File size: 820 Bytes
885747d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 |
import torch
import numpy as np
import gradio as gr
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("juliensimon/autonlp-imdb-demo-hf-16622775")
model = AutoModelForSequenceClassification.from_pretrained("juliensimon/autonlp-imdb-demo-hf-16622775")
def predict(review):
inputs = tokenizer(review, padding=True, truncation=True, return_tensors="pt")
outputs = model(**inputs)
predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)
predictions = predictions.detach().numpy()[0]
index = np.argmax(predictions)
score = predictions[index]
return "This revied os {:.3f}% {}".format(100*score, "negative" if index == 0 else "positive")
iface = gr.Interface(fn=predict, inputs='text', outputs='text')
iface.launch()
|