From abd4f9a10ed31d1664405318b7256bb54b6fdb22 Mon Sep 17 00:00:00 2001
From: Aman Gupta Karmani <aman@tmm1.net>
Date: Mon, 4 Sep 2023 17:43:52 -0400
Subject: [PATCH] log when xentropy is not found

---
 src/axolotl/monkeypatch/llama_attn_hijack_flash.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/src/axolotl/monkeypatch/llama_attn_hijack_flash.py b/src/axolotl/monkeypatch/llama_attn_hijack_flash.py
index 33de909719..b0163a6556 100644
--- a/src/axolotl/monkeypatch/llama_attn_hijack_flash.py
+++ b/src/axolotl/monkeypatch/llama_attn_hijack_flash.py
@@ -57,7 +57,9 @@ def replace_llama_attn_with_flash_attn(packed: Optional[bool] = False):
             CrossEntropyLoss, inplace_backward=True
         )
     except ImportError:
-        pass
+        LOG.info(
+            "optimized flash-attention CrossEntropyLoss not found (run `pip install git+https://github.com/Dao-AILab/flash-attention.git#egg=xentropy_cuda_lib&subdirectory=csrc/xentropy`)"
+        )
 
 
 # Disable the transformation of the attention mask in LlamaModel as the flash attention