Skip to content

feat: add bf16_loss training argument for VRAM-efficient QLoRA

3a6bedf
Select commit
Loading
Failed to load commit list.
Sign in for the full log view
Closed

feat: add bf16_loss training argument for VRAM-efficient QLoRA by keeping loss in BF16 during training to save ~1.4 GB VRAM #45769

feat: add bf16_loss training argument for VRAM-efficient QLoRA
3a6bedf
Select commit
Loading
Failed to load commit list.

Annotations

1 warning
comment
succeeded May 3, 2026 in 12m 10s