WSOD if external moderation is enabled, and "extra block" warnings in watchdog

Created on 14 October 2024, about 1 month ago

Problem/Motivation

Hi guys,

I enabled the external moderation module (I'm on yesterday's dev release) and I get a WSOD if I configure it and run something through OpenAI using automators.

Also the watchdog is flooding with 'The "extra_field_block:node:g_dispensary:content_moderation_control" was not found' warning messages.

Steps to reproduce

Proposed resolution

Remaining tasks

User interface changes

API changes

Data model changes

🐛 Bug report
Status

Active

Version

1.0

Component

Other Submodules

Created by

🇹🇭Thailand AlfTheCat

Live updates comments and jobs are added and updated live.
Sign in to follow issues

Comments & Activities

  • Issue created by @AlfTheCat
  • 🇩🇪Germany marcus_johansson

    Could you give back how the external moderation is setup on this page: /admin/config/ai/ai_external_moderation

  • 🇩🇪Germany marcus_johansson

    I'm not sure if the other issue is connected to the AI module, its the Content Moderation module from what I can see.

  • 🇹🇭Thailand AlfTheCat

    Hi Marcus, yes it could be the content moderation module that's causing the watchdog errors. I just didn't see them before but I will check the setup for that one.

    With regards to the settings form, the provider is set to OpenAI, empty tags field, and OpenAI - text-moderation-latest is the model selected.

  • 🇩🇪Germany marcus_johansson

    Ah, I haven't actually tested that combination, that might cause issues. OpenAI is moderated by default, no need for the external moderation for that.

    The external moderation is for instance if you want to use OpenAI moderation infront of Anthropic or Llama or something.

    But that's good to know, I would add a validation there that you can't use the same provider and moderation provider.

  • 🇹🇭Thailand AlfTheCat

    Hi Marcus, awesome thank you. I didn't know moderation was enabled by default although it now makes sense because there is the moderation provider option in the general ai settings. So I was trying to solve a problem I didn't have by using the external moderation module :).

    Thanks for the insight and I think adding that validation is a good idea.

Production build 0.71.5 2024