Problems with hub.load

TensorFlow General Discussion
,
1

I’m using a hosted Flask service at A2 hosting and work from their CPanel. I have installed
tensorboard==2.16.2
tensorboard-data-server==0.7.2
tensorflow==2.16.1
tensorflow-hub==0.16.1
tensorflow-io-gcs-filesystem==0.37.0
tensorflow-text==2.16.1

In my py page

from flask import Flask, request, jsonify, Blueprint
import requests
import tensorflow as tf
import tensorflow_hub as hub
import os

tensorflowapi_bp = Blueprint(‘tensorflowapi’, name)

Load the Universal Sentence Encoder model

#use_model = hub.load(“@tensorflow-models/universal-sentence-encoder”)

#use_model = hub.load(“Google | universal-sentence-encoder | Kaggle”)

model_url = “Google | universal-sentence-encoder | Kaggle”
use_model = hub.load(model_url)

As soon as I try to hub.load I get a 500 error for the entire application. As soon as I comment it out the app runs fine… I have tried many different url’s but it seems that the hub.load is the issue. Is this normal? Thank for the help

3

Hi William

I’ve just tested this on colab and it worked fine:

import tensorflow as tf
import tensorflow_hub as hub

model_url = "https://www.kaggle.com/models/google/universal-sentence-encoder/tensorFlow2/universal-sentence-encoder/2?tfhub-redirect=true"
use_model = hub.load(model_url)

I’m using TF version 2.15, can you try with that?

and maybe there are some information here that might help you: TensorFlow Hub Moving to Kaggle Models - FAQs

4

I’m new to Python so be easy. I set the TF version to 2.15 in the requirements.txt page in my Flask app and ran it and got the following error:

ERROR: Cannot install -r requirements.txt (line 87) and wrapt==1.16.0 because these package versions have conflicting dependencies.

The conflict is caused by:
The user requested wrapt==1.16.0
tensorflow 2.15.1 depends on wrapt<1.15 and >=1.11.0

To fix this you could try to:

  1. loosen the range of package versions you’ve specified
  2. remove package versions to allow pip attempt to solve the dependency conflict

ERROR: ResolutionImpossible: for help visit Dependency Resolution - pip documentation v24.2.dev0
((api:3.9)) [pythonap@az1-ts114 api]$

Could this be my trouble?

5

I’m going around in circles-- The conflict is caused by:
The user requested tensorflow==2.15.1
tensorflow-text 2.15.0 depends on tensorflow<2.16 and >=2.15.0; platform_machine != “arm64” or platform_system != “Darwin”
tf-keras 2.16.0 depends on tensorflow<2.17 and >=2.16