Colendi AIResearch
Menu

Inference & Serving · 2025

BitNet b1.58 2B4T Technical Report

Shuming Ma, Hongyu Wang, Shaohan Huang, et al. (Microsoft Research)

Primary source

First open 2B-scale LLM trained natively in 1.58-bit ternary weights, matching full-precision peers of its size while cutting memory ~6x and running fast on CPU.

Editorial record

Plain-language summary

First open 2B-scale LLM trained natively in 1.58-bit ternary weights, matching full-precision peers of its size while cutting memory ~6x and running fast on CPU.

Knowledge graph

Relationships

Descendants

  • Depends onEvidence: Strongly supported

    Attention Is All You Need

    Builds on the inference serving lineage in the archive

    freshness sweep 2026

Source record

Provenance

Record ID
P-610
Record created
2026-07-13
Last reviewed
2026-07-14
Record version
2

Citation caveat: Citation metadata is approximate and marked unverified in the source dataset.