Alfonso² Peterssen

Saw this talk in Antwerp and loved the one-jar principle. Inspired by it, built some UIs to test Quixotic features in a Java/Spring Boot app: github.com/samie/quixot...

"Quixotic: Pure Java Inference at Native Speed" #Devoxx talk is now up. AI in a jar! youtu.be/S7x4DxwRnk4

Can Java compete with llama.cpp/ollama? YES! With @jitllm.io! jitLLM uses NVIDIA CUDA features and libraries (cuBLAS, cuDNN) and reaches 0.90× to 1.12× of llama.cpp's throughput! jitLLM is empowered by @tornadovm.org and will be soon in @langchain4j.dev and @quarkus.io. github.com/beehive-lab/...

Finally, AI in a jar. #Devoxx

"Write once, run anywhere!" Not "Write once, bridge to Python!" qxotic.ai#examples

AI in Java, at native speed, competitive with native engines on CPUs. qxotic.ai#benchmarks

AI in Java... quixotic, until now! End-to-end AI stack for the JVM, built from first principles. Will be presented at Devoxx Belgium, next week: "Quixotic: Pure Java Inference at Native Speed" m.devoxx.com/ev m.devoxx.com/events/dvbe2...

Meet Tok'n'Roll 🎸 the tokenization substrate of qxotic.ai Token-perfect. Pure Java. Zero dependencies.

The JVM is a marvel of software engineering. AI on the JVM: quixotic ... until now. Local, fast, just one command away. qxotic.ai

Here's it in action:

TruffleRuby runs Doom at, 🥁, 662 FPS! From the talk of @khasinski.bsky.social at #EuRuKo

It's release day! Finally, AI in jar.

Beep, beep! Early benchmarks of the Quixotic AI stack (JVM) vs. llama.cpp qxotic.ai

🚀 Gemma4 Java runner: • Single file, no deps • E2B→31B + MoE • GGUF + quant (F16→Q8) • Vector API ⚡ • CLI + thinking modes • GraalVM native + instant TTFT Pure Java 😏 by @mukel.bsky.social github.com/mukel/gemma4...

Over the Easter weekend, I got Gemma 4 running in pure Java. No Python. No JNI. No native code. Just the JVM. github.com/mukel/gemma4...

My @graalvm.org Native Image deep dive recording is already up: youtube.com/watch?v=1J6m... 🐰🚀 It includes the very public first demo of Project Crema, Open World for Native Image, at 2:19:54 😅 Thank you, @devoxx.com! All demos and notes are here: github.com/alina-yur/gr...

run an LLM with a supercharged engine powered by Java and GraalVM (ht @alina-yurenko.bsky.social ) www.youtube.com/shorts/7zSEa...

✅ @graalvm.org Native Image ✅ Llama3.java ✅ Vector API, FFM API ✅ Apple Silicon ❤️ — `git clone github.com/mukel/llama3... ` — `sdk install java 25.ea.17-graal` — `make native` (optionally preload a model for zero overhead) — Profit!🚀 #Java #GraalVM #LLM #LLama

I had lots of fun talking about my early days as a dev, how I caused double credit bookings all over Germany, stopped smoking to survive long GraalVM meetings and how to heat the room with Java inception. Already looking forward to part 2.

what's new in llama3.java and the upcoming @graalvm.org for JDK 24 🔥 Try it out here: github.com/mukel/llama3... @stephanjanssen.be @mukel.bsky.social #VDCERN #VoxxedDaysCERN

This will be a big deal 🪄🤩. Normally the debug experience of natively compiled languages get obfuscated due to compiler optimisations like inlining. Not here. All optimizations can be enabled to debug GraalVM native images and it looks just the same as without opts.

We just merged the current status of the upcoming JDWP support for @graalvm.org Native Image! 🥳 This will soon provide developers with the same debugging experience they are used to in Java, but for native images! Stay tuned for more details. github.com/oracle/graal...

https://buff.ly/40KmT0t Graal compiler: +10% faster inference with the latest early access build. New features: batched prompt processing & AVX512 support.