LunNova
Follow
Stars
This is a replicate of DeepSeek-R1-Zero and DeepSeek-R1 training on small models with limited data
Clean, accessible reproduction of DeepSeek R1-Zero
Multi-platform high-performance compute language extension for Rust.
Optax is a gradient processing and optimization library for JAX.
A collection of niche / personally useful PyTorch optimizers with modified code.
Minimalistic 4D-parallelism distributed training framework for education purpose
A single-line modification to any (dualizer-based) optimizer that allows the optimizer to adapt to the scale of the gradients as they change during training
Synthetic data derived by templating, few shot prompting, transformations on public domain corpora, and monte carlo tree search.
🔎 Monitor deep learning model training and hardware usage from your mobile phone 📱
A Pytorch implementation of stochastic addition.
Fast, Modern, Memory Efficient, and Low Precision PyTorch Optimizers
Official PyTorch Implementation for Paper "No More Adam: Learning Rate Scaling at Initialization is All You Need"
A PyTorch library for implementing flow matching algorithms, featuring continuous and discrete flow matching implementations. It includes practical examples for both text and image modalities.
supporting pytorch FSDP for optimizers
Code for CRATE (Coding RAte reduction TransformEr).
The Laminar Cannon is an experimental device for directing a cohesive stream of highly filtered air at an individual in motion.
Small improvements and tweaks for rM devices, covering both rM1 and rM2.
additional functionality via binary patching
Somewhere Systems' Autonomous Agent Engine