app.py

import spaces
import gradio as gr
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer


device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

def load_model():
    model_id = "stefan-it/nanochat-german-v1"

    tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=False)
    model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=False, dtype=torch.bfloat16).to(device)
    model.eval()

    return tokenizer, model

tokenizer, model = load_model()


@spaces.GPU
def generate(prompt, history, max_new_tokens, temperature, top_p, repetition_penalty, no_repeat_ngram_size):
    if len(history) > 0:
        messages = history + [
        {"role": "user", "content": prompt},
    ]
    else:
        messages = [
            {"role": "user", "content": prompt},
        ]

    print(history)
    inputs = tokenizer.apply_chat_template(
        messages,
        add_generation_prompt=True,
        tokenize=True,
        return_tensors="pt",
        return_dict=True,
    ).to(device)
    
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=max_new_tokens,
            temperature=temperature,
            top_p=top_p,
            repetition_penalty=repetition_penalty,
            no_repeat_ngram_size=no_repeat_ngram_size,
        )
    
    generated_tokens = outputs[0, inputs.input_ids.shape[1]:]
    output = tokenizer.decode(generated_tokens, skip_special_tokens=True)

    return output


demo = gr.ChatInterface(fn=generate,
                        type="messages",
                        title="German nanochat v1",
                        additional_inputs=[
                            gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
                            gr.Slider(minimum=0.1, maximum=4.0, value=0.8, step=0.1, label="Temperature"),
                            gr.Slider(minimum=0.1, maximum=1.0, value=0.9, step=0.05, label="Top-p"),
                            gr.Slider(minimum=1.0, maximum=2.0, value=1.2, step=0.1, label="Repetition penalty"),
                            gr.Slider(minimum=0, maximum=5, value=3, step=1, label="No repeat of ngrams"),
                        ])
demo.launch()