Official repository for the paper "Grokfast: Accelerated Grokking by Amplifying Slow Gradients"
☆582Jun 28, 2024Updated 2 years ago
Alternatives and similar repositories for grokfast
Users that are interested in grokfast are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- Explorations into the proposal from the paper "Grokfast, Accelerated Grokking by Amplifying Slow Gradients"☆104Dec 22, 2024Updated last year
- ☆136Aug 19, 2024Updated 2 years ago
- Code for NeurIPS'24 paper 'Grokked Transformers are Implicit Reasoners: A Mechanistic Journey to the Edge of Generalization'☆242Jul 19, 2025Updated last year
- ☆119Jul 23, 2025Updated last year
- Deep Networks Grok All the Time and Here is Why☆40Apr 20, 2026Updated 4 months ago
- Deploy on Railway without the complexity - Free Credits Offer • AdConnect your repo and Railway handles the rest with instant previews. Quickly provision container image services, databases, and storage volumes.
- Omnigrok: Grokking Beyond Algorithmic Data☆66Feb 24, 2023Updated 3 years ago
- Code to reproduce "Transformers Can Do Arithmetic with the Right Embeddings", McLeish et al (NeurIPS 2024)☆200May 28, 2024Updated 2 years ago
- The AdEMAMix Optimizer: Better, Faster, Older.☆188Sep 12, 2024Updated last year
- Implementation for MatMul-free LM.☆3,089Dec 2, 2025Updated 9 months ago
- Pretraining and inference code for a large-scale depth-recurrent language model☆935Dec 29, 2025Updated 8 months ago
- ☆28Feb 1, 2023Updated 3 years ago
- Schedule-Free Optimization in PyTorch☆2,325Jul 28, 2026Updated last month
- [ICLR 2025] Samba: Simple Hybrid State Space Models for Efficient Unlimited Context Language Modeling☆971Nov 16, 2025Updated 9 months ago
- GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection☆1,703Oct 28, 2024Updated last year
- GPU virtual machines on DigitalOcean Gradient AI • AdGet to production fast with high-performance AMD and NVIDIA GPUs you can spin up in seconds. The definition of operational simplicity.
- Quick implementation of nGPT, learning entirely on the hypersphere, from NvidiaAI☆300Jun 3, 2025Updated last year
- Code for Adam-mini: Use Fewer Learning Rates To Gain More https://arxiv.org/abs/2406.16793☆459May 13, 2025Updated last year
- Entropy Based Sampling and Parallel CoT Decoding☆3,432Nov 13, 2024Updated last year
- Code to reproduce key results accompanying "SAEs (usually) Transfer Between Base and Chat Models"☆13Jul 18, 2024Updated 2 years ago
- ☆314Jun 21, 2024Updated 2 years ago
- Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clusters☆135Dec 3, 2024Updated last year
- Memory layers use a trainable key-value lookup mechanism to add extra parameters to a model without increasing FLOPs. Conceptually, spars…☆379Dec 12, 2024Updated last year
- This repository contains the joint use of CPO and SimPO method for better reference-free preference learning methods.☆59Aug 13, 2024Updated 2 years ago
- Muon is an optimizer for hidden layers in neural networks☆2,823May 24, 2026Updated 3 months ago
- Virtual machines for every use case on DigitalOcean • AdGet dependable uptime with 99.99% SLA, simple security tools, and predictable monthly pricing with DigitalOcean's virtual machines, called Droplets.
- A pure and fast NumPy implementation of Mamba with cache support.☆18Jun 16, 2024Updated 2 years ago
- Convolutions for Sequence Modeling☆918Jun 13, 2024Updated 2 years ago
- DeMo: Decoupled Momentum Optimization☆202Dec 2, 2024Updated last year
- BitLinear implementation☆38Jul 8, 2026Updated last month
- Official implementation of Half-Quadratic Quantization (HQQ)☆958Feb 26, 2026Updated 6 months ago
- Code for BLT research paper☆2,060Nov 3, 2025Updated 10 months ago
- Stanford NLP Python library for Representation Finetuning (ReFT)☆1,580Mar 5, 2026Updated 6 months ago
- smol models are fun too☆94Nov 9, 2024Updated last year
- NanoGPT (124M) in 90 seconds☆5,758Aug 9, 2026Updated 3 weeks ago
- GPU virtual machines on DigitalOcean Gradient AI • AdGet to production fast with high-performance AMD and NVIDIA GPUs you can spin up in seconds. The definition of operational simplicity.
- The Gaussian Histogram Loss (HL-Gauss) proposed by Imani et al. with a few convenient wrappers for regression, in Pytorch☆81Updated this week
- Repository for Sparse Universal Transformers☆20Oct 23, 2023Updated 2 years ago
- Normalized Transformer (nGPT)☆213Nov 19, 2024Updated last year
- Kolmogorov Arnold Networks☆16,337Jan 19, 2025Updated last year
- Implementation of the training framework proposed in Self-Rewarding Language Model, from MetaAI☆1,410Apr 11, 2024Updated 2 years ago
- Minimal (400 LOC) implementation Maximum (multi-node, FSDP) GPT training☆132Apr 17, 2024Updated 2 years ago
- Official Code Repository for the paper "Key-value memory in the brain"☆34Feb 25, 2025Updated last year