LMEvalAccuracyEvaluatorConfig
=============================

**class** ``LMEvalAccuracyEvaluatorConfig`` **(** :doc:`BaseEvaluatorConfig` **)**

    Configuration for lm-eval accuracy evaluation (task-specific correctness).

**Polymorphic Type:**

    ``type: accuracy_lmeval``

    All ``BaseEvaluatorConfig`` types:

    - ``performance``: :doc:`PerformanceEvaluatorConfig`
    - ``accuracy_lmeval``: :doc:`LMEvalAccuracyEvaluatorConfig`

**Fields:**

``target_channels`` : *list* = ``['text']``
    List of ChannelModality values to evaluate.

``slos`` : *list* [ :doc:`BaseSloConfig` ] = ``[<ConstantSloConfig>]``
    List of SLO definitions to evaluate against request-level metrics.

    Available types:

    - ``constant``: :doc:`ConstantSloConfig`

``stream_metrics`` : *bool* = ``True``
    Enable real-time metric streaming

``stream_metrics_interval`` : *float* = ``5.0``
    Interval for streaming metrics in seconds

``bootstrap_iters`` : *int* = ``100000``
    Bootstrap iterations for confidence intervals