Starrick Liu
421db17d53
This release introduces significant performance optimizations, memory efficiency improvements, and enhanced system robustness: 🚀 Performance Optimizations: - Add three new fused CUDA kernels (rope_index, rot_pos_emb, get_window_index) for accelerated multimodal preprocessing - Implement FSDP2 support for distributed training with improved memory efficiency - Add Torch.compile integration for additional performance gains - Optimize memory usage: reduce peak allocation from 48GB to 24GB on 8-GPU setup 🔧 System Robustness: - Fix missing token position inputs in prediction pipeline - Add type-robust negation operations in RoPE CUDA kernels (half/bfloat16 support) - Fix dataset root parameter initialization in LeRobot data loader - Enhanced error handling and input validation across fusion operators 📚 Documentation & Usability: - Add comprehensive memory usage benchmarks and hardware recommendations - Update citation format with proper arXiv reference - Improve training configuration documentation with quick start guide - Add detailed API documentation for new fusion operators 🛠️ Technical Details: - Version bump to 1.0.1 - New CUDA kernels: rope_index.cu, rot_pos.cu, window_index.cu - FSDP2 state dict loading with distribute_tensor support - Enhanced multimodal RoPE with 3D position encoding - Window attention optimization for Vision Transformers Breaking Changes: None - all changes are backward compatible
Fusion Operators (CSRC)
High-performance CUDA kernels for accelerating model training, with specialized support for multimodal and MoE architectures.
Operators
Asymmetric Dual Expert GEMM
asym_dual_gmm: Simultaneous matrix multiplication for two experts- Supports all transpose combinations (NN, TN, NT, TT)
Token Permutation
permute: Token permutation for MoE routingunpermute: Token recovery after expert computationunpermute_bwd: Backward pass for token recovery
Multimodal RoPE
rope: Rotary Position Embedding forward passrope_bwd: RoPE backward passrope_index: Generates position indices for multimodal RoPErot_pos_emb: Fused rotary position embedding computation
Vision Transformer Optimization
get_window_index: Window attention index generation
Acknowledgments
The permute and unpermute operators are adapted from fanshiqing/grouped_gemm. Thanks for their open-source contributions.