使用 python 从 S3 存储桶下载后加载泡菜文件的问题

Question

I am trying to download my ML model as a pickle file from the S3 bucket and then load it to predict.

import boto3
import io
import pickle

REGION = 'us-east-1'
ACCESS_KEY_ID = ******
SECRET_ACCESS_KEY = ******

'BUCKET_NAME = 'aws-logs-914250087788-us-east-1'
KEY = input_name # file path in S3

s3c = boto3.client(
        's3',
        region_name = REGION,
        aws_access_key_id = ACCESS_KEY_ID,
        aws_secret_access_key = SECRET_ACCESS_KEY
    )
rfModel=s3c.download_file(BUCKET_NAME, model1, 'RF_model.pckl')

#load model and predict

lm = open('RF_model.pckl', 'rb')
clf=pickle.loads(lm)
rf_pred=clf.predict(x)
np.savetxt(output_name1, rf_pred, delimiter=',', fmt='%s')

But when I run my code, the following error occurs:

TypeError: a byte-like object is required, not _io.BufferedReader.

How can I fix this error?

Answer 1

I should replace this part:

lm = open('RF_model.pckl', 'rb')
clf=pickle.loads(lm)

with this:

with open('RF_model.pckl', 'rb') as file:
    clf = pickle.load(file)