MEMO: A Modular Framework for Training a Dedicated Memory Model on New Knowledge Without Modifying LLM Parameters

Large language models become static after pretraining. Their knowledge does not update as the world changes. Retraining a full LLM is too expensive at modern scales. Fine-tuning risks degrading previously learned knowledge. Retrieval-augmented generation (RAG) struggles when answers require reasoning across many documents.

A team of researchers from the National University of Singapore, MIT CSAIL, A*STAR, and the Singapore-MIT Alliance for Research and Technology (SMART) proposes a new approach called MEMO (Memory as a Model).

What Problem Does MEMO Solve?

Existing methods for integrating new knowledge into LLMs fall into three categories. Non-parametric methods like RAG retrieve documents at inference time. They are sensitive to retrieval noise and struggle with cross-document reasoning. Parametric methods such as continual pretraining or supervised fine-tuning internalize knowledge into model weights. They are computationally expensive and cause catastrophic forgetting, where new training degrades previously acquired knowledge. Latent memory methods compress knowledge into soft tokens. These representations are tightly bound to the model that produced them — a limitation the research team calls representation coupling which limits transferability across LLMs.

MEMORY as a Separate Model

MEMO separates memory from reasoning. The MEMORY model is a small, dedicated language model trained to internalize knowledge from a target corpus. The EXECUTIVE model is the main LLM — frozen and queried only through its standard input-output interface.

In experiments, the MEMORY model is Qwen2.5-14B-Instruct. The EXECUTIVE model is either Qwen2.5-32B-Instruct or Gemini-3-Flash, a proprietary closed-source model. Because MEMO treats the EXECUTIVE model as a black box, it does not require weight access or output logits.

How the MEMORY Model is Trained

Training begins with a five-step data synthesis pipeline guided by a GENERATOR model — Qwen2.5-32B-Instruct in experiments. The pipeline converts a raw document corpus into a reflection QA dataset: question-answer pairs that represent corpus knowledge under diverse query variations.

The five steps are:

Fact extraction — direct extraction of explicitly stated facts, and indirect extraction of inferred information, run in parallel per document chunk.
Consolidation — QA pairs sharing a common context (entity, time period, relationship) are merged into multi-fact pairs.
Verification and rewriting — each QA pair is checked for self-containment. Pairs with unresolved pronouns or implicit references are rewritten using the source chunk or discarded.
Entity surfacing — QA pairs are generated where questions encode entity attributes and relationships, and answers reveal entity identities. This targets the reversal curse, where models trained on “A is B” fail to infer “B is A.”
Cross-document synthesis — the GENERATOR model constructs QA pairs spanning multiple documents. It identifies two types of cross-document connections: converging clues (multiple documents about the same entity) and parallel properties (different entities sharing a common attribute or role).

Step-5 is the most critical component. A leave-one-out ablation shows that removing it drops accuracy from 24.00% to 6.37% on NarrativeQA. It is also the dominant source of training pairs in the final dataset.

The MEMORY model is then trained via supervised fine-tuning (SFT). The loss is computed over answer tokens only. Source documents are never provided at inference. The model must answer from internalized parametric knowledge.

Inference: The Structured Multi-Turn Protocol

At inference, the EXECUTIVE model queries the MEMORY model through a structured multi-turn protocol with three sequential stages.

Stage 1: Grounding. The EXECUTIVE model decomposes the query into atomic sub-questions. Each targets a single identifying constraint. The MEMORY model answers each independently.

Stage 2: Entity identification. Using the grounding responses, the EXECUTIVE model issues targeted follow-up sub-queries. It iteratively narrows down candidate entities until one is confirmed or the stage budget runs out.

Stage 3: Answer seeking and synthesis. Conditioned on the identified entity, the EXECUTIVE model queries the MEMORY model for supporting facts. It then synthesizes all retrieved responses into a final answer.

The MEMORY model’s responses are compact natural-language snippets. Their length is independent of corpus size, so retrieval cost does not scale with the number of documents. This contrasts with RAG, where inference cost grows with the corpus.

Experimental Results

MEMO is evaluated on three benchmarks: BrowseComp-Plus (multi-hop deep-research), NarrativeQA (discourse understanding over books and movie scripts), and MuSiQue (2–4 hop reasoning over Wikipedia paragraphs). Baselines include BM25, NV-Embed-V2, HippoRAG2, and Cartridges. Cartridges requires white-box access to the EXECUTIVE model and scored 0.00% on BrowseComp-Plus and 3.75% on NarrativeQA.

