Show HN: kulyk-uk-en and kulyk-en-uk (opens in new tab)

(huggingface.co)

2 pointsyehors8mo ago1 comments

A new lightweight model to do machine translation Ukrainian<->English using recently published LFM2 model by LiquidAI.

With llama-server it gives 60 tokens/sec. on Apple M1, with NVIDIA RTX 3090 Ti it's faster around 230 tokens/sec with batch size = 1.

1 comments

yehorsOP8mo ago

Here's demos:

j / k navigate · click thread line to collapse

(huggingface.co)

2 pointsyehors8mo ago1 comments

A new lightweight model to do machine translation Ukrainian<->English using recently published LFM2 model by LiquidAI.

With llama-server it gives 60 tokens/sec. on Apple M1, with NVIDIA RTX 3090 Ti it's faster around 230 tokens/sec with batch size = 1.