query langchain

app.py

@@ -1,15 +1,41 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+
+from query_data import query_data
+from create_database import split_text
+import os
+import shutil 
+
+
 import logging
 
+logging.basicConfig(filename='myapp.log',format='%(asctime)s %(message)s', datefmt='%m/%d/%Y %I:%M:%S %p')
 logger = logging.getLogger(__name__)
 
 """
 For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
 """
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
-logger.info("### Inference client: zephyr-7b-beta")
+
+CHROMA_PATH = "chroma"
+DATA_PATH = "./data"
+
+
+accesstoken = os.environ['HF_TOKEN']
+checkpoint = "HuggingFaceH4/zephyr-7b-beta"
+client = InferenceClient(checkpoint,token = accesstoken)
+
+def upload_file(file):
+    if not os.path.exists(DATA_PATH):
+        os.mkdir(DATA_PATH)
+
+    shutil.copy(file,DATA_PATH)
+    gr.Info("File uploading")
+
+
+logger.info("### Inference client: "+checkpoint)
+
+
 
 def respond(
     message,
@@ -30,28 +56,22 @@ def respond(
     messages.append({"role": "user", "content": message})
 
     logger.info(messages)
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
+    response = query_data(message)
+    yield response
 
 
 """
 For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
 """
-demo = gr.ChatInterface(
+with gr.Blocks() as demo: 
+
+    upload_button = gr.UploadButton("Click the button to upload")
+    upload_button.upload(upload_file,upload_button)
+
+    gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+        gr.Textbox(value="You are a friendly Chatbot that helps searching knowledge into scientific articles.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
@@ -60,8 +80,8 @@ demo = gr.ChatInterface(
             value=0.95,
             step=0.05,
             label="Top-p (nucleus sampling)",
-        ),
-    ],
+        )
+                ],
 )
 
 

query_data.py

@@ -0,0 +1,55 @@
+import argparse
+# from dataclasses import dataclass
+from langchain_chroma import Chroma
+from langchain_huggingface import HuggingFaceEmbeddings
+from langchain_huggingface import HuggingFaceEndpoint
+
+from langchain.prompts import ChatPromptTemplate
+
+from langchain.chains import LLMChain
+from langchain_core.prompts import PromptTemplate
+import os
+
+CHROMA_PATH = "chroma"
+
+PROMPT_TEMPLATE = """
+Answer the question based only on the following context:
+
+{context}
+
+---
+
+Answer the question based on the above context: {question}
+"""
+
+
+def query_data(query_text):
+
+    # Prepare the DB.
+    embedding_function = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
+    db = Chroma(persist_directory=CHROMA_PATH, embedding_function=embedding_function)
+
+    # Search the DB.
+    results = db.similarity_search_with_relevance_scores(query_text, k=3)
+    if len(results) == 0 or results[0][1] < 0.2:
+        print(f"Unable to find matching results.")
+        return
+
+    context_text = "\n\n---\n\n".join([doc.page_content for doc, _score in results])
+    prompt_template = ChatPromptTemplate.from_template(PROMPT_TEMPLATE)
+
+    repo_id = "HuggingFaceH4/zephyr-7b-beta"
+
+    llm = HuggingFaceEndpoint(
+        repo_id=repo_id,
+        max_length = 512,
+        temperature=0.5,
+        huggingfacehub_api_token=os.environ['HF_TOKEN'],
+    )
+    llm_chain = prompt_template | llm
+
+    response_text = llm_chain.invoke({"question": query_text, "context":context_text})
+    
+    sources = [doc.metadata.get("source", None) for doc, _score in results]
+    formatted_response = f"{response_text}\nSources: {sources}"
+    return formatted_response

requirements.txt

@@ -1 +1,6 @@
-huggingface_hub==0.25.2
+huggingface_hub==0.25.2
+tiktoken 
+langchain 
+langchain-community
+langchain_chroma
+langchain_huggingface