On NarrativeQA with Gemini-3-Flash, MEMO achieves 53.58%. HippoRAG2 reaches 23.21% on the same setup. On MuSiQue, MEMO achieves 60.20% against HippoRAG2’s 57.00%. On BrowseComp-Plus, MEMO achieves 66.67% against HippoRAG2’s 66.33%.

With Qwen2.5-32B-Instruct as EXECUTIVE model, MEMO achieves 54.22% on BrowseComp-Plus and 48.30% on MuSiQue. Switching to Gemini-3-Flash yields gains of 12.45%, 26.73%, and 11.90% on the three benchmarks. The MEMORY model is not retrained when the EXECUTIVE model changes.

Robustness to retrieval noise: The research team evaluates performance when distractor documents are added to the corpus. NV-Embed-V2 and HippoRAG2 drop by up to 6.22% on BrowseComp-Plus when one negative document is added per evidence document. MEMO’s accuracy on the same benchmark changes by +0.55% — within one standard deviation.

MEMORY model architecture robustness: The research team also tests three MEMORY model families at similar parameter scale: Qwen2.5-1.5B-Instruct, Gemma3-1B-IT, and LFM2.5-1.2B-Instruct (a hybrid state-space and transformer architecture). Performance is largely consistent across all three, indicating the framework is not sensitive to the specific pretraining lineage of MEMORY model.

Continual Knowledge Integration via Model Merging

MEMO supports incremental knowledge updates through model merging. When a new corpus arrives, a separate MEMORY model is trained on it independently. Its task vector — the parameter difference from the base model — is then merged with the existing MEMORY model in parameter space.

The research team test this on NarrativeQA using TIES merging (ρ=0.3). For K=2 corpora, merging accumulates 48 GPU-hours versus 72 GPU-hours for full retraining — a 33% reduction. At K=10, merging scales as Θ(K) while full retraining scales as Θ(K²), yielding a 5.5× saving (240 vs. 1,320 GPU-hours).

The merged MEMORY model trails full retraining by 11.04% under Qwen2.5-32B-Instruct (15.81% vs. 26.85%). It trails by 19.11% under Gemini-3-Flash (34.47% vs. 53.58%). Despite this gap, it outperforms all retrieval baselines on NarrativeQA.

Marktechpost’s Visual Explainer

Marktechpost — Research Explainer
MEMO: Memory as a Model

01 / 06 — The Problem
LLMs Freeze After Pretraining
Their knowledge becomes outdated as the world evolves.

Large language models are static once pretraining ends. For applications requiring up-to-date or domain-specific knowledge, three approaches exist — and each has a critical flaw.

🔍RAGSensitive to retrieval noise. Struggles when answers span multiple documents.

⚙Fine-TuningCauses catastrophic forgetting. Expensive. Cannot be used on proprietary LLMs.

💾Latent MemoryRepresentations are tightly coupled to one specific model architecture only.

MEMO — Memory as a Model — from researchers at NUS, MIT CSAIL, and A*STAR addresses all three limitations simultaneously.

02 / 06 — The Concept
Memory Separated From Reasoning
Two models. One frozen. One trained on new knowledge.

MEMO introduces two distinct model roles that operate together.

◆ MEMORY ModelA small, dedicated language model trained to internalize knowledge from a target corpus. It stores facts and cross-document relationships in its parameters. It never sees source documents at inference — it answers only from what it has learned.

◇ EXECUTIVE ModelThe main LLM — frozen and unchanged throughout. It queries the MEMORY model through targeted sub-questions, reasons over retrieved responses, and produces the final answer. Works with any LLM, including closed-source APIs.

In experiments: Qwen2.5-14B-Instruct as MEMORY model. Qwen2.5-32B-Instruct or Gemini-3-Flash as EXECUTIVE model. Only black-box API access required — no weights, no logits.

03 / 06 — Training
How the MEMORY Model Is Built
A five-step pipeline converts raw documents into a reflection QA dataset.

