trends.barnas.net

method · first seen 15 Oct 2026

ternary quantization

Also: ternary weights, ternary weight neural network

Compressing neural network weights to three values, such as -1, 0, and 1, to reduce memory and compute.

Mentions per week

0219 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Rank 109 on 2 Nov 2026 · 6 mentions in 28 days · 6 independent groups · papers → code · first seen 3 weeks ago · acceleration -1.39

Earliest sources

Latest sources

Mentioned together with

small language modelfrozen embedding tablepost-training quantizationtool callingmixture of expertsspeculative decodingself-hosted LLM deploymentsystolic arrayautomatic speech recognitionspiking neural networks