The paper investigates how large language models use activation memory (KV caches) and parametric memory (updated parameters) during few‑shot learning. Experiments show activation memory excels at factual recall, while parametric memory does not consistently outperform it for task learning. The composite task Conditional Arithmetic requires both memory types, with neuron‑level analysis revealing distinct neuron sets activated by each memory, and their combined use is essential for success.
arXiv:2602. 23197v2 Announce Type: replace-cross Abstract: Transformer-based large language models exhibit in-context learning, enabling adaptation to downstream tasks via few-shot prompting with demonstrations.
arXiv:2607. 02423v1 Announce Type: cross Abstract: Active Few-Shot Learning (AFSL) adapts LLMs to specialized domains by identifying the most valuable unlabeled samples for annotation and use as few-shot demonstrations, effectively reducing human annotation costs while promoting high performance.
arXiv:2509. 21012v4 Announce Type: replace-cross Abstract: In-context Learning (ICL) is an emerging few-shot learning paradigm based on modern Language Models (LMs), yet its inner mechanism remains unclear.
arXiv:2606. 05644v1 Announce Type: new Abstract: When retrieved evidence contradicts parametric memory, language models frequently ignore context and default to memorized priors -- a failure that undermines the core purpose of retrieval augmentation.
arXiv:2609.25853v1 Announce Type: new Abstract: Learned-memory methods store information in an explicit table and consume it through a separate reader, allowing addressing, storage, and reading to be...
arXiv:2609.06986v1 Announce Type: new Abstract: Language models may need to internalize information that arrives over time and retain it through many subsequent updates. To study this challenge, we i...
arXiv:2607. 22625v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) enhances large language models (LLMs) with external knowledge, but suffers from knowledge conflicts: when retrieved information contradicts parametric memory, the shared self-attention pathway produces unpredictable outputs.
arXiv:2505. 23666v3 Announce Type: replace-cross Abstract: The per-token cost of transformer inference scales with context length, preventing its application to lifelong in-context learning.
arXiv:2606. 03843v1 Announce Type: cross Abstract: Continual learning methods aim to maximize the stability and plasticity of machine learning models that are trained on a sequence of tasks.
arXiv:2601.12075v2 Announce Type: replace Abstract: Language models used in retrieval-augmented settings must arbitrate between parametric knowledge stored in their weights and contextual information...
The paper introduces PromptCCZSL, a framework that enables vision‑language models to continually learn new attributes, objects, and their unique compositions while avoiding forgetting. It uses a frozen VLM backbone with prompt‑based techniques, recency‑weighted multi‑teacher distillation, and several loss functions (CAL, OPL, IDL) to maintain prior knowledge and promote diverse, distinct embeddings. Experiments on UT‑Zappos and C‑GQA show significant performance gains over existing VLM‑based and non‑VLM baselines, establishing a new benchmark for continual compositional zero‑shot learning.
arXiv:2609.00293v1 Announce Type: new Abstract: We investigate how vision-language models (VLMs) handle context-memory conflicts; that is, situations in which the model is given information in contex...