Pith. sign in

REVIEW 2 cited by

Self-Supervised Alignment with Mutual Information: Learning to Follow Principles without Preference Labels

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2404.14313 v2 pith:DRVA5RXS submitted 2024-04-22 cs.CL

classification cs.CL
keywords modelprincipleslabelspreferencepretrainedconstitutionslanguagerates
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

When prompting a language model (LM), users often expect the model to adhere to a set of behavioral principles across diverse tasks, such as producing insightful content while avoiding harmful or biased language. Instilling such principles (i.e., a constitution) into a model is resource-intensive, technically challenging, and generally requires human preference labels or examples. We introduce SAMI, an iterative algorithm that finetunes a pretrained language model (without requiring preference labels or demonstrations) to increase the conditional mutual information between constitutions and self-generated responses given queries from a dataset. On single-turn dialogue and summarization, a SAMI-trained mistral-7b outperforms the initial pretrained model, with win rates between 66% and 77%. Strikingly, it also surpasses an instruction-finetuned baseline (mistral-7b-instruct) with win rates between 55% and 57% on single-turn dialogue. SAMI requires a model that writes the principles. To avoid dependence on strong models for writing principles, we align a strong pretrained model (mixtral-8x7b) using constitutions written by a weak instruction-finetuned model (mistral-7b-instruct), achieving a 65% win rate on summarization. Finally, we investigate whether SAMI generalizes to diverse summarization principles (e.g., "summaries should be scientific") and scales to stronger models (llama3-70b), finding that it achieves win rates of up to 68% for learned and 67% for held-out principles compared to the base model. Our results show that a pretrained LM can learn to follow constitutions without using preference labels, demonstrations, or human oversight.

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. What's on My Network? Using Large Language Models to Identify Real-World IoT Devices at Scale

    cs.LG 2025-09 conditional novelty 6.0 of 10

    An instruction-tuned LLaMA 3.1 8B model, trained on LLM-generated pseudo-labels, is claimed to identify IoT device vendors from passive network metadata with 98.25% top-1 accuracy across 2,015 vendors.

  2. Building Task Bots with Self-learning for Enhanced Adaptability, Extensibility, and Factuality

    cs.CL 2025-08 conditional novelty 2.0 of 10

    A thesis that combines self-learning from dialog logs, schema-guided prompting, and self-aligned factuality to build task bots with minimal human intervention.

Pith tools