lucidrains / grokfast-pytorchLinks
Explorations into the proposal from the paper "Grokfast, Accelerated Grokking by Amplifying Slow Gradients"
☆103Updated 11 months ago
Alternatives and similar repositories for grokfast-pytorch
Users that are interested in grokfast-pytorch are comparing it to the libraries listed below
Sorting:
- ☆82Updated last year
- Implementation of GateLoop Transformer in Pytorch and Jax☆91Updated last year
- Collection of autoregressive model implementation☆86Updated 7 months ago
- Pytorch implementation of the PEER block from the paper, Mixture of A Million Experts, by Xu Owen He at Deepmind☆131Updated last month
- Train a SmolLM-style llm on fineweb-edu in JAX/Flax with an assortment of optimizers.☆18Updated 4 months ago
- ☆53Updated last year
- Explorations into the recently proposed Taylor Series Linear Attention☆100Updated last year
- ☆105Updated 4 months ago
- Implementation of the Llama architecture with RLHF + Q-learning☆168Updated 10 months ago
- Implementation of Gradient Agreement Filtering, from Chaubard et al. of Stanford, but for single machine microbatches, in Pytorch☆25Updated 10 months ago
- ☆21Updated last year
- Tiled Flash Linear Attention library for fast and efficient mLSTM Kernels.☆74Updated last week
- A byte-level decoder architecture that matches the performance of tokenized Transformers.☆66Updated last year
- Implementation of Infini-Transformer in Pytorch☆113Updated 11 months ago
- Unofficial but Efficient Implementation of "Mamba: Linear-Time Sequence Modeling with Selective State Spaces" in JAX☆92Updated last year
- DeMo: Decoupled Momentum Optimization☆197Updated last year
- σ-GPT: A New Approach to Autoregressive Models☆70Updated last year
- Just some miscellaneous utility functions / decorators / modules related to Pytorch and Accelerate to help speed up implementation of new…☆125Updated last year
- Some personal experiments around routing tokens to different autoregressive attention, akin to mixture-of-experts☆119Updated last year
- Exploration into the proposed "Self Reasoning Tokens" by Felipe Bonetto☆57Updated last year
- H-Net Dynamic Hierarchical Architecture☆80Updated 2 months ago
- LayerNorm(SmallInit(Embedding)) in a Transformer to improve convergence☆61Updated 3 years ago
- Exploring an idea where one forgets about efficiency and carries out attention across each edge of the nodes (tokens)☆55Updated 8 months ago
- Minimal (400 LOC) implementation Maximum (multi-node, FSDP) GPT training☆132Updated last year
- Yet another random morning idea to be quickly tried and architecture shared if it works; to allow the transformer to pause for any amount…☆53Updated 2 years ago
- ☆61Updated last year
- Getting crystal-like representations with harmonic loss☆192Updated 8 months ago
- Griffin MQA + Hawk Linear RNN Hybrid☆89Updated last year
- PyTorch implementation of models from the Zamba2 series.☆186Updated 10 months ago
- ☆50Updated last year