nemo_gym.dataset_viewer#

Module Contents#

Classes#

DatasetViewerVerifyResponse

JsonlDatasetViewerConfig

Launch a Gradio interface to view and explore dataset rollouts interactively.

Functions#

format_function_call_output

format_function_call

format_reasoning

format_message

convert_single_message

rollout_to_messages

extra_info_to_messages

get_aggregate_metrics

build_jsonl_dataset_viewer

main

API#

class nemo_gym.dataset_viewer.DatasetViewerVerifyResponse(/, **data: typing.Any)[source]#

Bases: nemo_gym.base_resources_server.BaseVerifyResponse

model_config#

‘ConfigDict(…)’

nemo_gym.dataset_viewer.format_function_call_output(
m: openai.types.responses.response_input_param.FunctionCallOutput,
) → List[gradio.ChatMessage][source]#
nemo_gym.dataset_viewer.format_function_call(
m: openai.types.responses.response_input_param.ResponseFunctionToolCallParam,
) → List[gradio.ChatMessage][source]#
nemo_gym.dataset_viewer.format_reasoning(
m: openai.types.responses.response_input_param.ResponseReasoningItemParam,
) → List[gradio.ChatMessage][source]#
nemo_gym.dataset_viewer.format_message(
m: openai.types.responses.response_input_param.EasyInputMessageParam,
) → List[gradio.ChatMessage][source]#
nemo_gym.dataset_viewer.convert_single_message(
m: openai.types.responses.response_input_param.ResponseInputItemParam,
) → List[gradio.ChatMessage][source]#
nemo_gym.dataset_viewer.rollout_to_messages(
create_params: dict,
response: dict,
) → List[gradio.ChatMessage][source]#
nemo_gym.dataset_viewer.extra_info_to_messages(
d: nemo_gym.dataset_viewer.DatasetViewerVerifyResponse,
) → List[gradio.ChatMessage][source]#
class nemo_gym.dataset_viewer.JsonlDatasetViewerConfig(/, **data: typing.Any)[source]#

Bases: nemo_gym.config_types.BaseNeMoGymCLIConfig

Launch a Gradio interface to view and explore dataset rollouts interactively.

Examples:

# Launch viewer with default settings (accessible from localhost only)
ng_viewer +jsonl_fpath=weather_rollouts.jsonl

# Accept requests from anywhere (e.g., for remote access)
ng_viewer +jsonl_fpath=weather_rollouts.jsonl +server_host=0.0.0.0

# Use a custom port
ng_viewer +jsonl_fpath=weather_rollouts.jsonl +server_port=8080

Initialization

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

jsonl_fpath: str#

‘Field(…)’

server_host: str | None#

‘Field(…)’

server_port: int | None#

‘Field(…)’

nemo_gym.dataset_viewer.get_aggregate_metrics(
data: List[nemo_gym.dataset_viewer.DatasetViewerVerifyResponse],
) → Dict[str, Any][source]#
nemo_gym.dataset_viewer.build_jsonl_dataset_viewer(
config: nemo_gym.dataset_viewer.JsonlDatasetViewerConfig,
) → gradio.Blocks[source]#
nemo_gym.dataset_viewer.main()[source]#