Pith. sign in

REVIEW 7 cited by

Large Language Models with Controllable Working Memory

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

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 2211.05110 v1 pith:RYKRYSLF submitted 2022-11-09 cs.CL cs.AIcs.LG

Daliang Li , Ankit Singh Rawat , Manzil Zaheer , Xin Wang , Michal Lukasik , Andreas Veit , Felix Yu , Sanjiv Kumar This is my paper · ORCID
classification cs.CLcs.AIcs.LG
keywords modelcontextknowledgecontrollabilitylanguagemodelsrobustnessinformation
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Large language models (LLMs) have led to a series of breakthroughs in natural language processing (NLP), owing to their excellent understanding and generation abilities. Remarkably, what further sets these models apart is the massive amounts of world knowledge they internalize during pretraining. While many downstream applications provide the model with an informational context to aid its performance on the underlying task, how the model's world knowledge interacts with the factual information presented in the context remains under explored. As a desirable behavior, an LLM should give precedence to the context whenever it contains task-relevant information that conflicts with the model's memorized knowledge. This enables model predictions to be grounded in the context, which can then be used to update or correct specific model predictions without frequent retraining. By contrast, when the context is irrelevant to the task, the model should ignore it and fall back on its internal knowledge. In this paper, we undertake a first joint study of the aforementioned two properties, namely controllability and robustness, in the context of LLMs. We demonstrate that state-of-the-art T5 and PaLM (both pretrained and finetuned) could exhibit poor controllability and robustness, which do not scale with increasing model size. As a solution, we propose a novel method - Knowledge Aware FineTuning (KAFT) - to strengthen both controllability and robustness by incorporating counterfactual and irrelevant contexts to standard supervised datasets. Our comprehensive evaluation showcases the utility of KAFT across model architectures and sizes.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 7 Pith papers

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

  1. MINTQA: A Multi-Hop Question Answering Benchmark for Evaluating LLMs on New and Tail Knowledge

    cs.CL 2024-12 conditional novelty 7.0 of 10

    MINTQA provides 28,366 multi-hop QA pairs across popular/unpopular and old/new knowledge, with sub-questions, and shows that even the best LLMs achieve only about 62% accuracy even with retrieval.

  2. Teaching Smaller Language Models To Generalise To Unseen Compositional Questions (Full Thesis)

    cs.CL 2024-11 conditional novelty 7.0 of 10

    Smaller language models can generalize to unseen compositional questions when trained and evaluated with retrieval-augmented contexts, and combining Wikipedia retrieval with LLM-generated rationales improves accuracy.

  3. DeRAGEC: Denoising Named Entity Candidates with Synthetic Rationale for ASR Error Correction

    cs.CL 2025-06 conditional novelty 6.0 of 10

    DeRAGEC explicitly denoises retrieved named-entity candidates with phonetic scores, definitions, and synthetic rationales, improving ASR error-correction WER and NE hit ratio without additional training.

  4. KnowShiftQA: How Robust are RAG Systems when Textbook Knowledge Shifts in K-12 Education?

    cs.CL 2024-12 conditional novelty 6.0 of 10

    A new benchmark, KnowShiftQA, shows that retrieval-augmented LLMs drop 22 to 27 points in accuracy when textbook facts are hypothetically updated to conflict with the model's parametric knowledge.

  5. ICT: Image-Object Cross-Level Trusted Intervention for Mitigating Object Hallucination in Large Vision-Language Models

    cs.CV 2024-11 conditional novelty 6.0 of 10

    ICT steers attention-head activations toward visual information using blur-derived trusted and untrusted pairs, improving object hallucination benchmarks in LLaVA-v1.5 and Qwen-VL without slowing generation.

  6. Understanding the Dark Side of LLMs' Intrinsic Self-Correction

    cs.CL 2024-12 conditional novelty 5.0 of 10

    Intrinsic self-correction makes state-of-the-art LLMs overturn correct answers across four task types, and simple question repetition or tiny fine-tuning reduces this damage.

  7. SKETCH: Structured Knowledge Enhanced Text Comprehension for Holistic Retrieval

    cs.CL 2024-12 reject novelty 3.0 of 10

    SKETCH combines semantic chunking and a knowledge graph retriever, and the paper claims it tops Naive RAG, RAPTOR, semantic-only, and KG-only baselines on RAGAS metrics, though the reported results are internally inco...

Pith tools