Introducing Optimum: The Optimization Toolkit for Transformers at Scale
What changed
What do Tesla, Google, Microsoft and Facebook all have in common? Through Optimum, our new open source library, we aim to build the definitive toolkit for Transformers production performance, and enable maximum efficiency to train and run models on specific hardware. We will soon announce new Hardware Partners who have joined us on our journey toward Machine Learning efficiency.
Why it matters
A concrete addition to Practical AI: Tools, Models & Frameworks: it changes what's available to builders today rather than being general commentary.
How it compares
Related prior coverage to compare against:
- Accelerated Inference with Optimum and Transformers Pipelines
- Optimization story: Bloom inference
- Introducing Decision Transformers on Hugging Face 🤗
Sources
- Introducing Optimum: The Optimization Toolkit for Transformers at Scale (huggingface-blog)primary