import os import gradio as gr import pandas as pd import numpy as np import chromadb from chromadb.config import Settings from io import StringIO from sentence_transformers import SentenceTransformer import openai import plotly.express as px from sklearn.manifold import TSNE # Initialize Chroma client with DuckDB and Parquet for persistence chroma_client = chromadb.Client(Settings( chroma_db_impl="duckdb+parquet", persist_directory="./chroma_db" )) # Model Configuration for Dynamic Dropdown model_config = { "gpt-4": { "endpoint": "https://roger-m38jr9pd-eastus2.openai.azure.com/openai/deployments/gpt-4/chat/completions?api-version=2024-08-01-preview", "api_key": os.getenv("GPT4_API_KEY") }, "gpt-4o": { "endpoint": "https://roger-m38jr9pd-eastus2.openai.azure.com/openai/deployments/gpt-4o/chat/completions?api-version=2024-08-01-preview", "api_key": os.getenv("GPT4O_API_KEY") }, "gpt-35-turbo": { "endpoint": "https://rogerkoranteng.openai.azure.com/openai/deployments/gpt-35-turbo/chat/completions?api-version=2024-08-01-preview", "api_key": os.getenv("GPT35_TURBO_API_KEY") }, "gpt-4-32k": { "endpoint": "https://roger-m38orjxq-australiaeast.openai.azure.com/openai/deployments/gpt-4-32k/chat/completions?api-version=2024-08-01-preview", "api_key": os.getenv("GPT4_32K_API_KEY") } } # Function to process uploaded CSV def process_csv_text(temp_file): if isinstance(temp_file, str): df = pd.read_csv(StringIO(temp_file)) else: df = pd.read_csv(temp_file.name, header='infer', sep=',') return df, gr.Dropdown.update(choices=list(df.columns)) # Insert or update ChromaDB with embeddings def insert_or_update_chroma(col, table, model_name, similarity_metric, client=chroma_client): try: collection = client.create_collection(name="my_collection", embedding_function=SentenceTransformer(model_name), metadata={"hnsw:space": similarity_metric}) except Exception as e: print("Collection exists, deleting it") client.delete_collection(name='my_collection') collection = client.create_collection(name="my_collection", embedding_function=SentenceTransformer(model_name), metadata={"hnsw:space": similarity_metric}) if collection: try: collection.add( documents=list(table[col]), metadatas=[{"source": i} for i in range(len(table))], ids=[str(i + 1) for i in range(len(table))] ) return "Embedding calculations and insertions successful" except Exception as e: return "Error in embedding calculations" # Show plot with embeddings using t-SNE def show_fig(): collection = chroma_client.get_collection(name="my_collection") embeddings = collection.get(include=['embeddings', 'documents']) df = pd.DataFrame({ 'text': embeddings['documents'], 'embedding': embeddings['embeddings'] }) embeddings_np = np.array(df['embedding'].tolist()) tsne = TSNE(n_components=2, random_state=42) transformed = tsne.fit_transform(embeddings_np) df['tsne_x'] = transformed[:, 0] df['tsne_y'] = transformed[:, 1] fig = px.scatter(df, x='tsne_x', y='tsne_y', hover_name='text') return fig, transformed # Show test string figure def show_test_string_fig(test_string, tsne, model_name, similarity_metric): collection = chroma_client.get_collection(name="my_collection", embedding_function=SentenceTransformer(model_name)) collection.add( documents=[test_string], metadatas=[{"source": 'test'}], ids=['test_sample'] ) embeddings = collection.get(include=['embeddings', 'documents']) df = pd.DataFrame({ 'text': embeddings['documents'], 'embedding': embeddings['embeddings'], 'set': ['orig' if document != test_string else 'test_string' for document in embeddings["documents"]] }) embeddings_np = np.array(df['embedding'].tolist()) transformed = tsne.transform(embeddings_np) df['tsne_x'] = transformed[:, 0] df['tsne_y'] = transformed[:, 1] fig = px.scatter(df, x='tsne_x', y='tsne_y', hover_name='text', color='set') return fig, tsne # Function to interact with OpenAI's Azure API def ask_gpt(message, messages_history, embedding_model, system_prompt, temperature, max_tokens, chatgpt_model): if len(messages_history) < 1: messages_history = [{"role": "system", "content": system_prompt}] model_info = model_config[chatgpt_model] headers = { "Content-Type": "application/json", "api-key": model_info["api_key"] } message = retrieve_similar(message, embedding_model) messages_history += [{"role": "user", "content": message}] response = openai.ChatCompletion.create( model=chatgpt_model, messages=messages_history, temperature=temperature, max_tokens=max_tokens ) return response['choices'][0]['message']['content'], messages_history # Function to retrieve similar questions from ChromaDB def retrieve_similar(prompt, embedding_model, client=chroma_client): collection = client.get_collection(name="my_collection", embedding_function=SentenceTransformer(model_name=embedding_model)) results = collection.query(query_texts=prompt, n_results=10) additional_context = '' for i, document in enumerate(results['documents'][0]): if i == 0: additional_context = 'Information: \n' + str(i+1) + '. ' + document else: additional_context += '\n' + str(i+1) + '. ' + document prompt_with_context = additional_context + '\nQuestion: ' + prompt return prompt_with_context # Gradio App Setup with gr.Blocks() as demo: # Tab 1: Upload CSV and Display Data with gr.Tab("Upload data"): upload_button = gr.UploadButton(label="Upload csv", file_types=['.csv'], file_count="single") table = gr.Dataframe(type="pandas", max_rows='20', overflow_row_behaviour='paginate', interactive=True) cols = gr.Dropdown(choices=[], label='Dataframe columns') upload_button.upload(fn=process_csv_text, inputs=upload_button, outputs=[table, cols], api_name="upload_csv") # Tab 2: ChromaDB, Embeddings, and Plotting with gr.Tab("Select Column and insert embeddings to ChromaDb"): with gr.Row(): gr.Markdown("
") with gr.Row(): cols = gr.Dropdown(choices=['text_column_1_placeholder'], label='Dataframe columns') with gr.Row(): embedding_model = gr.Dropdown(value='all-MiniLM-L6-v2', choices=['all-MiniLM-L6-v2', 'intfloat/e5-small-v2', 'intfloat/e5-base-v2', 'intfloat/e5-large-v2','paraphrase-multilingual-MiniLM-L12-v2'], label='Embedding model to use') similarity_metric = gr.Dropdown(value='cosine', choices=['cosine', 'l2'], label='Similarity metric to use') with gr.Row(): embedding_button = gr.Button(value="Insert or update rows from selected column to embeddings db") text = gr.Textbox(label='Process status for Chroma', placeholder='This will be updated once you click "Process status for Chroma"') with gr.Row(): show_embeddings_button = gr.Button(value="Calculate 2d values from embeddings and show scatter plot") embeddings_plot = gr.Plot() with gr.Row(): tsne = gr.State(value=None) test_string = gr.Textbox(label='test string to try to embed', value="Insert test string here") with gr.Row(): calculate_2d_repr_button = gr.Button(value="See where text string is in 2d") embeddings_plot_with_text_string = gr.Plot() embedding_button.click(insert_or_update_chroma, inputs=[cols, table, embedding_model, similarity_metric], outputs=[text]) show_embeddings_button.click(show_fig, inputs=[], outputs=[embeddings_plot, tsne]) calculate_2d_repr_button.click(show_test_string_fig, inputs=[test_string, tsne, embedding_model, similarity_metric], outputs=[embeddings_plot_with_text_string, tsne]) # Tab 3: Chat with GPT Models with gr.Tab("Chat"): system_prompt = gr.Textbox(value="You are a helpful assistant.", label="System Message") chatgpt_model = gr.Dropdown(value="gpt-4", choices=list(model_config.keys()), label="ChatGPT Model to Use") temperature = gr.Slider(minimum=0, maximum=2, step=0.1, value=0.7, label="Temperature") max_tokens = gr.Slider(minimum=50, maximum=2000, step=50, value=300, label="Max Tokens") chatbot = gr.Chatbot(label="ChatGPT Chat") clear_button = gr.Button("Clear Chat History") msg = gr.Textbox() msg_log = gr.Textbox("Message history will be visible here", label='Message history') msg.submit(ask_gpt, [msg, chatbot], [msg, chatbot]) chatbot.submit(ask_gpt, [chatbot, system_prompt, embedding_model, temperature, max_tokens, chatgpt_model], [chatbot, system_prompt]) clear_button.click(fn=lambda: None, inputs=None, outputs=[chatbot]) # Launch Gradio interface demo.launch()