Frequently Asked Questions
From billing, to system requirements, to what to do when the unexpected happens, here is a growing list of guidance for common questions.
No. We designed the Zetane Engine to integrate with and augment your existing AI framework and project pipeline (also described as a “workflow in machine learning”); there is no need for radical changes or for you to adopt a slew of new software tools.
Data scientists and machine learning developers can incorporate the Zetane Engine directly within their framework by a simple "pip install" of our software. The Engine provides out-of-the-box complementarity with Keras, Tensorflow, Pytorch and ONNX; once incorporated, you can explore and display the inner workings and details of your models and datasets with ease, such as the inspection of model architecture and key metrics for internal tensors.
No, but we recommend using GPUs in order to obtain the highest level of performance. You can operate the Zetane Engine without a GPU and achieve a functional and reasonable level of performance. We note that we designed Zetane to function while avoiding interferance with your computation resources.
GPUs are now common components of personal computers (iGPU, NVIDIA, AMD) and are typically compatible with the three main operating systems required for Zetane: Windows, Mac and Linux.
We designed Zetane to include capacities for remote training and inference. You can train your models remotely and take “snapshots” during training so you can analyze the process in the Zetane Engine using platforms such as Google Collab. You can stream data in real time from remote training and inference to an inspector on your personal computer.
For the time being, you can view in real-time attributes of your model only from your personal computer; expect remote, real-time capacities in the near future.
The Zetane Viewer is coming soon. This visualizer complements the Zetane Engine and is for professionals other than the developers of AI projects that typically work with computer code. The Viewer enables non-coding members of a team to review a machine learning project without needing to install development dependencies.
