taxonomy node · level 2
kernel & quantization library
18 methods filed at this node or below it, from the sources of 10 models.
software implementation :: kernel & quantization library
Matching aids for the classifier: FlashInfer; AITER; CUTEDSL; TileLang; MiniTriton; AngelSlim; ModelOpt; Marlin kernels.
In this branch 18
Everything filed at this node or below it, with one collapsible heading per child node.
filed here 18
By model
Which of this branch's techniques each model's own documents describe, and how strongly. Under each model: its strongest treatment anywhere in the branch.
| Model | techniques |
|---|---|
| DeepSeek-V4.1-Flash optional | deepseek-recipe optional— |
| Hy4-preview used | AngelSlim usedFLASHMLA_SPARSE used— |
| NVIDIA-Nemotron-3-Ultra-550B-A55B used | Container-baked precompiled artifacts usedFlashInfer usedFlashInfer NVLinkOneSided usedNVIDIA Model-Optimizer usedCUTEDSL optionalCUTLASS optional— |
| GLM-5.3 used | AITER used— |
| Hy3 mentioned | AngelSlim mentioned— |
| MiniMax-M3 evaluated | flash-moba evaluatedFlash-Sparse-Attention evaluated— |
| Kimi K3 used | MiniTriton usedMiniTriton dual-mode tensor library usedTriton mentioned— |
| Qwen3.6-35B-A3B used | Marlin MoE backend used— |
| Qwen3.8-Flash-Next used | FlashQLA used— |
| Step-3.7-Flash used | FlashInfer TensorRT-LLM backend usedFlashInfer optional— |