diff --git a/train.py b/train.py
index e7ce028..e8792c6 100644
--- a/train.py
+++ b/train.py
@@ -443,7 +443,7 @@ WEIGHT_DECAY = 0.2      # cautious weight decay for Muon
 ADAM_BETAS = (0.8, 0.95) # Adam beta1, beta2
 WARMUP_RATIO = 0.0      # fraction of time budget for LR warmup
 WARMDOWN_RATIO = 0.7    # fraction of time budget for LR warmdown
-FINAL_LR_FRAC = 0.0     # final LR as fraction of initial
+FINAL_LR_FRAC = 0.05    # final LR as fraction of initial
 
 # Model size
 DEPTH = 9               # number of transformer layers