Pinned Loading

1. deepseek-ai/profile-data deepseek-ai/profile-data Public

   Analyze computation-communication overlap in V3/R1.

   1.2k 156

2. deepseek-ai/DeepGEMM deepseek-ai/DeepGEMM Public

   DeepGEMM: clean and efficient BLAS kernel library on GPU

   Cuda 9.1k 1.5k

3. deepseek-ai/DualPipe deepseek-ai/DualPipe Public

   A bidirectional pipeline parallelism algorithm for computation-communication overlap in DeepSeek V3/R1 training.

   Python 3k 340