Tim Janik (@TimJanik) header
Tim Janik (@TimJanik) profile picture

Working on jj-fzf (Jujutsu TUI), Anklang (DAW), Imagewmark, Audiowmark; https://teapawt.net/t.co/FCVYmOjgsW

Hamburg
Joined October 2011
CMSCOUT 0.2.0 is out! 🚀 🔍 Code-motion aware diffs for TS/JS/TSX/JSX, Go, Bash, C & C++ 📦 Matches functions, classes, methods, imports, macros, JSX across renames/moves 📊 Per-block similarity + change summaries ⚙️ Git external diff support + --summary #100DaysOfCode
1
1
64
Unbelievable, Qwen3.8-Flash-Next tops DSv4Pro 0813 on the Engineering Index and runs on an RTX 4090 with ~20tps (Q3kxl, 128k context, system with 96GB RAM). And it has vision… But, asking it about this chart, the model found the most interesting score to be the 27B xhigh ;-)
166
Replying to @MaxForAI
Not DeepSeek specific, here is Qwen3.8 contradicting itself in German, LLMs simply like persona settings (requested German, I don't read Chinese):
1
1
573
Recommended Keynote! Especially if you wonder where we are heading after the AI craze: After the AI Hype – What’s Real, and What’s Next - Richard Campbell - 2026 youtube.com/watch?v=uWnUnMph…
39
Are Local LLMs Ready for Production? I gave a local Qwen3.6-27B a try at a real world task. testbit.eu/2026/local_llms_f…
1
2
413
ROTFL 😂 Qwen3.6-27B is hilarious, opt to "have fun" in the midst of a coding session. My favorite model so far!
1
3
271
Replying to @davis7
Nice! A skill is so much easier to test. Just for searching docs, using `git clone --depth 1` could be more efficient. And maybe make `btca cleanup` explicit… ;-)
1
86
And we're back after the loss of signal!
1
87
Not a crescent moon, but a crescent Earth...
2
2
178
Imagewmark 0.6.0 🚀 🖼️ C++ #watermarking (much faster, lower RAM) 🔧 Split arch: C++ for embedding, Python for extraction ⚙️ New test‑suite attacks: gamma, brightness‑contrast 🐞 Fixes: .ppm/.tif support, sign‑preservation 🔍 Blind detection Contributor: @polysynth
1
1
2
368
jj-fzf 0.38.0 is out! 🚀 🎯 Auto-position on @ upon startup 🧠 Updated commit message prompt for smaller LLMs 📖 Man page now includes F11 preview toggle docs 🐛 Fixed SFX support for Ubuntu 22.04 🔧 Fixed quoting for jj split ⚠️ Requires jj-0.38.0 and fzf-0.67.0
1
1
139
Nice speedups in llama.cpp lately. WIth the right set of flags, Qwen3.5-27B runs at 50tok/s + VL embedder on a 4090 at 24GB VRAM: llama-server -c 0 -m Qwen3.5-27B.IQ4NL.gguf -ngl 99 --cache-type-k q4_0 --cache-type-v q4_0 -ub 1024 -b 2048 --mmproj Qwen3.5-27B.F16.mmproj.gguf
2
206