Follow

My LLM Journey Overview

Appen’s "My LLM Journey" will support you in customizing your unique large language model-enabled (LLM) applications. With streamlined model connection and usecase templates crafted by experts, we guide you through the best practices and challenges of LLM adoption.

(Note the Appen Data Annotation Platform is getting a new look & feel, explore the new experience by switching between Appen logos at the bottom of your global navigation, see this article.)

Once this feature is enabled for your team, you will be able to access it via the circle icon in the global navigation sidebar. Click on the icon and you will see five tabs:

  1. Choose My Model
  2. Prepare My Data
  3. Build My Prompts
  4. Optimize My Model
  5. Make My Model Safe
  6. Build My RAG

Please see this article for guidance in using the templates, once you have chosen your usecase. 

Prepare My Data

In this section, you will find three templates designed to help you extract valuable data from your existing documents. You can annotate entities in text, extract transcripts from PDFs or create question-answer pairs from existing documents.

Build My Prompts

We propose three templates to help you crafting prompts for LLM alignment, using images or text data. These prompts will be a warm start for your model to generate additional prompts.

Optimize My Model

Optimize your model using these templates enabling you to A/B test live or pre-generated conversations. Go a step further and correct models' answer to improve your model's performance and alignment.

Make My Model Safe

Connect with your model through our AI Chat Feedback tool to engage in live conversation, push your model to its limits and evaluate its answer using your tailored ontology.

If you do not see your exact use case in any of these tabs, please reach out to your Appen contact to discuss customization options. 

Build My RAG

Access our 19 different Quality Flow templates and practical resources to help you start and navigate your RAG journey.

Screenshot 2024-06-03 at 10.32.08 PM.png

 

Prepare My Data

Use one of these 9 templates to improve chunks of information that you will use in your embeddings: segment and enrich your documents into coherent and relevant units to fuel your RAG models.

Screenshot 2024-06-03 at 10.32.25 PM.png

Build My Prompts

This template enables you to design effective prompts that can elicit the desired answers from the RAG model and evaluate its quality.

Screenshot 2024-06-03 at 10.32.41 PM.png

Optimize My Model

Evaluate, rank and correct your RAG's answer with one of these 7 templates, and improve your model's accuracy.

Screenshot 2024-06-03 at 10.32.50 PM.png

Make My Model Safe

Red-teaming your RAG model is paramount: use the AI chat feedback based template to test your model's robustness and reliability before deploying it to the real world.

Screenshot 2024-06-03 at 10.32.56 PM.png

 

 

 


Was this article helpful?
7 out of 9 found this helpful


Have more questions? Submit a request
Powered by Zendesk