simplify

Author: cmunley1

resources_servers/verifiers/schemas.py

@@ -22,7 +22,7 @@
     BaseVerifyRequest,
     BaseVerifyResponse,
 )
-from nemo_gym.openai_utils import NeMoGymResponse, NeMoGymResponseCreateParamsNonStreaming, NeMoGymResponseOutputItem
+from nemo_gym.openai_utils import NeMoGymResponse, NeMoGymResponseCreateParamsNonStreaming
 
 
 class VerifiersResourcesServerConfig(BaseResourcesServerConfig):
@@ -66,27 +66,6 @@ class VerifiersCloseResponse(BaseModel):
     success: bool
 
 
-class VerifiersAgentConfig(BaseModel):
-    model_config = ConfigDict(extra="allow")
-
-    vf_env_id: str = Field(description="Default verifiers environment ID")
-    vf_env_args: dict = Field(default_factory=dict, description="Default environment arguments")
-    dataset_n: int = Field(default=-1, description="Number of examples to load")
-    dataset_seed: int | None = Field(default=None, description="Seed for dataset shuffling")
-    group_size: int = Field(default=1, description="Number of rollouts per example")
-    max_concurrent_generation: int = Field(default=-1, description="Max concurrent generation")
-    max_concurrent_scoring: int = Field(default=-1, description="Max concurrent scoring")
-
-
-class VerifiersAgentRunRequest(BaseModel):
-    model_config = ConfigDict(extra="allow")
-
-    task_idx: int = Field(description="Index of the example to run")
-    responses_create_params: NeMoGymResponseCreateParamsNonStreaming = Field(
-        default_factory=lambda: NeMoGymResponseCreateParamsNonStreaming(input=[])
-    )
-
-
 class VerifiersNeMoGymResponse(NeMoGymResponse):
     env_id: str
     group_id: str

responses_api_agents/verifiers_agent/app.py

@@ -12,8 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import logging
-import sys
 import traceback
+import uuid
 from typing import Any
 
 import aiohttp
@@ -34,7 +34,6 @@
     VerifiersAgentVerifyRequest,
     VerifiersAgentVerifyResponse,
     VerifiersNeMoGymResponse,
-    VerifiersSeedSessionResponse,
 )
 
 
@@ -71,7 +70,6 @@ async def create(self, *args: Any, **kwargs: Any) -> ChatCompletion:
         choice_dict = response_dict["choices"][0]
         message_dict = choice_dict.get("message", {})
 
-
         prompt_token_ids = message_dict.pop("prompt_token_ids", [])
         generation_token_ids = message_dict.pop("generation_token_ids", [])
         generation_log_probs = message_dict.pop("generation_log_probs", [])
@@ -164,7 +162,6 @@ async def _ensure_env_loaded(self, vf_env_id: str) -> tuple[vf.Environment, str,
         if vf_env_id in _ENVS_CACHE:
             return _ENVS_CACHE[vf_env_id], _ENV_IDS_CACHE[vf_env_id], _DATASET_ROWS_CACHE[vf_env_id]
 
-        import uuid
         env_id = f"{vf_env_id}-{uuid.uuid4().hex[:8]}"
         logger.info(f"Loading verifiers environment: {vf_env_id}")
 
@@ -226,9 +223,8 @@ def _convert_trajectory_to_output(self, state: dict) -> list:
         from nemo_gym.openai_utils import (
             NeMoGymEasyInputMessage,
             NeMoGymResponseOutputMessage,
-            NeMoGymResponseOutputText,
-            NeMoGymEasyInputMessageForTraining,
             NeMoGymResponseOutputMessageForTraining,
+            NeMoGymResponseOutputText,
         )
 
         output = []