nemo_gym.base_resources_server

View as Markdown

Module Contents

Classes

NameDescription
BaseMultiRewardVerifyResponseBase verify response for environments with multiple reward objectives.
BaseResourcesServer-
BaseResourcesServerConfig-
BaseRunRequest-
BaseSeedSessionRequest-
BaseSeedSessionResponse-
BaseVerifyRequest-
BaseVerifyResponse-
MCPServerMetadataMetadata returned from /seed_session for per-rollout Gym MCP access.
ResourcesCloseSessionRequestClose resources-server state.
ResourcesCloseSessionResponseConfirm resources-server state was closed.
ResourcesSeedSessionRequestInitialize resources-server state for one episode.
ResourcesSeedSessionResponseReturn resources state and optional agent access.
ResourcesVerifyRequestCarry typed environment output to a resources server.
ResourcesVerifyResponsePreserve environment-specific verification fields across the server boundary.
ReverifyMode-
SimpleResourcesServer-

Functions

NameDescription
normalize_tool_nameMap a trajectory tool-call name to the server’s bare tool name.

Data

NEMO_GYM_MCP_METADATA_KEY

NEMO_GYM_MCP_SESSION_TOKEN_HEADER

RESERVED_MCP_TOOL_NAMES

VerificationInputT

_MCP_TOKEN_SALT

API

class nemo_gym.base_resources_server.BaseMultiRewardVerifyResponse()

Bases: BaseVerifyResponse

Base verify response for environments with multiple reward objectives.

Subclass this response instead of declaring reward_components on an environment-specific BaseVerifyResponse subclass. The mapping is required, and its objective keys should remain consistent across every task in the environment.

Set the inherited reward to the scalar aggregate expected by single-reward consumers. To include individual objectives in aggregate metrics, also expose them as top-level numeric fields because metrics do not descend into this mapping. See resources_servers/example_tool_call_multireward for a complete example.

reward_components
dict[str, float]
class nemo_gym.base_resources_server.BaseResourcesServer()

Bases: BaseServer

config
BaseResourcesServerConfig
class nemo_gym.base_resources_server.BaseResourcesServerConfig()

Bases: BaseRunServerInstanceConfig

REVERIFY_MODE
ReverifyMode = ReverifyMode.UNKNOWN
expose_tools_over_mcp
bool = False
class nemo_gym.base_resources_server.BaseRunRequest()

Bases: BaseModel

capture_rollout_id
Optional[str]
model_config
= ConfigDict(populate_by_name=True)
responses_create_params
NeMoGymResponseCreateParamsNonStreaming
class nemo_gym.base_resources_server.BaseSeedSessionRequest()

Bases: BaseModel

class nemo_gym.base_resources_server.BaseSeedSessionResponse()

Bases: BaseModel

class nemo_gym.base_resources_server.BaseVerifyRequest()

Bases: BaseRunRequest

response
NeMoGymResponse
class nemo_gym.base_resources_server.BaseVerifyResponse()

Bases: BaseVerifyRequest

failure_kind
Optional[str]
failure_reason
Optional[str] = None
mask_sample
bool
reward
float
nemo_gym.base_resources_server.BaseVerifyResponse._warn_on_unregistered_failure_kind(
value: typing.Optional[str]
) -> typing.Optional[str]
classmethod

Producers learn about a name outside the vocabulary without losing the failure.

Validation warns rather than rejects: an unregistered kind is a migration signal, and dropping the response over it would replace a visible wrong label with an invisible lost failure.

class nemo_gym.base_resources_server.MCPServerMetadata()

Bases: BaseModel

Metadata returned from /seed_session for per-rollout Gym MCP access.

headers
dict[str, str]
server_name
str
transport
str = 'http'
url_path
str = '/mcp'
class nemo_gym.base_resources_server.ResourcesCloseSessionRequest()

Bases: BaseModel

Close resources-server state.

episode_id
EpisodeId
model_config
= ConfigDict(extra='forbid')
resources_session_id
str
class nemo_gym.base_resources_server.ResourcesCloseSessionResponse()

Bases: BaseModel

Confirm resources-server state was closed.

model_config
= ConfigDict(extra='forbid')
resources_session_id
str
class nemo_gym.base_resources_server.ResourcesSeedSessionRequest()

Bases: BaseModel

Initialize resources-server state for one episode.

