使用聊天机器人通用句子编码器时出错

Question

我正在尝试“chatbot.py”，但出现以下错误。

Traceback（最近一次调用最后一次）： 文件“chatbot.py”，第 11 行，在 嵌入 = 模型（df[“MESSAGE”].values）[“输出”] IndexError：只有整数、切片 (:)、省略号 (...)、numpy.newaxis (None) 和整数或布尔数组是有效的索引

###########################
# chatbot.py starts here
from urllib import response
from utils import *
# pip install pandas
import pandas as pd
import numpy as np
# pip install openpyxl

model = embed_useT(r"C:/Users/Hp/Downloads/universal-sentence-encoder_4")
df = pd.read_excel("chats.xlsx")
print ('printing df', df)
embeddings = model(df["MESSAGE"].values)["outputs"]
norm = np.linalg.norm(embeddings, axis = -1)

def reply(message):
    message_vector = model([message])["outputs"]
    similarities = cos_similarity(message_vector, embeddings, norm)
    index = np.argmax(similarities)
    response = df["RESPONSE"].values[index]
    return response

while True:
    message = input("Type your message: ")
    response = reply(message)
    print (response)
# chatbot.py ends here
###########################

###########################
# utils.py starts here
import numpy as np
# pip install tensorflow-hub
import tensorflow_hub as hub 
# pip install tensorflow

# import tensorflow as tf
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()

def cos_similarity(vector, matrix, matrix_norm):
    dot = np.matmul(matrix, vector.T)
    vector_norm = np.linalg.norm(vector)
    norms = (vector_norm * matrix_norm); norms=norms.reshape(norms.shape[0],1)
    return dot / norms

def embed_useT(module):
    with tf.Graph().as_default():
        sentences = tf.placeholder(tf.string)
        embed = hub.KerasLayer(module)
        embeddings = embed(sentences)
        session = tf.train.MonitoredSession()
        return lambda x: session.run(embeddings, {sentences : x})
# utils.py ends here
###########################

chats.xlsx如下图

Answer 1

model(df["MESSAGE"].values) 已经将嵌入作为 numpy 数组返回。省略 ["outputs"] 应该可以正常工作。