Skip to main content

CurateClick

Back to Blog

Tag: on-device

1 post found

Hy-MT1.5-1.8B-2bit: Tencent Open-Sources a 574MB On-Device Translation Model That Beats 72B Giants
Tencent Hunyuan Team releases Hy-MT1.5-1.8B-2bit, a breakthrough 2-bit quantized translation model. At just 574MB, it outperforms Tower-Plus-72B and commercial APIs across 33 languages. Full offline on-device deployment available.