Review or Reviews
테크, 개발, AI, 하드웨어 — 실사용 기반 리뷰와 가이드
최신 글
Local RAG Scored on 300 Questions: a 27B on One RTX 3090 Tied Claude Opus 5.5, and Most Misses Happened Before the Model
300 document questions with known answers, 7 local models vs Claude 5.5 on the same retrieved pages, two judges. qwen3.8 27B with thinking tied Opus 5.5; parsing and retrieval caused two thirds of the top models' misses.
로컬 RAG, 한국어 문서에 정말 맞게 답할까 — 정답 있는 300문항으로 채점해 보니 로컬 27B가 Opus 5.5와 비겼고, 남은 점수는 파싱과 검색에서 샜다
공개된 한국어 RAG 시험지 300문항으로 로컬 모델 7종과 Claude 5.5 세 모델을 같은 검색 결과 위에서 채점했습니다. 3090 한 장의 qwen3.8 27B는 thinking을 켜면 채점기 두 개 모두에서 Opus 5.5와 구분되지 않았고, 상위 모델이 틀린 문항의 약 3분의 2는 생성이 아니라 PDF 파싱과 검색 단계에서 근거를 놓친 문항이었습니다.
로컬 LLM 설치 가이드 2026 — Ollama 설치부터 모델 고르기까지 (GTX 1080 Ti 실측)
Ollama 0.40.5 설치, 첫 모델 실행, VRAM별 모델 고르기를 GTX 1080 Ti 두 장에서 직접 재며 정리했습니다. 모델 8개의 속도·실제 VRAM, 정답이 있는 한국어 문제 10개 채점, 처음 쓰는 사람이 걸리는 함정 4가지.
더 보기
Local AI Music on an RTX 3090: ACE-Step 1.5 vs LeVo 2, Measured — and the Pitfalls Nobody Mentions
ACE-Step 1.5 (MIT) made a 36-second track in about 32 s on an RTX 3090; LeVo 2 took 414 s for four songs and is research-only. The agent couldn't hear, so every check was a number. Ten pitfalls, measured.
GGUF Quantization on a GTX 1080 Ti and RTX 3090, Measured: Q8_0 to Q3_K_M, With and Without imatrix
Seven GGUF quants of Qwen3-8B measured against BF16 on a GTX 1080 Ti and an RTX 3090: KL divergence, speed, VRAM, English and Korean. IQ4_XS was the fastest 4-bit quant on both, and imatrix halved the 4-bit drift.
I Handed Game Asset Production to an AI Agent. My Job Turned Out to Be Judgment
An AI coding agent produced six commercially usable, game-ready 3D props on my RTX 3090 in two rounds. It did the work, the measuring and most of the debugging. Four of the turning points came from me looking at something and saying 'no'. Here is where those happened and the review habits behind them.
One Image to a Roblox-Ready 3D Prop on a Single RTX 3090, Using Only Commercially Usable Tools
Six low-poly game props from FLUX schnell → TRELLIS.2 → Blender on one RTX 3090: about 80 seconds per model and an 18.5 GB VRAM peak. Seven things went wrong, from an inner shell that drew a red X on a kitten's mouth to a normal map that was 37% inside-out, and here is how each one was fixed.
I Fine-Tuned a 0.8B Classifier on Korean Business Plans. It Tied My Hand-Written Rules, Then Lost on New Data
A day on one RTX 3090 trying to make a 0.8B 'picker' model judge Korean startup grant applications. Synthetic training data scored 92/150; real text with targeted edits scored 126/150, tying the hand-written rules. Then a check I wrote down in advance failed on six new documents (48 vs 52).
Qwen-Image-2.1 on One RTX 3090: 495 Generations Into Pixel-Art Game Sprites, Measured (and 6 Gotchas)
I moved every sprite for two small Android apps onto a local Qwen-Image-2.1 running on a single RTX 3090. 495 generations later: ~87 s per 1024² image, ~16 GB VRAM with CPU offload, and six gotchas that only showed up once the images had to work as real game assets.