Update app.py

app.py

@@ -12,9 +12,8 @@ import os
 
 import torch
 from peft import PeftModel, PeftConfig
-from transformers import AutoModelForCausalLM,AutoTokenizer, LlamaForCausalLM
+from transformers import AutoModelForCausalLM,AutoTokenizer
 
-import webbrowser
 import joblib
 from deployML import predd
 
@@ -22,21 +21,21 @@ import time
 
 #load the chatbot
 
-model = LlamaForCausalLM.from_pretrained(
-    "medalpaca/medalpaca-7b",
-    return_dict=True,
-    load_in_8bit=True,
-    device_map="auto",
-)
+peft_model_id = "medalpaca/medalpaca-7b"
+
+config = PeftConfig.from_pretrained(peft_model_id)
 
-tokenizer = AutoTokenizer.from_pretrained("medalpaca/medalpaca-7b")
+model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, return_dict=True, load_in_8bit=True, device_map='auto')
+tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
+
+# Load the Lora model
+model = PeftModel.from_pretrained(model, peft_model_id)
 
 
 #load the first interface
 
 
 def fn(*args):
-  global flag
   global symptoms
   all_symptoms = [symptom for symptom_list in args for symptom in symptom_list]
 
@@ -46,14 +45,12 @@ def fn(*args):
     raise gr.Error("Please select at least 3 symptoms.")
 
   symptoms = all_symptoms  # Update global symptoms list
-  #webbrowser.open_new_tab(url)
-  flag=1
-  return symptoms
+  return predd(loaded_rf,symptoms)
+
 
 
 symptoms = []
 
-flag=0
 
 
 demo = gr.Interface(
@@ -72,20 +69,13 @@ demo = gr.Interface(
         gr.CheckboxGroup(['knee pain', 'hip joint pain', 'swelling joints'], label='Joint and Bone Issues'),
         gr.CheckboxGroup(['spinning movements', 'unsteadiness'], label='Neurological Movements')
     ],
-    outputs=None,
+    outputs="textbox",allow_flagging="never"
 )
 
 
 
 
 
-def wait_until_variable_changes( target_value):
-  global flag
-  while flag != target_value:
-      time.sleep(1)
-
-
-
 
 def predict(message, history):
   prompt = f"""
@@ -96,7 +86,7 @@ def predict(message, history):
   batch = tokenizer(prompt, return_tensors='pt')
   with torch.cuda.amp.autocast():
 
-    output_tokens = model.generate(**batch, max_new_tokens=60)
+    output_tokens = model.generate(**batch, max_new_tokens=100)
 
   return tokenizer.decode(output_tokens[0], skip_special_tokens=True).replace(prompt,"")
 
@@ -105,20 +95,10 @@ loaded_rf = joblib.load("model_joblib")
 Fmessage="hello im here to help you!"
 
 
-if __name__ == "__main__":
-    demo.launch()
-
-wait_until_variable_changes(1)
-
-
-
-if symptoms:
-  Fmessage=predd(loaded_rf,symptoms)
-
-
-chatbot=gr.ChatInterface(predict,chatbot=gr.Chatbot(value=[(None, Fmessage)],), clear_btn=None, retry_btn=None, undo_btn=None)
 
+chatbot=gr.ChatInterface(predict, clear_btn=None, retry_btn=None, undo_btn=None)
 
 
-demo.close()
-chatbot.launch()
+gr.TabbedInterface(
+    [demo, chatbot], ["symptoms checker", "chatbot"]
+).launch()