Skip to content

tinker.types.ForwardBackwardOutput

class tinker.types.ForwardBackwardOutput(**)

Fields:

  • loss_fn_output_type – The class name of the loss function output records (e.g., 'TorchLossReturn', 'ArrayRecord').
  • loss_fn_outputs – List of per-datum dicts mapping field names to TensorData.

  • metrics – Training metrics as key-value pairs.

    The following metrics are recorded only during MoE (Mixture of Experts) training.

    • e_frac_with_tokens:mean: Fraction of experts that received at least one token, averaged across layers. A value of 1.0 means every expert got work; 0.5 means half were idle. Decreasing over time is concerning (routing collapse).

    • e_frac_oversubscribed:mean: Fraction of experts receiving more tokens than perfect balance, averaged across layers. Increasing over time is concerning.

    • e_max_violation:mean: How much the most overloaded expert exceeds perfect balance, as a fraction of perfect balance, averaged across layers. Computed as (max_tokens - perfect_balance) / perfect_balance. A value of 2.0 means the busiest expert got 3x the fair share. Increasing over time is concerning.

    • e_max_violation:max: Same as e_max_violation:mean but takes the max across layers instead of the mean.

    • e_min_violation:mean: How much the least loaded expert is below perfect balance, as a fraction of perfect balance, averaged across layers. Typically negative; decreasing (more negative) is concerning.