Home > Workload Solutions > Data Analytics > White Papers > GenAI on Dell APEX File Storage for Azure using Databricks, HuggingFace, and MosaicML > Terminology
The following table provides definitions for some of the terms that are used in this document.
Table 1. Terminology
Term | Definition |
Fine-tuning dataset | Fine-tuning dataset refers to a specific set of data used during the process of fine-tuning a pretrained machine learning model. Fine-tuning refines a pre-trained model initially trained on a large dataset for a general task, such as language understanding. It involves additional training on a smaller, more specific dataset tailored to a particular task or domain. This smaller dataset is referred to as the fine-tuning dataset. |
Pretrained model | A pretrained model is a machine learning model trained on a large dataset for specific tasks before being made accessible for use. This pretraining allows the model to learn general patterns and features that are relevant to the task. Once pretrained, the model can be further fine-tuned on a smaller, specific dataset to adapt it to a particular application or domain. Pretrained models serve as a foundation for various AI applications, saving time and resources in the development process. |
Model fine-tuning | Model fine-tuning is the process of taking a pretrained machine learning model and further training it on a smaller, task-specific dataset. Fine-tuning helps the model adapt to a particular application or domain. Fine-tuning leverages the knowledge the model has gained during its initial training on a large dataset for a more general task. By using a fine-tuning dataset related to the specific task at hand, the model can improve its performance and accuracy for that targeted application. This process is crucial for customizing pretrained models to perform in specialized tasks or domains. |
Evaluation of fine-tuned model | Evaluating a fine-tuned model assessing its performance on a specific task or in a particular domain after it has undergone the fine-tuning process. Evaluation typically entails using a separate dataset, distinct from the one used for fine-tuning, to measure the model's accuracy, precision, recall, F1 score, or other relevant metrics. The evaluation aims to ensure that the fine-tuned model performs well in real-world scenarios and meets the wanted level of accuracy and effectiveness for the targeted application. This process is essential for verifying the success of the fine-tuning process and validating the model's suitability for the intended task or domain. |
Model inference | Model inference is the process of using a trained machine learning model to make predictions or decisions based on new, unseen data. During inference, the model processes input data to generate an output. This output could be a classification, regression, or other prediction, based on the model's nature and the task it was trained for. Unlike the training phase, where the model learns from labeled data, model inference applies the learned knowledge to make real-time or batch predictions on new data. It is a critical step in deploying machine learning models for practical applications example image recognition, language translation, and many other tasks. |