm_cli/entrypoint.py

@@ -277,7 +277,7 @@ class OpenLLMCommandGroup(BentoMLCommandGroup):
   def list_commands(self, ctx: click.Context) -> list[str]:
     return super().list_commands(ctx) + t.cast('Extensions', extension_command).list_commands(ctx)
 
-  def command(self, *args: t.Any, **kwargs: t.Any) -> t.Callable[[t.Callable[..., t.Any]], click.Command]:  # type: ignore[override] # XXX: fix decorator on BentoMLCommandGroup
+  def command(self, *args: t.Any, **kwargs: t.Any) -> t.Callable[[t.Callable[..., t.Any]], click.Command]:
     """Override the default 'cli.command' with supports for aliases for given command, and it wraps the implementation with common parameters."""
     if 'context_settings' not in kwargs:
       kwargs['context_settings'] = {}
@@ -457,7 +457,6 @@ def start_command(
     quantize=quantize,
     serialisation=serialisation,
     torch_dtype=dtype,
-    trust_remote_code=check_bool_env('TRUST_REMOTE_CODE'),
   )
   backend_warning(llm.__llm_backend__)
 
@@ -635,6 +634,7 @@ def process_environ(
       'OPENLLM_BACKEND': llm.__llm_backend__,
       'OPENLLM_CONFIG': config.model_dump_json(flatten=True).decode(),
       'TORCH_DTYPE': str(llm._torch_dtype).split('.')[-1],
+      'TRUST_REMOTE_CODE': str(llm.trust_remote_code),
     }
   )
   if llm.quantise: