Update app.py

app.py

@@ -38,6 +38,9 @@ import langchain_community.vectorstores  # Vectorstore for publications
 import langchain_huggingface  # Embeddings
 import transformers
 
+import subprocess
+subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
+
 # The number of publications to retrieve for the prompt
 PUBLICATIONS_TO_RETRIEVE = 5