Fine-Tuning Models Overview

The Fine-Tuning Models page allows users to view and manage all the fine-tuned models that have been created. This page provides an overview of the models, their current status, and actions that can be performed on them.

Key Features:

• Filter Models: Users can filter models based on their status: All, Successful, or Failed.
• Model Details: Each model is listed with essential information like its name, fine-tuned model details, and status.
• Actions: Users can take actions on the model, like re-running fine-tuning or editing hyperparameters.

tip

Fine-tuning allows users to improve model performance for specific tasks by training the model with additional data. Use it to enhance task-specific accuracy.

How to Use:

1. Browse Created Models: Scroll through the models listed on this page, filter them based on success or failure, and select any model for further actions.
2. Create New Model: Click the Create button in the top-right corner to initiate a new fine-tuning process.

Explore the Fine-Tuning Models section to optimize model performance based on specific datasets and parameters.

Field	Description
Model Name	The name of the fine-tuned model
Finetuned Model	The fine-tuned model ID and associated data
Status	The current status (Succeeded, Failed)
Actions	Options to view evaluation, delete, etc.

---

Fine-Tuning Evaluation

The Fine-Tuning Evaluation page provides a detailed evaluation of the fine-tuned model's performance. Users can track the checkpoints created, snapshots generated, and view a graphical representation of the model’s progress.

Evaluation Features:

• Graphical Representation: A chart showcasing the model's performance throughout the fine-tuning process.
• Timeline: A timeline view of when checkpoints were created and snapshots were generated during the training process.
• Detailed Model Information: Model details such as Model Name, Created At, Finished At, and Fine-Tuned Model ID are displayed.

info

Evaluation data can help you assess model accuracy and detect overfitting during the training process.

How to View the Fine-Tuning Evaluation:

1. Open Evaluation: Click on the eye icon from the Fine-Tuning Models page.
2. View Graph: The evaluation page displays a chart that visually represents the progress of the fine-tuning process.
3. Checkpoints and Snapshots: Details about checkpoints and snapshots are listed, including the step number and snapshot IDs.

Model Attribute	Details
Model Name	Name of the original base model
Created At	Timestamp of when fine-tuning started
Finished At	Timestamp of when fine-tuning was completed
Fine-Tuned Model	Identifier of the fine-tuned model
Status	Final status of the fine-tuning process (Succeeded or Failed)
Error (if applicable)	Error message or details in case of failure

---

Example Page:

In the Fine-Tuning Models page, if no models have been created yet, a No Data placeholder will be shown. Once models are created, they will be listed with their respective statuses and actions.

In the Fine-Tuning Evaluation page, users can see detailed information about the fine-tuning process, including checkpoints and the progress graph.