services/llm.py

from langchain_ollama import OllamaLLM
from llama_cpp import Llama
from langchain_huggingface import HuggingFaceEmbeddings
import streamlit as st


@st.cache_resource
def initialize_llm(model_name, temperature, top_p, max_tokens):
    # # Configure the LLM with additional parameters
    # llm = OllamaLLM(
    #     model=model_name,
    #     base_url="https://deepak7376-ollama-server.hf.space",
    #     temperature=temperature,   # Controls randomness (0 = deterministic, 1 = max randomness)
    #     max_tokens=max_tokens,   # Limit the number of tokens in the output
    #     top_p=top_p          # Nucleus sampling for controlling diversity
    # )
    llm = Llama.from_pretrained(
	repo_id="bartowski/DeepSeek-R1-Distill-Qwen-1.5B-GGUF",
	filename="DeepSeek-R1-Distill-Qwen-1.5B-IQ4_XS.gguf",
    n_ctx=max_tokens
    
    )
    return llm

@st.cache_resource
def initialize_embeddings():
    embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
    return embeddings