app.py

# import packages
import gradio as gr
import copy
from llama_cpp import Llama
from huggingface_hub import hf_hub_download
import chromadb
from chromadb.utils.embedding_functions import OpenCLIPEmbeddingFunction
from chromadb.utils.data_loaders import ImageLoader
from chromadb.config import Settings
from datasets import load_dataset
import numpy as np
from tqdm import tqdm
import shutil
import os
from chromadb.utils import embedding_functions
import gradio as gr
from PIL import Image
import requests
from io import BytesIO
from transformers import pipeline
from bark import SAMPLE_RATE, generate_audio, preload_models

# Initialize the Llama model
llm = Llama(
    ## original model
    # model_path=hf_hub_download(
    #     repo_id="microsoft/Phi-3-mini-4k-instruct-gguf",
    #     filename="Phi-3-mini-4k-instruct-q4.gguf",
    # ),
    ## compressed model
    model_path=hf_hub_download(
        repo_id="TheBloke/CapybaraHermes-2.5-Mistral-7B-GGUF",
        filename="capybarahermes-2.5-mistral-7b.Q2_K.gguf",
    ),
    n_ctx=2048,
    n_gpu_layers=50,  # Adjust based on your VRAM
)

# use of clip model for embedding
client = chromadb.PersistentClient(path="DB")

embedding_function = OpenCLIPEmbeddingFunction()
image_loader = ImageLoader() # must be if you reads from URIs

# initialize separate collection for image and text data
collection_images = client.create_collection(
    name='collection_images',
    embedding_function=embedding_function,
    data_loader=image_loader)

collection_text = client.create_collection(
    name='collection_text',
    embedding_function=embedding_function,
    )

# Get the uris to the images
IMAGE_FOLDER = 'images'


image_uris = sorted([os.path.join(IMAGE_FOLDER, image_name) for image_name in os.listdir(IMAGE_FOLDER) if not image_name.endswith('.txt')])
ids = [str(i) for i in range(len(image_uris))]

collection_images.add(ids=ids, uris=image_uris)