NX-AI / xlstm-jaxLinks
Official JAX implementation of xLSTM including fast and efficient training and inference code. 7B model available at https://huggingface.co/NX-AI/xLSTM-7b.
โ98Updated 6 months ago
Alternatives and similar repositories for xlstm-jax
Users that are interested in xlstm-jax are comparing it to the libraries listed below
Sorting:
- โ206Updated 8 months ago
- ๐งฑ Modula software packageโ216Updated last week
- Dion optimizer algorithmโ193Updated this week
- Cost aware hyperparameter tuning algorithmโ166Updated last year
- โ275Updated last year
- The simplest, fastest repository for training/finetuning medium-sized GPTs.โ149Updated last month
- The AdEMAMix Optimizer: Better, Faster, Older.โ184Updated 10 months ago
- A State-Space Model with Rational Transfer Function Representation.โ79Updated last year
- Efficient optimizersโ252Updated last week
- โ298Updated 6 months ago
- Simple, minimal implementation of the Mamba SSM in one pytorch file. Using logcumsumexp (Heisen sequence).โ120Updated 9 months ago
- Getting crystal-like representations with harmonic lossโ192Updated 4 months ago
- Official repository for the paper "Grokfast: Accelerated Grokking by Amplifying Slow Gradients"โ559Updated last year
- A MAD laboratory to improve AI architecture designs ๐งชโ123Updated 7 months ago
- โ224Updated last month
- Latent Program Network (from the "Searching Latent Program Spaces" paper)โ93Updated 4 months ago
- Annotated version of the Mamba paperโ487Updated last year
- Quick implementation of nGPT, learning entirely on the hypersphere, from NvidiaAIโ289Updated 2 months ago
- Understand and test language model architectures on synthetic tasks.โ221Updated 3 weeks ago
- supporting pytorch FSDP for optimizersโ84Updated 7 months ago
- Evaluating the Mamba architecture on the Othello gameโ48Updated last year
- Tiled Flash Linear Attention library for fast and efficient mLSTM Kernels.โ65Updated last week
- Accelerated First Order Parallel Associative Scanโ184Updated 11 months ago
- โ347Updated this week
- Normalized Transformer (nGPT)โ185Updated 8 months ago
- โ100Updated 2 weeks ago
- โ115Updated last month
- โ65Updated 8 months ago
- CIFAR-10 speedruns: 94% in 2.6 seconds and 96% in 27 secondsโ274Updated 2 weeks ago
- Library for text-to-text regression, applicable to any input string representation and allows pretraining and fine-tuning over multiple rโฆโ122Updated this week