app.py

import gradio as gr
from fastapi import FastAPI, Request
import uvicorn
from sentence_transformers import SentenceTransformer
from sentence_transformers.util import cos_sim
from sentence_transformers.quantization import quantize_embeddings


import spaces



app = FastAPI()


@spaces.GPU
def embed(text):
        
    query_embedding = Embedder.encode(text)
    return query_embedding.tolist();
    
    
    
@app.post("/v1/embeddings")
async def openai_embeddings(request: Request):
    body = await request.json();
    print(body);
    
    model = body['model']
    text = body['input'];
    embeddings = embed(text)
    return {
		'object': "list"
		,'data': [{
			'object': "embeddings"
			,'embedding': embeddings
			,'index':0
		}]
		,'model':model
		,'usage':{
			 'prompt_tokens': 0
			,'total_tokens': 0
		}
	}

def fn(text):
    embed(text);

with gr.Blocks(fill_height=True) as demo:
    text = gr.Textbox();
    embeddings = gr.Textbox()
    
    text.submit(fn, [text], [embeddings]);
    

print("Loading embedding model");
Embedder = SentenceTransformer("mixedbread-ai/mxbai-embed-large-v1")

# demo.run_startup_events() 
 
 
#demo.launch(
#    share=False,
#    debug=False,
#    server_port=7860,
#    server_name="0.0.0.0",
#    allowed_paths=[]
#)

print("Demo run...");
(app,url,other) = demo.launch(prevent_thread_lock=True);

print("Mounting app...");  
GradioApp = gr.mount_gradio_app(app, demo, path="/", ssr_mode=False); 

if __name__ == '__main__':
    print("Running uviconr...");
    uvicorn.run(GradioApp)