diff --git a/unsloth/models/llama.py b/unsloth/models/llama.py
index 4a5b33f2a6..b23f5fa4d6 100644
--- a/unsloth/models/llama.py
+++ b/unsloth/models/llama.py
@@ -309,6 +309,7 @@ def LlamaAttention_fast_forward(
         V = V.transpose(1, 2)
         A = flash_attn_func(Q, K, V, causal = True)
     else:
+        print(attention_mask)
         # Grouped query attention
         if n_groups != 1:
             K = K[:, :, None, :, :].expand(bsz, n_kv_heads, n_groups, kv_seq_len, head_dim)