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。

I tried OpenAI GPT-6 Astra, which became generally available on Amazon Bedrock

2026.09.09

I thought about AI agent Memory from the perspectives of correction, inspection, and deletion

森茂洋 / Hiroshi Morishige

2026.09.07

ちょっと話題の記事

I compared Qwen3 8B and GLM Z1 9B Flash across 4 configurations by use case on DGX Spark

森茂洋 / Hiroshi Morishige

2026.09.06

I thought about a configuration for delivering reasoning models to the field on NVIDIA Jetson

森茂洋 / Hiroshi Morishige

2026.09.05

話題の記事

I want you to check out WebMCP because it's hot right now

2026.09.03

Amazon Bedrock now has Claude Fable 5.1 available, so I tried calling it from the Converse API

2026.09.02

I tried running GLM-5.3-Flash 320B on 2 DGX Spark units and measured whether using 2 units is worth it

森茂洋 / Hiroshi Morishige

2026.09.01

I tried repurposing Nemotron Lightning as a dedicated judge for a team assistant using SFT and Capability Cards

森茂洋 / Hiroshi Morishige

2026.09.01

I investigated the cause of access_denied being returned with GPT-5.6 on Amazon Bedrock

2026.08.30

The Story of How I, a Business Professional Who Has Inhabited the IT Industry for a Quarter Century, Built My Own App

2026.08.30

ちょっと話題の記事

I tested the practical boundary by running 320B GLM-5.3-Flash on a single DGX Spark

森茂洋 / Hiroshi Morishige

2026.08.30

I thought about the ideal form of AI Gateway and router for using coding agents as a team

森茂洋 / Hiroshi Morishige

2026.08.30

I thought about how to use NVIDIA NeMo Relay and NeMo Switchyard as an AI Gateway

森茂洋 / Hiroshi Morishige

2026.08.29

Automatically routing Slack agent requests with NeMo Switchyard and a post-trained judge model

嶋田怜央成 / Reona Shimada

2026.08.28

I looked into how the prefill router in NVIDIA NeMo Switchyard works

森茂洋 / Hiroshi Morishige

2026.08.22

I tried Grok 4.6 on Amazon Bedrock with Grok Build CLI

2026.08.19

ちょっと話題の記事

I tried xAI's Grok 4.6, which became available on Amazon Bedrock, using Mantle

2026.08.19

I tried NVIDIA's new agent framework NOOA with a local LLM on DGX Spark

森茂洋 / Hiroshi Morishige

2026.08.18

I tried setting up a development environment by connecting the Pi coding agent to Switchyard

森茂洋 / Hiroshi Morishige

2026.08.17

I tried fine-tuning Nemotron 3.5 Lightning with LoRA post-training to make it a judge for an LLM router

森茂洋 / Hiroshi Morishige

2026.08.16

I tried connecting factory floor videos with process documents to make it searchable for abnormality candidates in operations

tanaka-takeru@製造BT部

2026.08.14

注目の記事

Why can tacit knowledge of veterans not be passed on even when documented?

2026.08.13

I compared three types of speculative decoding (DSpark / DFlash / MTP) of NVIDIA Nemotron 3.5 Lightning on DGX Spark

Guri / Hajime Oguri

2026.08.12

I tried NVIDIA Nemotron 3.5 Lightning 30B-A3B-NVFP4

森茂洋 / Hiroshi Morishige

2026.08.12

I tried NeMo Switchyard v0.2.0 reborn in Rust

森茂洋 / Hiroshi Morishige

2026.08.12

I tried attaching eyes and ears to DeepSeek V4 Flash, which doesn't support vision, using Qwen-MM-Plugins

森茂洋 / Hiroshi Morishige

2026.08.11

I tried running Meta's open-weight Muse Glimmer 30B on DGX Spark

森茂洋 / Hiroshi Morishige

2026.08.11

ちょっと話題の記事

Tried running DeepSeek V4 Flash-0731 at 284B on two DGX Spark units

森茂洋 / Hiroshi Morishige

2026.08.10

Open-weight model team AI environment covering everything from development to business use

森茂洋 / Hiroshi Morishige

2026.08.08

I tried building a team-shared RAG using NVIDIA RAG Blueprint × DGX Spark and connecting it with MCP

森茂洋 / Hiroshi Morishige

2026.08.07

話題の記事

Tried running Kimi K3 2.8T on a single DGX Spark

森茂洋 / Hiroshi Morishige

2026.08.06

I attended NVIDIA Japan's NPN Partner Model Customization Bootcamp

森茂洋 / Hiroshi Morishige

2026.08.06

I tried NVIDIA's agent execution runtime NeMo Relay

森茂洋 / Hiroshi Morishige

2026.08.06

話題の記事

Tried running DeepSeek V4 Flash-0731 on a single DGX Spark with llama.cpp

森茂洋 / Hiroshi Morishige

2026.08.02

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