Hugging Face And Arize AI Partner To Help Train Unstructured Models
Arize AI and Hugging Face have partnered in the goal of making those who want to take advantage of Hugging Face's pre-trained models for transfer learning have a smooth operation with a powerful toolset to identify issues and train effectively.
Created on August 10|Last edited on August 10
Comment
With the struggles of developing models from scratch, and especially when you lack the dataset needed to perform such a task, Hugging Face is a great place to find and download pre-trained models to use as a starting point in transfer learning. Though, even when using a high quality model in transfer learning, adapting it to your specific needs can introduce biases which should be avoided.
That's where Arize AI wants to step in, and with this company partnership, they aim to support those who rely on the models hosted on Hugging Face for transfer learning in the goal of mitigating bias introduced in the training process by offering a platform which is purpose-built to detect and analyze model shortcomings with powerful performance tracking, dataset optimization, and model comparison tools.
Arize's platform was built so that ML developers could quickly identify model issues on both structured and unstructured data through tools like intuitive 3D-space embedding visualizations and filterable data displays. All this data visualization works in tandem with automated monitoring tools to catch potential issues like drift or data quality issues. Once identified, model comparisons over time as well before and after changes or subsequent training let you see if you're heading in the right direction.
Find out more
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.