added mistralAI

app.py

@@ -1,21 +1,58 @@
 import streamlit as st
 from utils import generate_text_embeddings
+from mistral7b import mistral
+
+import time
+
 
-st.title("Echo Bot")
 
 if "messages" not in st.session_state:
     st.session_state.messages = []
-    
 
-with st.sidebar :
-    st.markdown("# Inference Analytics")
+if "tokens_used"  not in st.session_state :
+    st.session_state.tokens_used = 0
+
+if "inference_time" not in st.session_state :
+    st.session_state.inference_time = [0.00]
+
+if "model_settings" not in st.session_state :
+    st.session_state.model_settings = {
+        "temp" : 0.9,
+        "max_tokens" : 512,
+    }
+
+if "history" not in st.session_state :
+    st.session_state.history = []
+
+if "top_k" not in st.session_state :
+    st.session_state.top_k = 5
+
+with st.sidebar:
+    st.markdown("# Model Analytics")
+    st.write("Tokens used :", st.session_state['tokens_used'])
+
+    st.write("Average Inference Time: ", round(sum(st.session_state["inference_time"]) / len(st.session_state["inference_time"]), 3)) 
+    st.write("Cost Incured :",round( 0.033 * st.session_state['tokens_used']/ 1000, 3), "INR")
+
     st.markdown("---")
-    st.markdown("Tokens used :")
-    st.markdown("Average Querying Time :")
-    st.markdown("Average Inference Time :")
-    st.markdown("Cost Incurred :")
+    st.markdown("# Retrieval Settings")
+    st.slider(label="Documents to retrieve", min_value=1, max_value=10, value=3)
+    st.markdown("---")
+    st.markdown("# Model Settings")
+    selected_model = st.sidebar.radio('Select one:', ["Mistral 7B", "GPT 3.5 Turbo", "GPT 4",  "Llama 7B"])
+    selected_temperature = st.slider(label="Temperature", min_value=0.0, max_value=1.0, step=0.1, value=0.5)
+    st.write(" ")
+    st.info("**2023 ©️ Pragnesh Barik**")
+
+
 
+st.image("ikigai.svg")
+st.title("Ikigai Chat")
 
+with st.expander("What is Ikigai Chat ?"):
+    st.info("""Ikigai Chat is a vector database powered chat agent, it works on the principle of 
+                of Retrieval Augmented Generation (RAG), Its primary function revolves around maintaining an extensive repository of Ikigai Docs and providing users with answers that align with their queries. 
+                This approach ensures a more refined and tailored response to user inquiries.""")
 
 
 for message in st.session_state.messages:
@@ -23,12 +60,24 @@ for message in st.session_state.messages:
         st.markdown(message["content"])
 
 
-if prompt := st.chat_input("What is up?"):
-    query_embeddings = generate_text_embeddings(prompt)
+if prompt := st.chat_input("Chat with Ikigai Docs?"):
     st.chat_message("user").markdown(prompt)
     st.session_state.messages.append({"role": "user", "content": prompt})
+    
+    tick = time.time()
+    response = mistral(prompt, st.session_state.history, temperature=st.session_state.model_settings["temp"] , max_new_tokens=st.session_state.model_settings["max_tokens"])
+    tock = time.time()
 
-    response = f"Echo: {prompt}"
-    with st.chat_message("assistant"):
+    
+    st.session_state.inference_time.append(tock - tick)
+    response = response.replace("</s>", "")
+    len_response = len(response.split())
+
+    st.session_state["tokens_used"] =  len_response + st.session_state["tokens_used"] 
+    
+    with st.chat_message("assistant"): 
         st.markdown(response)
-    st.session_state.messages.append({"role": "assistant", "content": response})
+    st.session_state.history.append([prompt, response])
+    st.session_state.messages.append({"role": "assistant", "content": response})
+    
+

mistral7b.py

@@ -0,0 +1,47 @@
+from huggingface_hub import InferenceClient
+import os
+from dotenv import load_dotenv
+load_dotenv()
+
+API_TOKEN = os.getenv('HF_TOKEN')
+client = InferenceClient(
+    "mistralai/Mistral-7B-Instruct-v0.1",
+    token=API_TOKEN
+)
+
+
+def format_prompt(message, history):
+  prompt = "<s>"
+  for user_prompt, bot_response in history:
+    prompt += f"[INST] {user_prompt} [/INST]"
+    prompt += f" {bot_response}</s> "
+  prompt += f"[INST] {message} [/INST]"
+  return prompt
+
+def mistral(
+    prompt, history, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0,
+):
+    temperature = float(temperature)
+    if temperature < 1e-2:
+        temperature = 1e-2
+    top_p = float(top_p)
+
+    generate_kwargs = dict(
+        temperature=temperature,
+        max_new_tokens=max_new_tokens,
+        top_p=top_p,
+        repetition_penalty=repetition_penalty,
+        do_sample=True,
+        seed=42,
+    )
+
+    formatted_prompt = format_prompt(prompt, history)
+
+    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
+    output = ""
+
+    for response in stream:
+        # print(response)
+        output += response.token["text"]
+        # yield output
+    return output

requirements.txt

@@ -4,6 +4,7 @@ av==10.0.0
 bitarray==2.8.1
 blinker==1.6.3
 cachetools==5.3.1
+huggingface-hub==0.16.4
 certifi==2023.7.22
 charset-normalizer==3.2.0
 click==8.1.7
@@ -53,7 +54,7 @@ PyPDF2==3.0.1
 pyreadline3==3.4.1
 python-dotenv==1.0.0
 pytz==2023.3.post1
-PyYAML==6.0.1
+PyYAML==6.0.1git
 readme-renderer==42.0
 referencing==0.30.2
 regex==2023.8.8

utils.py

@@ -1,5 +1,3 @@
-import json
-import requests
 import os
 from dotenv import load_dotenv
 from sentence_transformers import SentenceTransformer