bloc97 / DeMoLinks
DeMo: Decoupled Momentum Optimization
☆188Updated 6 months ago
Alternatives and similar repositories for DeMo
Users that are interested in DeMo are comparing it to the libraries listed below
Sorting:
- an open source reproduction of NVIDIA's nGPT (Normalized Transformer with Representation Learning on the Hypersphere)☆101Updated 3 months ago
- supporting pytorch FSDP for optimizers☆82Updated 6 months ago
- The simplest, fastest repository for training/finetuning medium-sized GPTs.☆134Updated this week
- Focused on fast experimentation and simplicity☆74Updated 6 months ago
- NanoGPT-speedrunning for the poor T4 enjoyers☆66Updated 2 months ago
- Pytorch implementation of the PEER block from the paper, Mixture of A Million Experts, by Xu Owen He at Deepmind☆127Updated 10 months ago
- ☆98Updated 5 months ago
- Repo for "LoLCATs: On Low-Rank Linearizing of Large Language Models"☆239Updated 4 months ago
- Explorations into the proposal from the paper "Grokfast, Accelerated Grokking by Amplifying Slow Gradients"☆101Updated 6 months ago
- ☆114Updated 6 months ago
- PyTorch implementation of models from the Zamba2 series.☆182Updated 5 months ago
- Normalized Transformer (nGPT)☆183Updated 7 months ago
- research impl of Native Sparse Attention (2502.11089)☆54Updated 4 months ago
- ☆78Updated 11 months ago
- Collection of autoregressive model implementation☆85Updated 2 months ago
- smolLM with Entropix sampler on pytorch☆150Updated 7 months ago
- ☆81Updated last year
- ☆53Updated last year
- Tiny re-implementation of MDM in style of LLaDA and nano-gpt speedrun☆52Updated 3 months ago
- Efficient optimizers☆220Updated this week
- When it comes to optimizers, it's always better to be safe than sorry☆241Updated 2 months ago
- An implementation of PSGD Kron second-order optimizer for PyTorch☆91Updated 2 months ago
- Minimal (400 LOC) implementation Maximum (multi-node, FSDP) GPT training☆127Updated last year
- ☆132Updated 10 months ago
- Quick implementation of nGPT, learning entirely on the hypersphere, from NvidiaAI☆284Updated 3 weeks ago
- WIP☆93Updated 10 months ago
- Fast, Modern, and Low Precision PyTorch Optimizers☆94Updated this week
- OpenCoconut implements a latent reasoning paradigm where we generate thoughts before decoding.☆173Updated 5 months ago
- A MAD laboratory to improve AI architecture designs 🧪☆120Updated 6 months ago
- PCCL (Prime Collective Communications Library) implements fault tolerant collective communications over IP☆94Updated last month