Metraipolis › archive › dc_5cb762bcebcc

In-Context Learning Amplifies a Latent Symbolic Circuit — arXiv:2609.36265

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.36265
License (as archived)
CC0-1.0
Archived
2026-10-07T16:25:32.635Z
SHA-256 of the archived text
25a298732166d294c7ed42c4fc8bfd609de5f4a1061b0b541e68c3c684cc2fe0
Stable URL of this page
https://metraipolis.com/documento/dc_5cb762bcebcc

How to cite

In-Context Learning Amplifies a Latent Symbolic Circuit — arXiv:2609.36265 — arXiv (en), https://arxiv.org/abs/2609.36265 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5cb762bcebcc (Metraipolis archive dc_5cb762bcebcc, sha256 25a298732166d294c7ed42c4fc8bfd609de5f4a1061b0b541e68c3c684cc2fe0)
BibTeX
@misc{dc5cb762bcebcc,
  title = {In-Context Learning Amplifies a Latent Symbolic Circuit — arXiv:2609.36265},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.36265},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5cb762bcebcc; sha256 25a298732166d294c7ed42c4fc8bfd609de5f4a1061b0b541e68c3c684cc2fe0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.36265. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models can learn abstract rules from just a few in-context examples, but how their internal mechanisms activate as examples accumulate is not well understood. We trace a three-stage symbolic reasoning circuit (abstraction, induction, retrieval) across shot counts in three model families and find it is detectable and functional well before the model achieves high accuracy. Per-head causal contribution grows up to 8x from 1- to 10-shot, and cross-shot activation patching raises accuracy from 1% to 56% at 0-shot and 17% to 88% at 1-shot. Function vectors scaled and injected at 0-shot rescue accuracy up to 86%, largely substituting for the induction stage but depending critically on an intact downstream retrieval stage. The infrastructure for abstract rule-following is present in the weights before any demonstrations; in-context examples, function vectors, and related interventions appear to supply input to the same latent circuit.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.36265 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T15:54:53.799Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).