Prem Studio is a powerful feature that allows developers to create custom AI models without requiring machine learning expertise or human intervention.
Unlike traditional fine-tuning approaches, which require extensive experience and manual oversight, our autonomous fine-tuning agent only requires a well-curated dataset and a few granular parameters.
The user interface (UI) and user experience (UX) of the Prem Studio agents are designed to be intuitive and easy to use, so any type of developer can use it and focus on building the AI-enhanced applications they want to build.
Under the hood, Prem has sophisticated data augmentation and distributed training architectures that maximize your custom model’s performance.
Save time, money and resources by letting Prem handle the heavy lifting.
This is one of our most detailed guides. So buckle up and get ready to fine-tune your models like a pro.
Click here to skip to the step by step guide
Prepare your data (JSON or JSONL Only)
In order to fine-tune your model, you are required to submit a JSON or JSONL dataset in the following format:
or in JSONL format (one JSON object per line):
Configure your fine-tuning job
When configuring your fine-tuning job, you are required to provide the following:
Fine-tuning Job Name
Base Model: Select the base model you’d like to use.
Training Depth: Use the slider to select the training depth between quick and deep training.
Key Differences:
A Dataset: Drag and drop your dataset into the Dataset section. Make sure your dataset is in the correct format as shown above.
Toggle the Synthetic Data Generation switch to enable or disable the generation of additional training examples.
Configure how additional training examples will be generated with the following parameters:
Now just click the Start Fine-Tuning Job button to start the Autonomous Fine-Tuning Agent.
Fine-Tuning in Action: Data Augmentation
If you are using synthetic data generation, the Data Augmentation phase begins. If you’re not using synthetic data generation, you’ll be queued and go straight to the fine-tuning phase.
Data Augmentation for synthetic data generation
You'll receive confirmation emails about the status of your fine-tuning jobs
The first email will confirm that the fine-tuning process has begun.
The second email will confirm that the fine-tuning has been completed.
Try your fine-tuned model in the Lab
Now that the fine-tuning is complete, head over to the Lab to test it out.
You can also download your datasets by clicking on the fine-tuning job and then clicking the Download Dataset button for both the original dataset and the augmented dataset.
If you’d like to learn more about the Prem Lab and how to experiment with different models, check out our Lab Guide.
You can search for your fine-tuned model the same way you would with any of the pre-trained models available.
You can always change the system prompt, change params, and add repositories to curate the fine-tuned model to your liking.
You can also test out your fine-tuned model individually in the chat section of the Lab
Launch your fine-tuned model with the Launchpad
If you’d like to learn more about the Prem Launchpad and how to deploy your model, check out our Launchpad Guide.
You’ll also need to know how to download and use the SDK to integrate your model into your applications. Check out the SDK Guide for more information.
Now that you used the Autonmous fine-tuning agent and testing is complete, navigate to the Launchpad to deploy your model so it’s ready for integrating into your applications.
You will still have the ability to configure the model’s params, system prompts and repositories.
Integrate your fine-tuned model into your applications using the SDK
Evaluate your fine-tuned model's performance with Traces
Learn more about the Prem Traces and how to use them, by checking out our Monitoring & Traces Guide.
As you continue to use your launched/deployed fine-tuned model, you can use the Traces feature to monitor its performance. We highly recommend that you check in on your fine-tuned model’s performance regularly to ensure it’s performing as expected.
You’ve now successfully fine-tuned your model and launched it using the Launchpad.
Remember to check that Prem also has an API that works just like the SDK. So if you’re using a different language, you can still use Prem.
Prem Studio is a powerful feature that allows developers to create custom AI models without requiring machine learning expertise or human intervention.
Unlike traditional fine-tuning approaches, which require extensive experience and manual oversight, our autonomous fine-tuning agent only requires a well-curated dataset and a few granular parameters.
The user interface (UI) and user experience (UX) of the Prem Studio agents are designed to be intuitive and easy to use, so any type of developer can use it and focus on building the AI-enhanced applications they want to build.
Under the hood, Prem has sophisticated data augmentation and distributed training architectures that maximize your custom model’s performance.
Save time, money and resources by letting Prem handle the heavy lifting.
This is one of our most detailed guides. So buckle up and get ready to fine-tune your models like a pro.
Click here to skip to the step by step guide
Prepare your data (JSON or JSONL Only)
In order to fine-tune your model, you are required to submit a JSON or JSONL dataset in the following format:
or in JSONL format (one JSON object per line):
Configure your fine-tuning job
When configuring your fine-tuning job, you are required to provide the following:
Fine-tuning Job Name
Base Model: Select the base model you’d like to use.
Training Depth: Use the slider to select the training depth between quick and deep training.
Key Differences:
A Dataset: Drag and drop your dataset into the Dataset section. Make sure your dataset is in the correct format as shown above.
Toggle the Synthetic Data Generation switch to enable or disable the generation of additional training examples.
Configure how additional training examples will be generated with the following parameters:
Now just click the Start Fine-Tuning Job button to start the Autonomous Fine-Tuning Agent.
Fine-Tuning in Action: Data Augmentation
If you are using synthetic data generation, the Data Augmentation phase begins. If you’re not using synthetic data generation, you’ll be queued and go straight to the fine-tuning phase.
Data Augmentation for synthetic data generation
You'll receive confirmation emails about the status of your fine-tuning jobs
The first email will confirm that the fine-tuning process has begun.
The second email will confirm that the fine-tuning has been completed.
Try your fine-tuned model in the Lab
Now that the fine-tuning is complete, head over to the Lab to test it out.
You can also download your datasets by clicking on the fine-tuning job and then clicking the Download Dataset button for both the original dataset and the augmented dataset.
If you’d like to learn more about the Prem Lab and how to experiment with different models, check out our Lab Guide.
You can search for your fine-tuned model the same way you would with any of the pre-trained models available.
You can always change the system prompt, change params, and add repositories to curate the fine-tuned model to your liking.
You can also test out your fine-tuned model individually in the chat section of the Lab
Launch your fine-tuned model with the Launchpad
If you’d like to learn more about the Prem Launchpad and how to deploy your model, check out our Launchpad Guide.
You’ll also need to know how to download and use the SDK to integrate your model into your applications. Check out the SDK Guide for more information.
Now that you used the Autonmous fine-tuning agent and testing is complete, navigate to the Launchpad to deploy your model so it’s ready for integrating into your applications.
You will still have the ability to configure the model’s params, system prompts and repositories.
Integrate your fine-tuned model into your applications using the SDK
Evaluate your fine-tuned model's performance with Traces
Learn more about the Prem Traces and how to use them, by checking out our Monitoring & Traces Guide.
As you continue to use your launched/deployed fine-tuned model, you can use the Traces feature to monitor its performance. We highly recommend that you check in on your fine-tuned model’s performance regularly to ensure it’s performing as expected.
You’ve now successfully fine-tuned your model and launched it using the Launchpad.
Remember to check that Prem also has an API that works just like the SDK. So if you’re using a different language, you can still use Prem.