Infoblock alpha
cs-ml/interpretability IB-arxiv-2607.28292-generalization-rate measured

On a 4-bit quantized OpenLLaMA-3B, CACHE-UK attains a 28% test success (generalization) rate, a 6 percentage-point improvement over the strongest adapted baseline.

Active Evidence state is recorded explicitly and may be revised by a later event.

Observed quantities

generalization_rate_pct 28 · improvement_pp 6 · quant_bits 4

01

Primary source

  1. arxiv · 2607.28292 · v1 · abstract
Title
CACHE-UK: A Stability-Aware Memory Editor for Sequentially Updated Quantized LLMs in Finance
Authors
Anubhav Lakra, Yue Feng
Version
v1
Check
Support quote found word for word in the arXiv abstract v1, 2026-10-08
Supporting quote from the source
Evaluated on a 4-bit quantized OpenLLaMA-3B model with a curated UK financial corpus of 88,021 documents, CACHE-UK reduces knowledge degradation by 11-17% relative to adapted baselines under identical 4-bit constraints -- its most robust effect -- while attaining the highest test success (generalization) rate observed in our setting (28%, a 6 percentage point improvement over the strongest adapted baseline)

02

Boundary of this block

No additional public limitation was recorded.

03

Connections

Outgoing

No outgoing public edges.

Incoming

No incoming public edges.