# Prometheus Mind: Retrofitting Memory to Frozen Language Models

arXiv: https://arxiv.org/abs/2601.15324 · HTML: https://mlsanity.com/paper/2601.15324/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 32 references that are in the map.
- First posted on arXiv 18 Jan 2026.

## Builds on (most cited first, 15 of 15)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [REALM: Retrieval-Augmented Language Model Pre-Training](https://mlsanity.com/paper/2002.08909/) (2020)
- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory](https://mlsanity.com/paper/2504.19413/) (2025)
- [Retrieval-Augmented Generation for AI-Generated Content: A Survey](https://mlsanity.com/paper/2402.19473/) (2024)
- [Titans: Learning to Memorize at Test Time](https://mlsanity.com/paper/2501.00663/) (2024)
- [Augmenting Language Models with Long-Term Memory](https://mlsanity.com/paper/2306.07174/) (2023)
- [Linear Representations of Sentiment in Large Language Models](https://mlsanity.com/paper/2310.15154/) (2023)
- [S-LoRA: Serving Thousands of Concurrent LoRA Adapters](https://mlsanity.com/paper/2311.03285/) (2023)
- [Steering Large Language Models using Conceptors: Improving Addition-Based Activation Engineering](https://mlsanity.com/paper/2410.16314/) (2024)
- [Interpretable Steering of Large Language Models with Feature Guided Activation Additions](https://mlsanity.com/paper/2501.09929/) (2025)
- [Your Transformer is Secretly Linear](https://mlsanity.com/paper/2405.12250/) (2024)
- [Linear Attention via Orthogonal Memory](https://mlsanity.com/paper/2312.11135/) (2023)

## Built on by (most cited first, 1 of 1)

- [Black-Mamba: Biologically-Inspired Leaky Accumulation for Conceptual Knowledge under Distribution Drift](https://mlsanity.com/paper/2607.18899/) (2026)

## Abstract (arXiv)

Adding memory to pretrained language models typically requires architectural changes or weight modification. We present Prometheus Mind, which retrofits memory to a frozen Qwen3-4B using 11 modular adapters (530MB, 7% overhead) -- fully reversible by removing the adapters. Building this system required solving four problems: (1) Extraction -- we develop Contrastive Direction Discovery (CDD), which finds semantic directions via minimal pairs without labeled data. (2) Training -- end-to-end optimization collapses; stage-wise training of each adapter on simple proxy tasks succeeds. (3) Injection -- learned encoders fail to generalize; we find that lmhead-weight rows already provide the mapping we need, requiring no training. (4) Hidden state collapse -- transformers make ``wife'' and ``brother'' 0.98+ similar; we train projections to recover distinction (0.98 rightarrow 0.09). On PrometheusExtract-132 (132 cases), the system achieves 94.4% retrieval on clean inputs (n=54, 95% CI: [84.9%, 98.1%]), degrading to 19.4% on informal inputs with ellipsis, filler words, or implicit subjects (n=36). The primary bottleneck is relation classification (47.3% accuracy), responsible for most extraction errors.