AI and machine learning

PyTorch

PyTorch is what I train models in, including the transformer segmentation models behind my IEEE Access paper.

How I use it

My published research trained transformer-based segmentation models on a custom road-defect dataset in PyTorch. That is the framework I use for anything that needs a custom model rather than an API call. Most client work does not need one, and I say so when it does not.

Offered through

Shipped with PyTorch

All projects →

Questions clients ask about PyTorch

Do I need a custom model or can I use an API?
Usually an API or a pretrained model plus good data handling gets you there faster and cheaper. A custom model is justified when the domain is narrow and the data is yours. I have done both and will tell you which you need.

Work with me

Need PyTorch work done?

I work remotely with clients across the US, Canada, the UK, the UAE, Brazil and Australia. Tell me what you are building and I will reply with questions, a plan and an estimate.

Related