add install.sh

app.py

@@ -29,7 +29,7 @@ def install_vllm_from_patch():
         return False
 
 #install vllm from patch file
-install_vllm_from_patch()
+#install_vllm_from_patch()
 
 
 # load model and tokenizer
@@ -76,7 +76,11 @@ def respond(
 
     model_inputs = tokenizer([text], return_tensors="pt", return_token_type_ids=False).to(model.device)
 
-    model_inputs.update(dict(max_new_tokens=512,streamer=streamer))
+    model_inputs.update(dict(max_new_tokens=max_tokens,
+                             temperature = temperature,
+                             top_p = top_p,
+                             presence_penalty = 1.5,
+                             streamer=streamer))
 
     # Start a separate thread for model generation to allow streaming output
     thread = Thread(