app.py

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "inclusionAI/Ring-mini-2.0"
print(f"load model {model_name}")
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto",
    trust_remote_code=True
)

print(f"load tokenizer {model_name}")
tokenizer = AutoTokenizer.from_pretrained(model_name)


def text_gen(msg):
    
    prompt = "Give me a short introduction to large language models."
    messages = [
    {"role": "system", "content": "You are Ring, an assistant created by inclusionAI"},
    {"role": "user", "content": prompt}
    ]
    text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=True
    )
    model_inputs = tokenizer([text], return_tensors="pt", return_token_type_ids=False).to(model.device)

    generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=8192
    )
    generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
    ]

    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
    return response

    
from flask import Flask  
from flask import request

app = Flask(__name__)  
print(f"Flask app")
# Главная страница
@app.route("/")
def home():
    return f"<h1>Главная страница</h1><p>Добро пожаловать!</p><p>current model {model_name}</p><p>/about /contact /gen?msg</p>"

# Страница "О нас"
@app.route("/about")
def about():
    return "<h1>О нас</h1><p>Мы изучаем Flask!</p>"

# Страница "Контакты"
@app.route("/contact")
def contact():
    return "<h1>Контакты</h1><p>Свяжитесь с нами: email@example.com</p>"

# Страница "пут"
@app.route("/gen", methods=['POST', 'GET'])
def gen_msg():
    print('gen')
    answer="<p>{answer}</p>"
    if request.args.get('msg'):
        answer = text_gen(request.args['msg'])
        
    return f"<p>{answer}</p>"

if __name__ == "__main__":
    app.run(debug=False, host='0.0.0.0', port=7860)