ELI5 GenAI Day 03 - NeMo-GuardRails

去年年底在 [DevFest 2023 Taipei] 講過一場 [Future-Proofing Enterprise Conversations: Integrating NeMo-GuardRails & PaLM2 - JimmyLiao]

探討如何運用 NeMo-GuardRails 確保溝通的合規性與安全性，同時借助 Gemini/Gemma 模型的強大自然語言處理能力，開發出更具人性化的對話系統，提升企業級對話平台的成效

不多說，直接來看一個 Blacklist 黑名單功能

`moderation.co`

`actions.py`

`block_list.txt`

這樣就可以在對話中，過濾掉一些不當的用語，讓對話更加合規。

### 本篇同步發表於 [iThome] 與 [個人電子報](https://memo.jimmyliao.net/) -> 訂閱訂起來

## English Version

Last year, I gave a talk at DevFest 2023 Taipei on [Future-Proofing Enterprise Conversations: Integrating NeMo-GuardRails & PaLM2 - JimmyLiao]

The talk explored how to use NeMo-GuardRails to ensure the compliance and security of communications. By leveraging the powerful natural language processing capabilities of the Gemini/Gemma models, we developed a more humanized conversational system to enhance the effectiveness of enterprise-level conversation platforms.

Let's take a look at a Blacklist feature.

By using a Blacklist, you can filter out inappropriate language in conversations to make them more compliant.

Please check above for the code snippets.

This post is simultaneously published on [iThome] and my personal [newsletter] -> Subscribe now!