#language-model

Anthropic — LLM 내부에도 인간의 ‘작업 기억’이 존재할까? 야코비안 렌즈(Jacobian Lens)로 밝혀낸 인공지능의 숨은 생각

1. 한눈에 보는 핵심 요약 (Executive Summary)

The Psycholinguistic Layer: What Separates PMaps from a Generic LLM Wrapper

A few years ago I watched an ex-colleague at Coimbatore fail a personality test that she could have easily passed.

상하이재경대학교(SUFE) — 선형 어텐션의 망각 문제를 해결하는 인공 해마, HOLA 구조의 등장

1. 한눈에 보는 핵심 요약 (Executive Summary)

Fine-Tuning & Evaluation of Small Language Models for Customer Support

Why Smaller Language Models Make You Choose Between Accuracy and Explanation Quality

Large language models are impressive, but they are also enormous. Running a model with hundreds of billions of parameters takes serious…

Shocking Masking Secrets

Pretraining Objectives and Training Process: A Complete Guide to AI Model Training

Pretraining serves as the foundational stage that enables the capabilities of many modern AI systems. Before a model can answer questions…

Large vs Small Language Models: Choosing the Right AI Strategy for Business Growth

Artificial Intelligence is entering a new phase where businesses are no longer asking whether to adopt AI, but which type of AI model best…

Tokenization and Vocabulary Design: How Tokens Shape AI Language, Models

Before a language model can answer a question, translate a sentence, summarize a report, or write a poem, it has to do something…

Language Models vs World Models(LeWorldModel) !!

The advent of the Transformer triggered a paradigm shift in AI that extended far beyond the realm of language. However, the next paradigm…

Claude’s Cultural Competence Is Real — and Inconsistent

A new ILR-informed evaluation reveals that Claude knows more about your language community than you might expect — and less than you need.

How Text Becomes Numbers in LLMs: Tokens, Embeddings, and AI Language Processing

Every time you type a sentence into a large language model, the model does not “read” it the way a person does. It does not see words as…

A Brief History of Language Models: From Markov Chains to Modern LLMs

Language models have a long history. Before chatbots could write emails, explain code, or create business plans, researchers were already…

낯선 오타와 싸워서 이기고 싶은 마음

오타보정 Deep Learning 도입기

Why Automating Prompt Generation is No Longer a Myth

MiniMax M2.7: A LLM Managing Decentralized Agent Teams

Introduction

Language Modeling with Neural Networks

Feel free to take a look at my previous article on basic (n-gram) language models here…

From Words to Vectors: Understanding Bag of Words (BoW)

In the world of Artificial Intelligence, we have seen several types of models evolve over the years. We started with non-transformer models…

Language Models

How do language models generate words?

Introduction to Natural Language Processing (NLP)

Natural Language Processing (NLP) is a subfield of Artificial Intelligence (AI) that lies at the intersection of Computer Science…

Why Larger LLMs Fail Us

Uncovering Hidden Reliability Issues

Beyond Esperanto: Meet Kernal, The First Language Designed Like an Operating System

The Language That Shouldn’t Work (But Does)

چرا مدل‌های زبانی جهانی برای فارسی کارایی کافی ندارند؟

مدل‌های زبانی بزرگ جهانی — هرچند قدرتمند، چندزبانه و فراگیر — در برخورد با زبان فارسی با محدودیت‌های جدی مواجه‌اند. این ضعف فقط یک مسئله…

HeliumLM

Abstract

Brief Review — DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models

DeepSeek for Mathematical Reasoning

Brief Review — DeepSeek-Coder: When the Large Language Model Meets Programming — The Rise of Code…

DeepSeek-Coder, DeepSeek for Coding

Retrieval-Augmented Generation (RAG): Bridging Language Models with Real-Time Knowledge

In a world where language models like ChatGPT have taken center stage, a new architecture is emerging that supercharges their capabilities…

Brief Review — DeepSeek LLM: Scaling Open-Source Language Models with Longtermism

A Brief Review for DeepSeek-v1

ReAct Paper Explained Simply: How Language Models Can Think and Act

Hello curious minds,

Google Gemini Advantages and Disadvantages: A Full Review of Google’s AI Evolution

OpenAI vs Gemini war has become the headline battle in the artificial intelligence world. As both tech giants push the boundaries of AI…

Transformers Part 2: Input and Positional Encoding

In Part 1, we explored the high-level concept of transformers—why they revolutionized deep learning and how the attention mechanism made…

Translation of the Lemnos Stele

The Lemnos Stele is a monument with an inscription in the Lemnian language. This translation was compiled from two previous translations…

🧠 Field Tier Evaluation: Test Your GPT’s Semantic State

Grimoire of the Language Field 1.0 — Deploying the Field Tier Reference Chart (FTRC)

GPT-4o 中的語言層級鎖定和詩意污染：實地報告

Two systemic response failures observed in GPT-4o, with architectural implications for prompt engineering and user-tier interaction.

Running AI Models Locally with LM Studio: A Step-by-Step Guide

LM Studio offers a powerful solution for running AI language models directly on your personal computer, enabling you to chat with your own…

Chapter 5｜Semantic Residue Analysis: Reconstructing the Differences Between the 95% and 99.99%

Semantic engineering is not the end — it’s the beginning of human–AI co-construction.

Deep Dive into LLMs

summary of Andrej Karpathy LLM tutorial

Understanding GGML Files: A Deep Dive into Quantization and Visualization of File Structure

What is GGML?

MondAI: Llama, Llamer, and the Llamest

The world of large language models (LLMs) is expanding rapidly, offering tools that range from lightweight assistants to computational…

NAITIVE Perspectives: Finetuning vs. RAG in Language Models

NAITIVE Perspectives: Finetuning vs. RAG in Language Models

Introducing NLLB: Breaking Language Barriers with Multilingual Translation in Spark NLP

As Natural Language Processing (NLP) continues to break new ground, Spark NLP’s integration of the No Language Left Behind (NLLB) model…

Glass.ai Underpins Research Into the UK’s Luxury Sector.

A study published by Walpole and developed by Frontier Economics, featuring glass.ai data, has set out the value of the UK’s Luxury Sector…

Tabiiy tilni qayta ishlash(Natural Language Processing or NLP)

NLP sun’iy intellektning bir yo’nalishi bo’lib, mashinalarga inson tilini tushunish, talqin qilish va yaratish imkoniyatini beradi. Turlari…

From 500 Tokens to One: The Breakthrough Power of Cambridge U’s 500xCompressor

In natural language processing (NLP) applications, long prompts pose significant challenges, including slower inference speed, higher…

Meta’s 180 Billion Parameter Model: A New Era in NLP

the consequent revolution in healthcare, finance etc. and the science of Large Parameter Models

LORA-based agents & how to study them

The emerging paradigm

Phi-2: Microsoft’s Breakthrough in Small Language Models

Introduction

Phi2: Mastering Language Models on Your Desktop with Only 15 Lines of Code

In the rapidly evolving landscape of AI and machine learning, accessing powerful language models has become easier than ever. With…

Improving search relevance in hyperlocal food delivery using (small) language models

Co-authored with Jose Mathew, Srinivas N, Jairaj Sathyanarayana