Writing procedural and stylistic capabilities into quantization cells

Determine whether procedural or stylistic capabilities can be written into the quantization cells of a frozen 4-bit language-model artifact, beyond factual recall on synthetic biographies.

Background

The paper evaluates CellFill and related bounded in-cell update methods only on synthetic biography facts, where novelty is guaranteed and the information content is countable. It does not evaluate reasoning, instruction following, coding, mathematics, or other downstream capabilities.

The authors identify skills as the most important unresolved question concerning the method’s reach. Their concern is that procedural or stylistic capabilities may require coordinated parameter displacements that are poorly accommodated by independent per-weight box constraints.

References

Whether procedural or stylistic capability can be written into the cells is untested, and we regard it as the single largest open question about the method's reach: the per-weight box bounds magnitude, and skills may require coordinated displacement that the box shapes badly.

— Nothing Changed but the Model: CellFill -- Bounded In-Cell Learning for Bit-Identical, Revocable Updates to Quantized LLMs  (2608.20873 - Liu et al., 21 Aug 2026) in Section Discussion and outlook, subsection “Open problems, ranked” (1)