csrc/README.md

# Fusion Operators (CSRC)

High-performance CUDA kernels for accelerating model training.

## 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 routing
- `unpermute`: Token recovery after expert computation
- `unpermute_bwd`: Backward pass for token recovery

### Multimodal RoPE
- `rope`: Rotary Position Embedding forward pass
- `rope_bwd`: RoPE backward pass
- Support for multimodal inputs with configurable sections

## Acknowledgments

The `permute` and `unpermute` operators are adapted from [fanshiqing/grouped_gemm](https://github.com/fanshiqing/grouped_gemm). Thanks for their open-source contributions.
Init 2025-09-07 14:59:17 +08:00			`# Fusion Operators (CSRC)`

			`High-performance CUDA kernels for accelerating model training.`

			`## 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 routing
[lint] Update lint (#16 ) 2025-09-11 13:18:33 +08:00			- `unpermute`: Token recovery after expert computation
Init 2025-09-07 14:59:17 +08:00			- `unpermute_bwd`: Backward pass for token recovery

			`### Multimodal RoPE`
			- `rope`: Rotary Position Embedding forward pass
			- `rope_bwd`: RoPE backward pass
			`- Support for multimodal inputs with configurable sections`

			`## Acknowledgments`

[lint] Update lint (#16 ) 2025-09-11 13:18:33 +08:00			The `permute` and `unpermute` operators are adapted from [fanshiqing/grouped_gemm](https://github.com/fanshiqing/grouped_gemm). Thanks for their open-source contributions.