app.py

import gradio as gr
from transformers import AutoTokenizer,AutoModelForCasualLM
def haiku_generate():
    model_id = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
    tokenizer = AutoTokenizer.from_pretrained(model_id)
    model = AutoModelForCasualLM.from_pretrained{
        model_id,
        device_map="auto",
        low_cpu_mem_usage=True
    }
    model.eval()

    prompt ="俳句だけを生成して"
    inputs = tokenizer(prompt,return_tensors="pt")
    if not torch.cuda.is_available():
        inputs ={k:v.to("cpu") for k,v in inputs.items()}
    with torch.no_glad():
        outputs = model.generate{
            **inputs,
            max_new_tokens=25,
            do_sample=True,
            temperature=0.7,
            top_p=0.9
        }
    return tokenizer.decode(outputs[0],skip_special_tokens=True)
def greet(name):
    return "Hello " + name + "!!"

demo = gr.Interface(fn=haiku_generate, outputs="text")
demo.launch()