MuLabPKU / TransMLALinks
TransMLA: Multi-Head Latent Attention Is All You Need (NeurIPS 2025 Spotlight)
☆426Updated 3 months ago
Alternatives and similar repositories for TransMLA
Users that are interested in TransMLA are comparing it to the libraries listed below
Sorting:
- Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs☆200Updated last month
- Lightning Attention-2: A Free Lunch for Handling Unlimited Sequence Lengths in Large Language Models☆339Updated 10 months ago
- 🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"☆956Updated 10 months ago
- [ICLR 2025] DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads☆519Updated 11 months ago
- [NeurIPS 2025 Spotlight] TPA: Tensor ProducT ATTenTion Transformer (T6) (https://arxiv.org/abs/2501.06425)☆445Updated last week
- Official implementation of "Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding"☆790Updated last month
- Parallel Scaling Law for Language Model — Beyond Parameter and Inference Time Scaling☆467Updated 8 months ago
- 🔥 A minimal training framework for scaling FLA models☆335Updated 2 months ago
- DeepSeek Native Sparse Attention pytorch implementation☆111Updated last month
- Efficient LLM Inference over Long Sequences☆393Updated 6 months ago
- SeerAttention: Learning Intrinsic Sparse Attention in Your LLMs☆187Updated 3 months ago
- Efficient triton implementation of Native Sparse Attention.☆260Updated 7 months ago
- Speed Always Wins: A Survey on Efficient Architectures for Large Language Models☆389Updated 2 months ago
- [ICLR 2025] COAT: Compressing Optimizer States and Activation for Memory-Efficient FP8 Training☆257Updated 5 months ago
- Implementation of the sparse attention pattern proposed by the Deepseek team in their "Native Sparse Attention" paper☆793Updated 5 months ago
- Chain of Experts (CoE) enables communication between experts within Mixture-of-Experts (MoE) models☆227Updated 2 months ago
- Official codebase for "Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling".☆280Updated 11 months ago
- qwen-nsa☆87Updated 3 months ago
- Implementation of FlashAttention in PyTorch☆180Updated last year
- ☆209Updated 2 months ago
- [ICLR 2025] Official PyTorch Implementation of Gated Delta Networks: Improving Mamba2 with Delta Rule☆429Updated 4 months ago
- Implementation for FP8/INT8 Rollout for RL training without performence drop.☆283Updated 2 months ago
- [NeurIPS 24 Spotlight] MaskLLM: Learnable Semi-structured Sparsity for Large Language Models☆183Updated last year
- [ICML 2025] XAttention: Block Sparse Attention with Antidiagonal Scoring☆265Updated 6 months ago
- Efficient Mixture of Experts for LLM Paper List☆156Updated 3 months ago
- A family of compressed models obtained via pruning and knowledge distillation☆363Updated 2 months ago
- [CoLM'25] The official implementation of the paper <MoA: Mixture of Sparse Attention for Automatic Large Language Model Compression>☆153Updated last week
- Unofficial implementation for the paper "Mixture-of-Depths: Dynamically allocating compute in transformer-based language models"☆176Updated last year
- Training library for Megatron-based models with bi-directional Hugging Face conversion capability☆363Updated this week
- Code for paper: [ICLR2025 Oral] FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference☆160Updated 3 months ago