ginc-memory-matrix-dimensions
IN premise — summaries/2026/08/24/xie-2021-icl-bayesian-sR-references-chunk-2.md
Created 2026-08-25T02:58:56+00:00
The GINC dataset uses a 10×10 memory matrix (10 entities × 10 properties) with 100 entries, where the first column is a fixed delimiter token and each entry maps to a vocabulary token.
Summary
The GINC dataset is a small, fixed grid of 10 entities crossed with 10 properties, giving 100 cells where the first column is always the same placeholder and every other cell holds a single token from a bounded vocabulary. This means any downstream processing—indexing, lookup, model inference—is constrained by that rigid 10-by-10 shape, with no room for variable-length rows, extra columns, or free-text content.