tokenizers icon indicating copy to clipboard operation
tokenizers copied to clipboard

Proposal: Add Golang Bindings for tokenizers

Open Nav31 opened this issue 9 months ago • 3 comments

  • Performance-Critical Applications – Many performance-intensive apps use Go for concurrency and efficiency. Native bindings allow seamless integration without overhead.
  • Server-Side NLP – Go is widely used in backend services. Golang bindings would make it easier to use Hugging Face tokenizers in APIs, microservices, and inference systems.
  • Better Compatibility – Many Go-based projects rely on Python/C++ workarounds for NLP tasks. Direct bindings would reduce complexity and improve developer experience.
  • Improved Inference Pipelines – Many real-time AI systems are built in Go. Native tokenizers would remove Python bottlenecks, making pipelines faster and more reliable

Nav31 avatar Mar 17 '25 21:03 Nav31

How do you imagine that? extern and cgo combination?

Or something more high level, like uniffi-rs, rust2go or fcplug?

strategy155 avatar Mar 31 '25 07:03 strategy155

Wow ahha that's a lot of reactions!

ArthurZucker avatar May 27 '25 10:05 ArthurZucker

we unfortunately don't have the man power to maintain this 😿

ArthurZucker avatar May 27 '25 10:05 ArthurZucker