Quickstart
This guide will show you how to customize an LLM for your application. We’ll cover how to find the best model for your application and fine-tune it. We'll also look at where to go next to find all the information you need to take full advantage of our suite of tools.
You can use the Data Composer tool to generate a dataset if you do not have one. Otherwise, you can format your own according to the specifications below or use our test dataset.
Find the Best Model
We can use the MyxMatch utility to help us find the best model for any use case. Use a prompt like you did above to view a ranking of open source models to find the best candidate for fine-tuning. Click the MyxMatch and fill in the form with a name to label the job and a prompt that resembles the samples in your dataset. You can click on the Myxboards tab where you'll see your ranking job's status. Once it is complete, click on the job and view the ranking results. Clicking any of the train buttons will redirect you to the tune tool with your chosen model.
![remyx model matching example](https://remyx.ai/images/docs/myxmatch_example.gif)
Tune a Model
If you used MyxMatch to get a recommended model, you should see the Generate Text form to train your model. Otherwise, you can find this utility under the "Train" section in the home page by clicking the "Generate Text" option.
Fill in the details by providing a name for your training job. You can upload your generated dataset or point to a dataset in Hugging Face using the drop box or text box.
Ensure your data has two string columns for the model inputs and outputs like the following example:
prompt | response |
---|---|
sample prompt | sample response |
A recommended model will be automatically selected if you've been redirected from the MyxMatch utility but you can override your selection with the size/accuracy dial or under the "Advanced Settings" dropdown.
You can then navigate to the Models tab where you can see the progress status of your training job. Once it turns green, youcan access its model dashboard which includes details about its training and evaluation results and deployment options.
![remyx fine tuning example](https://remyx.ai/images/docs/generate_tune_example.gif)
What's next?
Great, you've customized an LLM model for your application! 🎉 You can explore other tools in the Remyx Studio like: