Metraipolis › archive › dc_0c1ef79c8d35
Behavioral Capacity Certificates for Quantized Language Models — arXiv:2609.37887
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.37887
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T17:31:51.755Z
- SHA-256 of the archived text
4e8331fbdb164da52ba9dbec37e2a915989beabe80a15a84cf81ee23a0311586- Stable URL of this page
https://metraipolis.com/documento/dc_0c1ef79c8d35
How to cite
Behavioral Capacity Certificates for Quantized Language Models — arXiv:2609.37887 — arXiv (en), https://arxiv.org/abs/2609.37887 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0c1ef79c8d35 (Metraipolis archive dc_0c1ef79c8d35, sha256 4e8331fbdb164da52ba9dbec37e2a915989beabe80a15a84cf81ee23a0311586)
BibTeX
@misc{dc0c1ef79c8d35,
title = {Behavioral Capacity Certificates for Quantized Language Models — arXiv:2609.37887},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.37887},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0c1ef79c8d35; sha256 4e8331fbdb164da52ba9dbec37e2a915989beabe80a15a84cf81ee23a0311586},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.37887. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Activation and key-value cache precision change what a quantized language model computes without altering its stored weights. Direct weight-code bounds, however, assign identical complexity to deployments that behave differently and charge separately for weight codes that behave identically. Behavioral Capacity Certificates (BCC) charge for behavior using the aggregate prior mass of complete implementations---weights, scales, activation and cache rules---that induce the same bounded loss. When quantization merges implementations, this shared mass lowers the complexity penalty, and a break-even law determines when the saving survives the cost of validating it. BCC supports a three-step deployment workflow, and our experiments verify each step. First, a forward-only screen shortlists per-layer bit-widths by how often candidate perturbations preserve the reference predictions, with quality comparable to Hessian-guided selection at lower preprocessing cost. Second, margin-certified cells identify weights that can be pruned or sign-flipped without changing the deployed behavior: every permitted combination preserves all declared predictions, and on OLMoE-1B-7B and SmolLM2-1.7B, independent probes bound the probability that any permitted combination changes a prediction on new text. Third, BCC bounds the population loss of the deployed model, nonvacuously for complete decoders and more tightly than the compressed-code route. At equal cache memory, giving keys higher precision than values yields lower NLL and higher prediction agreement on GPT-2, Qwen2.5, and SmolLM2, together with a tighter complexity bound in the GPT-2 audit. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.37887 — 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 2026-10-07T16:59:48.388Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).