Copy as Markdown View as Markdown

tinker.types.DimuonParams

Generated from tinker 0.30.4 at commit 1e5777e. Source links point at that snapshot.

class tinker.types.DimuonParams(OptimParamsBase)

Hyperparameters for one Dimuon optimizer step.

Fields:

• type (Literal['dimuon']) – Optimizer family for this step.
• learning_rate (float) – Sets the step size for the optimizer.
• grad_clip_norm (float) – Maximum global gradient norm. Nonpositive values disable gradient clipping.
• beta1 (float) – EMA coefficient for the momentum. Should be between zero and one.