Fact Extraction→
Consolidation→
Verification→
Entity Surfacing→
Cross-Doc Synthesis

Fact ExtractionDirect extraction of stated facts and indirect extraction of inferred information run in parallel per document chunk.

ConsolidationQA pairs sharing a common entity, time period, or relationship are merged into multi-fact pairs.

Verification & RewritingEach pair is checked for self-containment. Pairs with unresolved pronouns or implicit references are rewritten or discarded.

Entity SurfacingQA pairs are generated where questions encode entity attributes and answers reveal identities, targeting the reversal curse.

Cross-Document SynthesisThe most critical step. Removing it drops NarrativeQA accuracy from 24.00% to 6.37%. Constructs QA pairs spanning multiple documents via converging clues and parallel properties.

MEMORY model trained via supervised fine-tuning (SFT) — loss over answer tokens only. Source documents never provided at inference.

04 / 06 — Inference
Three-Stage Query Protocol
The EXECUTIVE model queries the MEMORY model through structured sub-questions.

Complex user queries are decomposed across three sequential stages. No documents are retrieved — all answers come from internalized parametric knowledge.

Grounding — Budget: 1 interactionThe user query is decomposed into atomic sub-questions, each targeting one identifying constraint. MEMORY model answers each independently.

Entity Identification — Budget: 7 interactionsUsing grounding responses, the EXECUTIVE model issues follow-up sub-queries to iteratively narrow candidate entities until one is confirmed.

Answer Seeking & Synthesis — Budget: 8 interactionsConditioned on the confirmed entity, the EXECUTIVE model gathers supporting facts then synthesizes all retrieved responses into a final answer.

MEMORY model responses are compact natural-language snippets. Retrieval cost is fixed and does not scale with corpus size — unlike RAG.

05 / 06 — Advantages
What MEMO Does Differently
Compared to RAG, fine-tuning, and latent memory methods.

Other Methods
Retrieval noise significantly degrades RAG accuracy
Fine-tuning causes catastrophic forgetting in the LLM
Latent memory tied to one specific model architecture
Retrieval cost grows with corpus size at inference
Cannot be used with proprietary closed-source LLMs
Adding new knowledge requires full retraining

MEMO
Accuracy changes ±1.77% under added distractor documents
Main LLM stays frozen; no catastrophic forgetting possible
Works across Qwen, Gemma, and LFM2.5 architectures
Fixed-size responses; cost independent of corpus size
Black-box compatible — works with any LLM including APIs
New corpora merged via model merging without full retraining

TIES merging (ρ=0.3) cuts compute by 33% at K=2 corpora and 5.5× at K=10 corpora vs full retraining.

06 / 06 — Results
Benchmark Performance
Qwen2.5-14B-Instruct as MEMORY model. Gemini-3-Flash as EXECUTIVE model.

53.58%NarrativeQAvs HippoRAG2: 23.21%

60.20%MuSiQuevs HippoRAG2: 57.00%

66.67%BrowseComp-Plusvs HippoRAG2: 66.33%

Switching EXECUTIVE model from Qwen2.5-32B-Instruct to Gemini-3-Flash yields gains of +12.45%, +26.73%, and +11.90% across the three benchmarks — without retraining the MEMORY model.

Under retrieval noise, HippoRAG2 drops 6.22% on BrowseComp-Plus. MEMO changes by +0.55% on the same benchmark — within one standard deviation.

Source: arXiv 2605.15156 — Quek, Lee, Leong, Verma et al., NUS / MIT CSAIL / A*STAR / SMART, May 2026.

1 / 6

Marktechpost — AI Research, Simplified for Engineers
arXiv: 2605.15156

Key Takeaways

MEMO trains a dedicated MEMORY model on new knowledge, keeping the main LLM frozen and unchanged.
A five-step data synthesis pipeline converts raw documents into a reflection QA dataset capturing cross-document relationships.
At inference, a structured multi-turn protocol decomposes complex queries into targeted sub-queries to the MEMORY model.
Retrieval cost is fixed at inference time — it does not scale with corpus size, unlike RAG.
Model merging cuts cumulative training compute by 33% at K=2 corpora and 5.5× at K=10, with a measurable accuracy trade-off.

Check out the Research Paper. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.

Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us

The post MEMO: A Modular Framework for Training a Dedicated Memory Model on New Knowledge Without Modifying LLM Parameters appeared first on MarkTechPost.