Update app.py

app.py

@@ -19,18 +19,18 @@ from llama_cpp_agent.chat_history.messages import Roles
 # subprocess.run('pip install llama-cpp-agent==0.2.10', shell=True)
 
 
-# hf_hub_download(
-#     repo_id="QuantFactory/Meta-Llama-3-8B-Instruct-GGUF",
-#     filename="Meta-Llama-3-8B-Instruct.Q8_0.gguf",
-#     local_dir = "./models"
-# )
-
 hf_hub_download(
-    repo_id="bartowski/Meta-Llama-3-70B-Instruct-GGUF",
-    filename="Meta-Llama-3-70B-Instruct-Q3_K_M.gguf",
+    repo_id="QuantFactory/Meta-Llama-3-8B-Instruct-GGUF",
+    filename="Meta-Llama-3-8B-Instruct.Q8_0.gguf",
     local_dir = "./models"
 )
 
+# hf_hub_download(
+#     repo_id="bartowski/Meta-Llama-3-70B-Instruct-GGUF",
+#     filename="Meta-Llama-3-70B-Instruct-Q3_K_M.gguf",
+#     local_dir = "./models"
+# )
+
 def process_document(pdf_path, page_ids=None):
     extracted_pages = extract_pages(pdf_path, page_numbers=page_ids)
 
@@ -87,7 +87,7 @@ def deidentify_doc(pdftext, maxtokens, temperature, top_probability):
     prompt = "In the following text replace any person name and any address with term [redacted], replace any Date of Birth and NHS number with term [redacted]. Output the modified text."
 
     llm = Llama(
-        model_path="models/Meta-Llama-3-70B-Instruct-Q3_K_M.gguf",
+        model_path="models/Meta-Llama-3-8B-Instruct.Q8_0.gguf",
         flash_attn=True,
         n_gpu_layers=81,
         n_batch=1024,