Skip to main content

Creating a New Fine-Tuning Model

The Create Fine-Tuning Model page provides the ability for users to fine-tune a base model using their own datasets and hyperparameters tailored to specific requirements. This process involves selecting a pre-trained model, uploading training and validation datasets, and configuring key settings.

Steps for Creating a Fine-Tuned Model:​

  1. Select a Base Model: Choose a pre-trained model from the provided list. Popular models like llama-2-70b, gpt-4-mini, and others are available for selection. Models that have already been fine-tuned are conveniently displayed at the top for easy access.

    • The fine-tuned models are listed with their completion status, making it easier to manage or reuse them.
    • The model selection gallery allows users to filter and search for specific models based on their needs.

Available Base Models:​

Here are some of the most popular models available for fine-tuning:

Model NameTypeUpdated AtSize
llama-2-70b-chatText GenerationAug 22, 2024325.7 GB
llama-2-70bText GenerationAug 16, 2024320.4 GB
llama-2-13b-chatText GenerationAug 20, 202478.1 GB
code-llama-34bCode GenerationAug 07, 202494.3 GB
gpt-4-miniChat CompletionJul 18, 2024120.5 GB
gpt-35-turbo-16kChat CompletionJul 28, 202414.7 GB
gpt-35-turbo-instructChat CompletionAug 02, 20248.9 GB
gpt-4o-2024-08-06Chat CompletionAug 06, 20249.1 GB
text-embedding-adaEmbeddingsMay 20, 20242.7 GB

You can filter, search, and choose base models depending on your project requirements.

  1. Name Your Model: Enter a name for your fine-tuned model in the Name field. A unique name helps in managing and distinguishing models, especially when working with multiple versions.

  2. Upload Training Dataset: Upload the dataset that will be used to train the model. You can either upload a new dataset or select from an existing dataset from your library. This flexibility ensures quick adaptation for various use cases.

  3. Upload Validation Dataset: In the next step, upload a validation dataset. This dataset is used to evaluate the model’s ability to generalize to unseen data during the fine-tuning process.

    • You can upload new data or select an existing validation set.
    • Alternatively, you can choose None if you don't want to use validation (though this is not recommended for best results).

  4. Configure Hyperparameters: Fine-tune key hyperparameters that influence the training process:

    • Batch Size: The number of samples processed before the model is updated.
    • Learning Rate Multiplier: Controls how much the model weights are adjusted in response to the error after each update.
    • Epochs: The number of times the model will pass through the entire training dataset.
    • These settings help in optimizing how fast the model learns and how accurately it fine-tunes.

  5. Additional Settings (Optional): Customize further settings:

    • Suffix: Add a custom suffix to help identify different experiments.
    • Seed: Set a seed value for reproducibility or leave it random. This ensures consistent training outcomes when needed.

Key Advantages:​

  • Model Personalization: Fine-tuning allows you to modify models based on specific datasets, improving the relevance and accuracy of generated outputs.
  • Easy Dataset Management: Quickly upload new datasets or reuse existing ones from your dataset library.
  • Comprehensive Hyperparameter Control: Full control over key parameters like batch size, learning rate, and epochs for performance optimization.
tip

Carefully adjusting hyperparameters can significantly speed up the fine-tuning process, improving efficiency based on the dataset size and model complexity.

Example Workflow:​

  1. Select Base Model: For instance, select llama-2-70b-chat, a model optimized for text generation tasks such as chatbot development.
  2. Upload Data: Upload a custom training dataset, such as a set of chat dialogues, to fine-tune the model for conversational capabilities.
  3. Set Hyperparameters: Choose a learning rate and batch size suitable for the hardware you have available and ensure efficient model fine-tuning.
  4. Final Step: After configuring all settings, click Save to begin the fine-tuning process.