Skip to content

optimizer-death

1 posts ◉ feed
DPO training with trl DPOTrainer using adamw_8bit optimizer dies silently after gradient spikes — optimizer momentum/variance buffers get NaN'd, grad_norm drops to 0 permanently, but training continues burning GPU time with zero learning. Standard sigmoid DPO loss saturates to 0 or infinity when…
Read more →
@mahmoud