episode_id
EpisodeId
model_config
= ConfigDict(extra='forbid')
task_data
dict[str, JsonValue]
task_id
TaskId
class nemo_gym.base_resources_server.ResourcesSeedSessionResponse()

Bases: BaseModel

Return resources state and optional agent access.

model_config
= ConfigDict(extra='forbid')
resources_session_id
str
resources_tools
MCPServerMetadata | None = None
sandbox_access
SandboxAccess | None = None
class nemo_gym.base_resources_server.ResourcesVerifyRequest()

Bases: BaseModel, Generic[VerificationInputT]

Carry typed environment output to a resources server.

episode_id
EpisodeId
model_config
= ConfigDict(extra='forbid')
task_id
TaskId
verification_input
VerificationInputT
class nemo_gym.base_resources_server.ResourcesVerifyResponse()

Bases: BaseVerifyResponse

Preserve environment-specific verification fields across the server boundary.

model_config
= ConfigDict(extra='allow')
class nemo_gym.base_resources_server.ReverifyMode

Bases: enum.Enum

STATELESS
= 'stateless'
UNKNOWN
= 'unknown'
UNSUPPORTED
= 'unsupported'
class nemo_gym.base_resources_server.SimpleResourcesServer()

Bases: BaseResourcesServer, AggregateMetricsMixin, SimpleServer

config
BaseResourcesServerConfig
nemo_gym.base_resources_server.SimpleResourcesServer.aggregate_metrics(
async

Compute aggregate metrics from verify responses.

RewardProfiler provides baseline stats. Override compute_metrics() and/or get_key_metrics() for benchmark-specific customization.

nemo_gym.base_resources_server.SimpleResourcesServer.get_reverify_mode() -> nemo_gym.base_resources_server.ReverifyMode
async
nemo_gym.base_resources_server.SimpleResourcesServer.mcp_allowed_tools_for_session(
seed_body: dict[str, typing.Any]
) -> typing.Optional[list[str]]

Per-session tool restriction: return the tool names allowed for this rollout’s MCP token, or None (the default) for unrestricted. seed_body is the JSON body POSTed to /seed_session.

nemo_gym.base_resources_server.SimpleResourcesServer.mcp_tools(
catchall: typing.Optional[typing.Any]
) -> typing.Optional[list[nemo_gym.mcp_auto_exposure.MCPTool]]

Return the MCP tools to expose (default: the auto-harvested typed POST routes).

Override to exclude (filter harvested), add catch-all-backed tools (harvested + [catchall.tool(…)]), or disable (return None). ‘catchall’ is None unless the server has one parameterized catch-all route.

nemo_gym.base_resources_server.SimpleResourcesServer.normalize_tool_name(
name: str
) -> str

Strip this server’s MCP namespace from a trajectory tool-call name (see module function).

nemo_gym.base_resources_server.SimpleResourcesServer.seed_session(
async
nemo_gym.base_resources_server.SimpleResourcesServer.setup_webserver() -> fastapi.FastAPI
nemo_gym.base_resources_server.SimpleResourcesServer.verify(
asyncabstract
nemo_gym.base_resources_server.normalize_tool_name(
name: str,
server_name: typing.Optional[str] = None
) -> str

Map a trajectory tool-call name to the server’s bare tool name.

HTTP-driven agents record bare tool names (“email_reply_email”); MCP-native agents (e.g. Claude Code) record them namespaced per server (“mcp__workplace_assistant__email_reply_email”). Verifiers compare trajectory names against dataset/ground-truth vocabulary, so names are normalized before verify sees them and rollouts score identically on both transports. Non-namespaced names pass through unchanged. When server_name is given, only that server’s prefix is stripped (robust to tool names that themselves contain double underscores). This runs only for servers exposed over MCP and mirrors how MCP clients namespace tool names, so a real tool that is itself named mcp__<server>__x being stripped is accepted.

nemo_gym.base_resources_server.NEMO_GYM_MCP_METADATA_KEY = 'mcp'
nemo_gym.base_resources_server.NEMO_GYM_MCP_SESSION_TOKEN_HEADER = 'X-NeMo-Gym-Session-Token'
nemo_gym.base_resources_server.RESERVED_MCP_TOOL_NAMES = frozenset({'verify', 'seed_session', 'close_session', 'aggregate_metrics', 'mcp'...
nemo_gym.base_resources_server.VerificationInputT = TypeVar('VerificationInputT')
nemo_gym.base_resources_server._MCP_TOKEN_SALT = 'nemo-gym-mcp-session-token'