flash attn installation in runtime

app.py

@@ -1,3 +1,11 @@
+import subprocess
+
+subprocess.run(
+    "pip install flash-attn --no-build-isolation",
+    env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
+    shell=True,
+)
+
 import os
 from time import time_ns
 

requirements.txt

@@ -4,5 +4,4 @@ llama-cpp-python
 gradio
 requests
 sentencepiece
-spaces
-https://github.com/tridao/flash-attention-wheels/releases/download/v2.3.5.post7/flash_attn_wheels_test-2.3.5.post7+cu122torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl
+spaces