app.py

import gradio as gr
import time
import ctypes #to run on C api directly 
import llama_cpp
from llama_cpp import Llama
from huggingface_hub import hf_hub_download #load from huggingfaces 


llm = Llama(model_path= hf_hub_download(repo_id="TheBloke/StableBeluga-7B-GGML", filename="stablebeluga-7b.ggmlv3.q6_K.bin"))

def generate_text(input_text):
    output = llm(f"Q: {input_text}  A:", max_tokens=521, stop=["Q:", "\n"], echo=True)
    return output['choices'][0]['text']

input_text = gr.inputs.Textbox(lines= 10, label="Enter your input text")
output_text = gr.outputs.Textbox(label="Output text")

description = "bro neil it currently dosent work two people sending it request at the same time so going to fix that but currently running ggml models with llama.cpp implementation in python [https://github.com/abetlen/llama-cpp-python]"

examples = [
    ["What is the capital of France? ", "The capital of France is Paris."],
    ["Who wrote the novel 'Pride and Prejudice'?", "The novel 'Pride and Prejudice' was written by Jane Austen."],
    ["What is the square root of 64?", "The square root of 64 is 8."]
]


demo = gr.Interface(fn=generate_text, inputs=input_text, outputs=output_text, title="Llama Language Model", description=description, examples=examples).queue
demo.queue(concurrency_count=3)
demo.launch()