app.py

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

model = AutoModelForCausalLM.from_pretrained(
    "Trelis/Llama-2-7b-chat-hf-sharded-bf16",
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
    low_cpu_mem_usage=True,
    device_map = {"": 0}
)
tokenizer = AutoTokenizer.from_pretrained("Trelis/Llama-2-7b-chat-hf-sharded-bf16", trust_remote_code=True,)


def generate_text(input_text):
    input_ids = tokenizer.encode(input_text, return_tensors="pt")
    attention_mask = torch.ones(input_ids.shape)

    output = model.generate(
        input_ids,
        attention_mask=attention_mask,
         max_new_tokens=400, top_p = 0.7, top_k = 50,
        do_sample=True,
        num_return_sequences=1,
        eos_token_id=tokenizer.eos_token_id,
    )

    output_text = tokenizer.decode(output[0], skip_special_tokens=True)
    print(output_text)

    # Remove Prompt Echo from Generated Text
    cleaned_output_text = output_text.replace(input_text, "")
    return cleaned_output_text


text_generation_interface = gr.Interface(
    fn=generate_text,
    inputs=[
        gr.inputs.Textbox(label="Input Text"),
    ],
    outputs=gr.inputs.Textbox(label="Generated Text"),
    title="Falcon-7B Instruct",
).launch()