|
import gradio as gr |
|
import torch |
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
tokenizer = AutoTokenizer.from_pretrained("Salesforce/xgen-7b-8k-base", trust_remote_code=True) |
|
model = AutoModelForCausalLM.from_pretrained("Salesforce/xgen-7b-8k-base", torch_dtype=torch.bfloat16) |
|
|
|
|
|
|
|
def gentext(user_input="The world is"): |
|
inputs = tokenizer(user_input, return_tensors="pt") |
|
sample = model.generate(**inputs, max_length=128) |
|
|
|
return {"output": tokenizer.decode(sample[0])} |
|
|
|
|
|
gr.Interface( |
|
gentext, |
|
inputs="text", |
|
outputs="text", |
|
title="Testing out salesforce XGen 7B", |
|
).launch() |