RobertCsordas / moe_layerLinks
sigma-MoE layer
☆20Updated last year
Alternatives and similar repositories for moe_layer
Users that are interested in moe_layer are comparing it to the libraries listed below
Sorting:
- Official repository for the paper "Approximating Two-Layer Feedforward Networks for Efficient Transformers"☆38Updated 3 months ago
- Fine-Tuning Pre-trained Transformers into Decaying Fast Weights☆19Updated 3 years ago
- ☆31Updated last year
- ☆48Updated last year
- Reference implementation of "Softmax Attention with Constant Cost per Token" (Heinsen, 2024)☆24Updated last year
- ☆56Updated last year
- [NeurIPS 2023] Sparse Modular Activation for Efficient Sequence Modeling☆40Updated last year
- ☆20Updated last year
- [NeurIPS 2023 spotlight] Official implementation of HGRN in our NeurIPS 2023 paper - Hierarchically Gated Recurrent Neural Network for Se…☆65Updated last year
- RWKV model implementation☆38Updated 2 years ago
- Here we will test various linear attention designs.☆61Updated last year
- Engineering the state of RNN language models (Mamba, RWKV, etc.)☆32Updated last year
- Official Repository for Efficient Linear-Time Attention Transformers.☆18Updated last year
- My explorations into editing the knowledge and memories of an attention network☆35Updated 2 years ago
- Official repository for the paper "SwitchHead: Accelerating Transformers with Mixture-of-Experts Attention"☆99Updated last year
- [EMNLP 2023] Official implementation of the algorithm ETSC: Exact Toeplitz-to-SSM Conversion our EMNLP 2023 paper - Accelerating Toeplitz…☆14Updated last year
- Using FlexAttention to compute attention with different masking patterns☆45Updated last year
- Triton Implementation of HyperAttention Algorithm☆48Updated last year
- HomebrewNLP in JAX flavour for maintable TPU-Training☆50Updated last year
- Combining SOAP and MUON☆16Updated 7 months ago
- Blog post☆17Updated last year
- Implementation of Token Shift GPT - An autoregressive model that solely relies on shifting the sequence space for mixing☆50Updated 3 years ago
- ☆29Updated 3 years ago
- A fusion of a linear layer and a cross entropy loss, written for pytorch in triton.☆70Updated last year
- 32 times longer context window than vanilla Transformers and up to 4 times longer than memory efficient Transformers.☆49Updated 2 years ago
- ☆53Updated 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 last year
- Official repository of paper "RNNs Are Not Transformers (Yet): The Key Bottleneck on In-context Retrieval"☆27Updated last year
- [NeurIPS 2022] Your Transformer May Not be as Powerful as You Expect (official implementation)☆32Updated 2 years ago
- Code for the paper "The Impact of Positional Encoding on Length Generalization in Transformers", NeurIPS 2023☆135Updated last year