Mattermost Inappropriate filter Plugin (Beta)

This plugin allows you to censor inappropriate messages on your Mattermost server. The plugin checks all messages a machine learning model before they are posted to any channel. Inappropriate messages are then blocked until user confirms they want to send them.

Supported Mattermost Server Versions: 5.2+

Plugin Marketplace

1. Go to Main Menu -> Plugin Marketplace in Mattermost.
2. Search for "Inappropriate Filter" or manually find the plugin from the list and click Install
3. Once the plugin has downloaded and been installed, click Configure.

Setup with AWS Sagemaker

1. Under Sagemaker console go to Notebook -> Notebook Instances and create a Jupyter instance
2. Open the notebook instance and create a new 'conda_tensorflow2_p36' notebook
3. Run the following script in a cell to load and deploy the model:

import sagemaker
sess = sagemaker.Session()
role = sagemaker.get_execution_role()

from sagemaker.huggingface.model import HuggingFaceModel

#Hub Model configuration. <https://huggingface.co/models>
hub = {
  'HF_MODEL_ID':'hamzaMM/Inappropriate-Filter', # model_id from hf.co/models
  'HF_TASK':'text-classification' 
}

role = IAM role ARN

#create Hugging Face Model Class
huggingface_model = HuggingFaceModel(
   env=hub, # configuration for loading model from Hub
   role=role, # iam role with permissions to create an Endpoint
   transformers_version="4.6", # transformers version used
   tensorflow_version="2.4", # tensorflow version used
   py_version='py37', # python version used
)

#deploy model to SageMaker Inference
predictor = huggingface_model.deploy(
   initial_instance_count=1,
   instance_type="ml.t2.medium" #Change to any instance type
   #Un comment to add elastic inference to speed up prediction latency
   #accelerator_type='ml.eia2.medium',  
)

After the model is deployed, it will create a model, endpoint configuration, and endpoint which will serve the model.

To use custom endpoint configuration create a endpoint using the sagemaker console, and add paramter endpoint_name with the deploy method. More information available here: https://sagemaker.readthedocs.io/en/stable/api/inference/model.html

Once the endpoint is serving you may close and stop the notebook instance.

Setup with local server

Coming Soon

Configuration in Mattermost webapp

Get Access Key from IAM -> Users -> Security Credentials -> Create Access Key

Get Endpoint Name from Sagemaker Console -> Inference -> Endpoint

Get region from Sagemaker Console -> Inference -> Models -> My_Model

Usage