Search results

Keyerror when resuming training with optimizer adamw_stable

Status: Open.

#12 In warner-benjamin/optimi;

· lspindler2509 opened on Dec 15, 2025

Is this LLM-only?

Status: Open.

#11 In warner-benjamin/optimi;

· ogencoglu opened on Dec 6, 2025

pure fp16 ?

Status: Open.

#8 In warner-benjamin/optimi;

· vince62s opened on Mar 8, 2025

Casting Existing FP32/FP16 model weights to BF16 + Kahan Summation

Status: Open.

#7 In warner-benjamin/optimi;

· zaptrem opened on Jul 30, 2024

[FeatureRequest]gradient-descent-the-ultimate-optimizer

Status: Open.

#6 In warner-benjamin/optimi;

· sdbds opened on Jul 2, 2024

When using accelerator with Gradient Release hook, Increasing VRAM consumption

Status: Open.

#5 In warner-benjamin/optimi;

· sdbds opened on Jun 26, 2024