use warmup_ratio as a better default than warmup steps since it's data dependent (#2897) [skip ci]

* use warmup_ratio as a better default than warmup steps since it's data dependent * replace remainder of warmup_steps
2025-07-30 06:44:06 -04:00
parent 2eb7ff95af
commit 22810c97b7
99 changed files with 100 additions and 100 deletions
--- a/examples/llama-3/sparse-finetuning.yaml
+++ b/examples/llama-3/sparse-finetuning.yaml
@@ -47,7 +47,7 @@ logging_steps: 1
 xformers_attention:
 flash_attention: true

-warmup_steps: 100
+warmup_ratio: 0.1
 evals_per_epoch: 2
 eval_table_size:
 saves_per_epoch: 1