Hugging Face details quantization trade-offs in Transformers.js

Quantization can shrink models to a fraction of their size while maintaining usefulness. Users control the size vs. quality trade-off via the dtype parameter in Transformers.js.