Update app.py

app.py

@@ -1,5 +1,6 @@
 import os
-from typing import Dict, List, Union
+from typing import List, Tuple, Dict, Union, Any
+import requests
 
 import numpy as np
 import openai
@@ -58,6 +59,65 @@ def ai_judge(prompt: str) -> float:
     return call_chatgpt(prompt)
 
 
+def query(payload: Dict[str, Any]) -> Dict[str, Any]:
+    """
+    Sends a JSON payload to a predefined API URL and returns the JSON response.
+    Args:
+        payload (Dict[str, Any]): The JSON payload to be sent to the API.
+    Returns:
+        Dict[str, Any]: The JSON response received from the API.
+    """
+
+    # API endpoint URL
+    API_URL = "https://sks7h7h5qkhoxwxo.us-east-1.aws.endpoints.huggingface.cloud"
+    
+    # Headers to indicate both the request and response formats are JSON
+    headers = {
+        "Accept": "application/json",
+        "Content-Type": "application/json"
+    }
+    
+    # Sending a POST request with the JSON payload and headers
+    response = requests.post(API_URL, headers=headers, json=payload)
+    
+    # Returning the JSON response
+    return response.json()
+
+
+def llama2_7b_ysa(prompt: str) -> str:
+    """
+    Queries a model and retrieves the generated text based on the given prompt.
+    This function sends a prompt to a model (presumably named 'llama2_7b') and extracts
+    the generated text from the model's response. It's tailored for handling responses
+    from a specific API or model query structure where the response is expected to be
+    a list of dictionaries, with at least one dictionary containing a key 'generated_text'.
+    Parameters:
+    - prompt (str): The text prompt to send to the model.
+    Returns:
+    - str: The generated text response from the model.
+    Note:
+    - The function assumes that the 'query' function is previously defined and accessible
+      within the same scope or module. It should send a request to the model and return
+      the response in a structured format.
+    - The 'parameters' dictionary is passed empty but can be customized to include specific
+      request parameters as needed by the model API.
+    """
+
+    # Define the query payload with the prompt and any additional parameters
+    query_payload: Dict[str, Any] = {
+        "inputs": prompt,
+        "parameters": {}
+    }
+
+    # Send the query to the model and store the output response
+    output = query(query_payload)
+
+    # Extract the 'generated_text' from the first item in the response list
+    response: str = output[0]['generated_text']
+
+    return response
+
+
 ## rag strategy 1
 # file_names = [f"output_files/file_{i}.txt" for i in range(131)]
 # # file_names = [f"output_files_large/file_{i}.txt" for i in range(1310)]
@@ -161,7 +221,7 @@ if prompt := st.chat_input("Tell me about YSA"):
         ref = pd.DataFrame(
             {
                 "idx": idx,
-                "question": [dataset["train"]['questions'][i] for i in idx],
+                "questions": [dataset["train"]['questions'][i] for i in idx],
                 "answers": [dataset["train"]['answers'][i] for i in idx],
                 "distances": results["distances"][0]
             }
@@ -177,6 +237,20 @@ if prompt := st.chat_input("Tell me about YSA"):
             ref_from_db_search = ref["answers"]
             final_ref = ref
 
+        try:
+            llm_response = llama2_7b_brk_letters(question)
+        except:
+            llm_response = "Sorry, the inference endpoint is temporarily down. 😔"
+
+        finetuned_llm_guess = {
+            "idx": "from_llm",
+            "questions": question,
+            "answers": llm_response,
+            "distances": 0
+        }
+        final_ref = final_ref.append(finetuned_llm_guess)
+
+        # add ai judge as additional rating
         independent_ai_judge_score = []
         for i in range(final_ref.shape[0]):
             this_quest = question
@@ -188,7 +262,7 @@ if prompt := st.chat_input("Tell me about YSA"):
 
                 From 0 to 10, rate how well the content address the user's question. 
 
-                Only produce a number from 0 to 10. 
+                Only produce a number from 0 to 10 while 10 being the best at addressing user's question.
             """
             this_score = ai_judge(prompt_for_ai_judge)
             independent_ai_judge_score.append(this_score)