Mathew

@lumalabsai prev @unslothai ex quant @statestreet @jpmorgan

[Mathew @DoubleMathew]

One of the most interesting projects I’ve worked on with a cracked team. We taught a 27B VLM to control robots with GRPO, entirely in simulation.

Lots more to explore in sim-only RL.

[Ben ✈️ CoRL 2026 @bentropy_ai]

You don't need a frontier model to control a robot. An open VLM already knows what to do; simulation can teach it how.

Learn more at https://t.co/1oPYvS8YYG

(10/11)

[Luma @LumaLabsAI]

Exact sequence parallelism for packed multimodal masks.

It picks ring attention, a sparse all-to-all or Ulysses per layer, matches single-GPU numerics, and beats a FlashMask + Ulysses hybrid in 47 of 48 cases on up to 128 H100s.

https://t.co/VFujPDSDgE

[Mathew @DoubleMathew]

@_DivyaMakkar @adityamakkar0 Nice work!!

[Divya Makkar @_DivyaMakkar]

Spent some time creating a lightweight, performant async RL stack in pure JAX with @AdityaMakkar0! We share a work log covering inference, RDMA weight transfer, memory optimizations, sharding and a few more JAX engineering insights we've seen along the way! It’s a bit of a long read, but I hope it sheds some light on the systems side of scaling RL!

[Mathew @DoubleMathew]

@djsaunde “But what happened next, nobody expected.”

[Mathew @DoubleMathew]

Thank you pytorch

[Mathew @DoubleMathew]

We added a UI for finetuning and chat!

[Unsloth AI @UnslothAI]

Introducing Unsloth Studio ✨
A new open-source web UI to train and run LLMs.

• Run models locally on Mac, Windows, Linux
• Train 500+ models 2x faster with 70% less VRAM
• Supports GGUF, vision, audio, embedding models
• Auto-create datasets from PDF, CSV, DOCX
• Self-healing tool calling and code execution
• Compare models side by side + export to GGUF
GitHub: https://t.co/2kXqhhvLsb
Blog and Guide: https://t.co/ENuTWal5AA

Available now on Hugging Face, NVIDIA, Docker and Colab.

[Unsloth AI @UnslothAI]

Merry Christmas from Unsloth! 🎄🎁

Thank you for all the support this year!

We’re excited to keep shipping open-source next year 🥰

[ElevenLabs Developers @ElevenLabsDevs]

Today, we launch the ElevenLabs OSS Engineers Fund - a program that provides sustained support to the open-source projects that help power our work.

Over the next six months, we are contributing $22,000 to projects our engineers rely on.

[Mathew @DoubleMathew]

You know those tasks where you’d like to load massive context? You can now finetune on those long context tasks in Unsloth.

[Unsloth AI @UnslothAI]

You can now do 500K context length fine-tuning with Unsloth!

Train gpt-oss-20b to extend its context window to 530K on 80GB VRAM & 750K+ on 192GB - no accuracy loss.

Unsloth's new algorithms + Tiled MLP = 72% less VRAM & 6x more context

Blog + Notebook: https://t.co/BkExGTX24k

[Unsloth AI @UnslothAI]

We made a free notebook that fine-tunes IBM Granite 4.0 into a powerful support agent!

This agent will enable real-time analysis & solving of customer interactions.

You'll also learn how to train models using data from Google Sheets.

Colab Notebook: https://t.co/yRqKJKOGX8

[Unsloth AI @UnslothAI]

Join us, @Pytorch and @AMD for a Virtual Hackathon on Oct 18-20. 🔥

Win $10K in prizes by training the best AI agent via Unsloth!

⭐ Sign up here: https://t.co/HeN5DeSnAn

[Mathew @DoubleMathew]

@UnslothAI 🙌

[Unsloth AI @UnslothAI]

We're teaming up with @MistralAI and @NVIDIA for an Unsloth event on Tues, Oct 21 at @YCombinator's office! 🦥

Join us in San Francisco for a night of talks, merch and more.

Food & drinks provided. RSVP required!
⭐ https://t.co/PsnAd47VBy

[Daniel Han @danielhanchen]

We managed to save >50% VRAM for multimodal RL in @UnslothAI with vLLM weight sharing! GSPO, Dr GRPO and much longer contexts are also possible!

Our notebook shows how you can design good reward functions to teach a VLM to answer hard maths and logic questions with RL

Currently Gemma 3, Qwen 2.5 VL, Llama Vision are supported & Mistral 3 + others are incoming!

Docs: https://t.co/7evxz8eGAO
Colab: https://t.co/eqvLeFnBxl

[Akshay 🚀 @akshay_pachaar]

Google just dropped a new LLM!

You can run it locally on just 0.5 GB RAM.

Let's fine-tune this on our own data (100% locally):

[Mathew @DoubleMathew]

@johnowhitaker Maybe try explicitly setting use_cache or flash attention

[Mathew @DoubleMathew]

@AlexReibman You must be offloading to disk. With 16 gb you’d need to run a quantized version

[Ahmad @TheAhmadOsman]

qwen coder 30b, specifically the Q4_K_XL dynamic quant from unsloth

has been my daily driver on the pc

i basically route everything that i don't wanna waste claude tokens on through it

and it's a mfing killer model

[Unsloth AI @UnslothAI]

The @GoogleDeepMind Gemma 3n Challenge ($150,000 in prizes) ends in 7 days!

We've made 2 new fine-tuning Gemma 3n Kaggle notebooks (Vision & Audio) to spark your creativity. Your fine-tuned model can compete for any prize.

Notebooks + Challenge Details: https://t.co/oGzfKF94AH