Note: You may find it easier to read about Intel Transfer Learning tool, follow the Get Started guide, and browse the API material from our published documentation site https://intel.github.io/transfer-learning.
Transfer learning workflows use the knowledge learned by a pre-trained model on a large dataset to improve the performance of a related problem with a smaller dataset.
Intel® Transfer Learning Tool makes it easier and faster for you to create transfer learning workflows across a variety of AI use cases. Its open-source Python* library leverages public pretrained model hubs, Intel-optimized deep learning frameworks, and your custom dataset to efficiently generate new models optimized for Intel hardware.
This project documentation provides information, resource links, and instructions for the Intel Transfer Learning Tool as well as Jupyter* notebooks and examples that demonstrate its usage.
Features:
Intel Optimizations:
The Intel Transfer Learning Tool lets you train AI models with TensorFlow or PyTorch using either no-code command line interface (CLI) commands at a bash prompt, or low-code application programming interface (API) calls from a Python script.
Use your own dataset or select an existing image or text dataset listed in the public datasets documentation. Construct your own CLI or API commands for training, evaluation, and optimization using the TensorFlow or PyTorch framework, and finally export your saved model optimized for inference on Intel CPUs.
An overview of the Intel Transfer Learning Tool flow is shown in this figure:
Intel Transfer Learning Tool Flow
Check out the Get Started Guide which will walk you through the steps to check system requirements, install, and then run the tool with a couple of examples showing no-code CLI and low-code API approaches. After that, you can check out these additional CLI and API Examples.
As described in the Get Started Guide, once you have a Python environment set up, you do a basic install of the Intel Transfer Learning Tool. Note that the default installation includes PyTorch and not TensorFlow. Here are some examples of commands you will find in the Get Started Guide:
pip install intel-transfer-learning-tool
Then you can use the Transfer Learning Tool CLI interface (tlt) to train a
PyTorch image classification model (RenderedSST2), download and use an
existing built-in dataset (tf_flowers), and save the trained model to
/tmp/output
using this one command:
tlt train --framework pytorch --model-name efficientnet_b0 --dataset-name RenderedSST2 \
--output-dir /tmp/output --dataset-dir /tmp/data
TensorFlow can be used by installing with the following command:
pip install intel-transfer-learning-tool[tensorflow]
Use tlt --help
to see the list of CLI commands. More detailed help for each
command can be found using, for example, tlt train --help
.
Intel Transfer Learning Tool provides standard evaluation metrics such as accuracy and loss for validation/test/train sets. While important, it's essential to acknowledge that these metrics may not explicitly capture biases. Users should be cautious and consider potential biases by analyzing disparities in the data and model prediction. Techniques such as confusion matrices, PR curves, ROC curves, local attribution-based and gradCAM
explanations, can all be good indicators for bias. Clear documentation of model behavior and performance is also crucial for iterative bias mitigation. Intel® Explainable AI Tools provides components that demonstrate the aformentioned techniques with Explainer, a simple API providing post-hoc model distillation and visualization methods, as well as The Model Card Generator which provides an interactive HTML report that containing these workflows and demonstrations of model behavior.
The Intel Transfer Learning Tool team tracks bugs and enhancement requests using GitHub issues. Before submitting a suggestion or bug report, search the existing GitHub issues to see if your issue has already been reported.
See Legal Information for Disclaimers, Trademark, and Licensing information.