deepseek-ai/FlashMLA
MLA decoding kernel for Hopper GPUs
View on index · View in 3D Map
Related repos
- DeepEP — communication library for Mixture-of-Experts and expert parallelism, with all-to-all GPU kernels and low-precision support
- DeepSeek-R1 GitHub repo with model code and info
- Not new, but still interesting
- FP8 matrix multiplication library for NVIDIA Hopper, supports dense and MoE GEMMs
- n8n node for DeepSeek AI automations
- DeepClaude: Combines DeepSeek R1 and Claude 3.5 for AI reasoning and coding
- Toolkit for optimizing and deploying deep learning models across CPUs, GPUs, and NPUs
- DeepSeek-V3, a Mixture-of-Experts language model with 671B parameters and 37B active per token, focused on math and coding tasks
- LLM API and chat UI combining DeepSeek R1 reasoning traces with Claude models
- unified multimodal model for both understanding and generating images and text
