great work by Mario Sieg, already integrated into PCCL to make quantization of pseudo-gradients in DiLoCo lightning fast

Awesome work by Mario Sieg to accelerate quantization of pseudo-gradients in decentralized training settings like DiLoCo - already integrated in pccl (prime collective communication library)

To implement a GPT-2 in my custom PyTorch-like ML framework, I added boolean tensors. Boolean tensors are used for filtering, indexing and as attention and loss masks and much more. The main logical operators AND, OR, XOR and NOT are now supported. Another more step towards LLM

This is not PyTorch. It’s Magnetron - my tiny ML framework with a PyTorch-like API, designed for microcontrollers and IoT. Now supports nn.Module, nn.Linear, nn.Sequential, nn.ModuleList, nn.ModuleDict, and more. The API got very close to PyTorch the last month, more to come!

Seems like I've come to a point where my C code crashes a modern LLVM compiler and makes it spit out LLVM IR 🙄

Launching SYNTHETIC-2: our next-gen open reasoning dataset and planetary-scale synthetic data generation run. Powered by our P2P inference stack and DeepSeek-R1-0528, it verifies traces for the hardest RL tasks. Contribute towards AGI via open, permissionless compute.

Our fast quantization library piquant will support 2-bit quantization and new 4-bit kernels for even higher performance on AVX-512 CPUs in the next release. Get ready to crunch those packed integers!

Sometimes I have random creative "attacks" where I build random stuff. Last time it was techno music generated with pure code, this time it's a small cryptocurrency... It's not about money, it's about exploring, learning and having fun. This approach taught me 99% of what I

Upcoming features of piquant - our blazingly fast quantization library - int2 quantization - direct quanitization of bf16 tensors - sign quantization - SIMD kernels for stochastic rounding

Chris Barber gabriel Noam Brown Jason Wei rohan anil will brown finbarr Jeremy Howard xjdr doomslide Casey Flint great list, missing (totally unbiased selection): Johannes Hagemann samsja Justus Mattern Grad Jackmin mike64_t Mario Sieg

magnetron's GPT-2 inference example is working!! one year ago I wrote the first C file to build a small pytorch clone, today LLMs can be implemented by it. Took a lot of hours and work to get everything right and I can't wait to continue with even more advanced models like

My attempt at creating a creepy "Jarvis like" sci-fi horror AI assisant. Built him a year ago in C++ with raylib for rendering and whisper.cpp for speech recognition. The mouth animation definitely needs some work

my matrix multiplication kernels for magnetron now beat PyTorch's performance on my CPU: Magnetron matmul: 1955.2 GFLOP/s Torch matmul: 1752.3 GFLOP/s check out the kernel code: github.com/MarioSieg/magn… magnetron detects cache sizes and uses state of the art block tuning