automl / unlocking_state_trackingLinks
Expanding linear RNN state-transition matrix eigenvalues to include negatives improves state-tracking tasks and language modeling without added training or inference costs.
☆15Updated 4 months ago
Alternatives and similar repositories for unlocking_state_tracking
Users that are interested in unlocking_state_tracking are comparing it to the libraries listed below
Sorting:
- Experiments on the impact of depth in transformers and SSMs.☆32Updated 8 months ago
- ☆49Updated last year
- Efficient PScan implementation in PyTorch☆16Updated last year
- ☆56Updated last year
- Official repository of paper "RNNs Are Not Transformers (Yet): The Key Bottleneck on In-context Retrieval"☆27Updated last year
- Here we will test various linear attention designs.☆62Updated last year
- Official Code Repository for the paper "Key-value memory in the brain"☆27Updated 5 months ago
- Reference implementation of "Softmax Attention with Constant Cost per Token" (Heinsen, 2024)☆24Updated last year
- Xmixers: A collection of SOTA efficient token/channel mixers☆11Updated 3 weeks ago
- Parallel Associative Scan for Language Models☆18Updated last year
- Official code for the paper "Attention as a Hypernetwork"☆40Updated last year
- Stick-breaking attention☆58Updated last month
- ☆32Updated last year
- ☆20Updated last year
- ☆37Updated last year
- ☆38Updated 3 months ago
- ☆33Updated last year
- [NeurIPS 2023 spotlight] Official implementation of HGRN in our NeurIPS 2023 paper - Hierarchically Gated Recurrent Neural Network for Se…☆66Updated last year
- ☆11Updated last year
- Combining SOAP and MUON☆16Updated 5 months ago
- Open-sourcing code associated with the AAAI-25 paper "On the Expressiveness and Length Generalization of Selective State-Space Models on …☆14Updated 2 months ago
- ☆53Updated last year
- Code for "Theoretical Foundations of Deep Selective State-Space Models" (NeurIPS 2024)☆15Updated 6 months ago
- ☆32Updated 9 months ago
- ☆24Updated last week
- ☆23Updated 10 months ago
- Blog post☆17Updated last year
- Using FlexAttention to compute attention with different masking patterns☆44Updated 10 months ago
- HGRN2: Gated Linear RNNs with State Expansion☆55Updated 11 months ago
- ☆27Updated last year