improve handling and error if fa3 requested but not installeD

src/axolotl/utils/models.py

@@ -636,6 +636,14 @@ class ModelLoader:
                 if torch.cuda.get_device_capability() >= (9, 0):
                     # FA3 is only available on Hopper GPUs and newer
                     use_fa3 = True
+                if not importlib.util.find_spec("flash_attn_interface"):
+                    use_fa3 = False
+            if use_fa3 and not importlib.util.find_spec("flash_attn_interface"):
+                # this can happen when use_flash_attention_3 is explicity set to True
+                # and flash_attn_interface is not installed
+                raise ModuleNotFoundError(
+                    "Please install the flash_attn_interface library to use Flash Attention 3.x"
+                )
             if use_fa3 and importlib.util.find_spec("flash_attn_interface") is not None:
                 from flash_attn_interface import flash_attn_func as flash_attn_func_v3
                 from flash_attn_interface import (