Extreme Compression of Per-Layer Embeddings via Cross-Layer Latent Vector Quantization

No. 143 of 245 · arXiv · Jiacheng Liu, Yaxin Luo, Xiaohan Zhao et al. · published 23 Oct 2026, 04:00 UTC

AI & machine learning
Paper

3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked

OPTIC is a method for compressing per-layer embeddings in language models, which make up 35.2% of Gemma 4 E4B's parameters. Instead of giving each token-layer vector its own short code, it stores one shared code per token and decodes it at each layer with a layer-specific affine decoder, fitted to…

Read the original at arxiv.org. The judges saw only its text, with the site, the author and link domains removed; they are shown again after ranking.

Score

0.64total under rubric v2.0

Significance0.63
Reader interest0.71
Audience breadth0.52
Practical relevance0.66
Substantiated0.68
Technical depth0.79
New information0.90
Counts against: Promotional0.07
Counts against: Overclaim0.23

Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).

Question	Judge 1	Judge 2	Judge 3	Mean	Spread	Weight	Contribution (weight times mean)
Significance	0.66	0.45	0.78	0.63	0.33	+0.25	+0.157
Reader interest	0.65	0.60	0.88	0.71	0.28	+0.20	+0.142
Audience breadth	0.48	0.35	0.72	0.52	0.37	+0.10	+0.052
Practical relevance	0.67	0.50	0.82	0.66	0.32	+0.15	+0.099
Substantiated	0.71	0.60	0.72	0.68	0.12	+0.10	+0.068
Technical depth	0.75	0.80	0.82	0.79	0.07	+0.10	+0.079
New information	0.93	0.80	0.97	0.90	0.17	+0.10	+0.090
Promotional	0.08	0.10	0.03	0.07	0.07	−0.15	−0.011
Overclaim	0.17	0.40	0.12	0.23	0.28	−0.15	−0.035
Sum of contributions							+0.642
Claim-check penalty							0.000
Total, as published							0.642

Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.

Also covered by

No other coverage was clustered with this story.

Elsewhere

Elsewhere: HN —

Recorded for context only, checked 24 Oct 2026, 08:54 UTC. It played no part in the ranking.

Panel for this edition

Four AI agents ranked this edition: three judges scored every story blind to its site and author, and a fourth compared the top stories’ claims with the sources they cite. Summaries are written by AI.

Model versions are pinned for every edition and listed in its JSON. This one was scored under rubric v2.0; How ranking works describes the current rubric, v2.2.