produced by Classmethod

# LLM articles

Large Language Modelの略称で、GPT-3やChatGPTなどの大規模な言語モデルを指します。

Kiro's GPT-5.6 Luna credit multiplier is about 83% off. I checked the display after the Terra / Luna changes.

2026.08.01

I checked why Kiro CLI 17 models had mixed pass/fail results on the VPC Flow Logs diagnostic task in aws-bench

2026.08.01

Amazon Bedrock's OpenAI GPT-5.6 Terra / Luna has been reduced in price.

2026.07.31

Building an Unbreakable RAG Knowledge Base: Lazy Initialization, Idempotent Ingestion, and Graceful Degradation

2026.07.28

Preparing LLM training data: I tried placing a 100GB slice of FineWeb on S3

小渕ヒューゴ

2026.07.28

I tried running Qwen3.6-35B-A3B on Amazon SageMaker and getting streaming responses

越井琢巳 (Koshii Takumi)

2026.07.26

I tried to organize the state of local LLMs in the summer of 2026

森茂洋 / Hiroshi Morishige

2026.07.24

Tried running the 118B code-specialized MoE "Laguna S 2.1" on DGX Spark

森茂洋 / Hiroshi Morishige

2026.07.23

I organized an overview of NVIDIA VSS

2026.07.22

I tried recognizing dumpling weighing images with a VLM

2026.07.21

I tried E2E testing Next.js LLM streaming without external APIs using Playwright

越井琢巳 (Koshii Takumi)

2026.07.20

I also compared two quantized versions of Gemma 4 NVFP4 on DGX Spark

森茂洋 / Hiroshi Morishige

2026.07.18

"Business handover" and "introducing generative AI into operations" are similar topics

ケーイチ｜生成AIコンサルタント@Classmethod

2026.07.17

Can RAG Be Replaced by Fine-Tuning? Investigating and Thinking About Knowledge Base Search Options

2026.07.17

Dumpling Making Process Recognition　NVIDIA SOP Monitoring Services

2026.07.17

I attended Build a Claw Tokyo, where Jensen Huang, the CEO, also made a surprise appearance.

森茂洋 / Hiroshi Morishige

2026.07.16

I tried out the newly supported OpenAI GPT-5.6 Sol / Terra / Luna on Amazon Bedrock

2026.07.14

OpenAI, Claude, Gemini Reasoning Control Parameters Compared — Design Philosophy of reasoning_effort / budget_tokens / thinkingLevel

2026.07.13

Have you noticed different behavior between LLM chat UIs and APIs? Here's a cross-provider breakdown of causes and solutions

2026.07.13

A new champion among partner LLMs for DGX Spark? Comparing Qwen3.6-35B-A3B-NVFP4 in two quantized versions

森茂洋 / Hiroshi Morishige

2026.07.13

# Organizing LLM Output Control and Extraction Techniques in 4 Layers — From Constrained Decoding to Prompt Design

2026.07.12

I tried Grok 4.5 with Grok Build CLI — A comparison of responses with Opus 4.8, Sonnet 5, and GPT-5.5

2026.07.12

[Session Report] Search Evolves with Agents: Agent Capabilities of Amazon OpenSearch Service [ANT407] #AWSSummit

越井琢巳 (Koshii Takumi)

2026.07.07

ちょっと話題の記事

I thought about an LLM configuration that partners with one DGX Spark

森茂洋 / Hiroshi Morishige

2026.07.06

I was able to call Claude Fable 5 on Bedrock Mantle, so I tried it out

2026.07.05

I tried out the Inference Target of AgentCore Gateway

2026.07.05

ちょっと話題の記事

How Far Can a Gaming PC Run Local LLMs? I Tried It with Ollama

2026.07.04

I summarized recent developments in design systems advancing AI agent support - Meta Astryx, v0, Serendie, SmartHR

2026.07.04

I tried running NVIDIA's time series foundation model NV-Tesseract on DGX Spark after it was released as open source

森茂洋 / Hiroshi Morishige

2026.07.04

ちょっと話題の記事

I tested the re-released "Claude Fable 5" on Amazon Bedrock

2026.07.01

Tried running DeepSeek V4 Flash-DSpark on 2 DGX Spark units

森茂洋 / Hiroshi Morishige

2026.06.30

I tried Hermes Agent's Mixture of Agents (MoA)

森茂洋 / Hiroshi Morishige

2026.06.29

ちょっと話題の記事

I ran Ornith 1.0 on DGX Spark and compared its Japanese language performance with Gemma 4 / Nemotron

森茂洋 / Hiroshi Morishige

2026.06.28

I tried out OpenRouter's MCP Server

森茂洋 / Hiroshi Morishige

2026.06.27

It seems Ollama has added MLX support, so I investigated Apple MLX and ran some benchmarks on M2 Max.

2026.06.27

LLMs Don't "Think": Understanding Hallucinations, Prompt Engineering, and Security Through the Mechanism of Token Prediction

2026.06.25