glassroom / heinsen_attentionLinks
Reference implementation of "Softmax Attention with Constant Cost per Token" (Heinsen, 2024)
☆24Updated last year
Alternatives and similar repositories for heinsen_attention
Users that are interested in heinsen_attention are comparing it to the libraries listed below
Sorting:
- ☆11Updated last year
- Here we will test various linear attention designs.☆62Updated last year
- ☆23Updated 11 months ago
- Official repository of paper "RNNs Are Not Transformers (Yet): The Key Bottleneck on In-context Retrieval"☆27Updated last year
- ☆49Updated last year
- ☆33Updated last year
- Official code for the paper "Attention as a Hypernetwork"☆40Updated last year
- Xmixers: A collection of SOTA efficient token/channel mixers☆11Updated last month
- HGRN2: Gated Linear RNNs with State Expansion☆55Updated last year
- ☆20Updated last year
- ☆32Updated last year
- ☆19Updated 3 months ago
- Efficient PScan implementation in PyTorch☆16Updated last year
- Griffin MQA + Hawk Linear RNN Hybrid☆88Updated last year
- 32 times longer context window than vanilla Transformers and up to 4 times longer than memory efficient Transformers.☆48Updated 2 years ago
- RADLADS training code☆27Updated 3 months ago
- Experiments on the impact of depth in transformers and SSMs.☆33Updated 9 months ago
- ☆56Updated last year
- [NeurIPS 2023 spotlight] Official implementation of HGRN in our NeurIPS 2023 paper - Hierarchically Gated Recurrent Neural Network for Se…☆66Updated last year
- Official Code Repository for the paper "Key-value memory in the brain"☆28Updated 6 months ago
- Awesome Triton Resources☆33Updated 3 months ago
- Combining SOAP and MUON☆15Updated 6 months ago
- Engineering the state of RNN language models (Mamba, RWKV, etc.)☆32Updated last year
- ☆106Updated last year
- sigma-MoE layer☆20Updated last year
- Using FlexAttention to compute attention with different masking patterns☆44Updated 11 months ago
- Triton implement of bi-directional (non-causal) linear attention☆51Updated 6 months ago
- ☆40Updated 4 months ago
- Expanding linear RNN state-transition matrix eigenvalues to include negatives improves state-tracking tasks and language modeling without…☆16Updated 5 months ago
- ☆24Updated 3 weeks ago