Unveiling the Learning Mind of Language Models: A Cognitive Framework and Empirical Studycs.AI updates on arXiv.org arXiv:2506.13464v3 Announce Type: replace-cross
Abstract: Large language models (LLMs) have shown impressive capabilities across tasks such as mathematics, coding, and reasoning, yet their learning ability, which is crucial for adapting to dynamic environments and acquiring new knowledge, remains underexplored. In this work, we address this gap by introducing a framework inspired by cognitive psychology and education. Specifically, we decompose general learning ability into three distinct, complementary dimensions: Learning from Instructor (acquiring knowledge via explicit guidance), Learning from Concept (internalizing abstract structures and generalizing to new contexts), and Learning from Experience (adapting through accumulated exploration and feedback). We conduct a comprehensive empirical study across the three learning dimensions and identify several insightful findings, such as (i) interaction improves learning; (ii) conceptual understanding is scale-emergent and benefits larger models; and (iii) LLMs are effective few-shot learners but not many-shot learners. Based on our framework and empirical findings, we introduce a benchmark that provides a unified and realistic evaluation of LLMs’ general learning abilities across three learning cognition dimensions. It enables diagnostic insights and supports evaluation and development of more adaptive and human-like models.
arXiv:2506.13464v3 Announce Type: replace-cross
Abstract: Large language models (LLMs) have shown impressive capabilities across tasks such as mathematics, coding, and reasoning, yet their learning ability, which is crucial for adapting to dynamic environments and acquiring new knowledge, remains underexplored. In this work, we address this gap by introducing a framework inspired by cognitive psychology and education. Specifically, we decompose general learning ability into three distinct, complementary dimensions: Learning from Instructor (acquiring knowledge via explicit guidance), Learning from Concept (internalizing abstract structures and generalizing to new contexts), and Learning from Experience (adapting through accumulated exploration and feedback). We conduct a comprehensive empirical study across the three learning dimensions and identify several insightful findings, such as (i) interaction improves learning; (ii) conceptual understanding is scale-emergent and benefits larger models; and (iii) LLMs are effective few-shot learners but not many-shot learners. Based on our framework and empirical findings, we introduce a benchmark that provides a unified and realistic evaluation of LLMs’ general learning abilities across three learning cognition dimensions. It enables diagnostic insights and supports evaluation and development of more adaptive and human-like models. Read More
Build an AI-powered website assistant with Amazon BedrockArtificial Intelligence This post demonstrates how to solve this challenge by building an AI-powered website assistant using Amazon Bedrock and Amazon Bedrock Knowledge Bases.
This post demonstrates how to solve this challenge by building an AI-powered website assistant using Amazon Bedrock and Amazon Bedrock Knowledge Bases. Read More
Migrate MLflow tracking servers to Amazon SageMaker AI with serverless MLflowArtificial Intelligence
Migrate MLflow tracking servers to Amazon SageMaker AI with serverless MLflowArtificial Intelligence This post shows you how to migrate your self-managed MLflow tracking server to a MLflow App – a serverless tracking server on SageMaker AI that automatically scales resources based on demand while removing server patching and storage management tasks at no cost. Learn how to use the MLflow Export Import tool to transfer your experiments, runs, models, and other MLflow resources, including instructions to validate your migration’s success.
This post shows you how to migrate your self-managed MLflow tracking server to a MLflow App – a serverless tracking server on SageMaker AI that automatically scales resources based on demand while removing server patching and storage management tasks at no cost. Learn how to use the MLflow Export Import tool to transfer your experiments, runs, models, and other MLflow resources, including instructions to validate your migration’s success. Read More
With a new year upon us, software and cybersecurity experts disagree on the utility of software bill of materials — in theory, SBOMs are great, but in practice, they’re a mess. Read More
A recently disclosed security vulnerability in MongoDB has come under active exploitation in the wild, with over 87,000 potentially susceptible instances identified across the world. The vulnerability in question is CVE-2025-14847 (CVSS score: 8.7), which allows an unauthenticated attacker to remotely leak sensitive data from the MongoDB server memory. It has been codenamed MongoBleed. “A […]
Cybersecurity researchers have disclosed details of what has been described as a “sustained and targeted” spear-phishing campaign that has published over two dozen packages to the npm registry to facilitate credential theft. The activity, which involved uploading 27 npm packages from six different npm aliases, has primarily targeted sales and commercial personnel at critical Read More
In December 2024, the popular Ultralytics AI library was compromised, installing malicious code that hijacked system resources for cryptocurrency mining. In August 2025, malicious Nx packages leaked 2,349 GitHub, cloud, and AI credentials. Throughout 2024, ChatGPT vulnerabilities allowed unauthorized extraction of user data from AI memory. The result: 23.77 million secrets were leaked through AI Read […]
A ransomware attack hit Oltenia Energy Complex (Complexul Energetic Oltenia), Romania’s largest coal-based energy producer, on the second day of Christmas, taking down its IT infrastructure. […] Read More
A former Coinbase customer service agent was arrested in India for helping hackers earlier this year steal sensitive customer information from a company database. […] Read More
Fortinet has warned customers that threat actors are still actively exploiting a critical FortiOS vulnerability that allows them to bypass two-factor authentication (2FA) when targeting vulnerable FortiGate firewalls. […] Read More
