flashrt/fp4-gemm
• Updated • 2 • 3
AI & ML interests
training & inference infra
flashrt/weight-only-ffn
• Updated • 435 • 1
flashrt/gated-delta-attention
• Updated • 216 • 3
flashrt/transformer-fused-ops
• Updated • 878 • 1
flashrt/causal-conv1d-state
• Updated • 127 • 2
flashrt/fp4-fused-ops
• Updated • 167 • 1
flashrt/sageattention3-blackwell
• Updated • 557 • 2
flashrt/sageattention2-blackwell
• Updated • 705 • 1
flashrt/fp8-gemm
• Updated • 27 • 2
flashrt/masked-mha-runtime
• Updated • 323 • 2
flashrt/flashrt-qkv-cache-rope
• Updated • 18 • 2
flashrt/flashrt-fp8-ffn
• Updated • 1.23k • 5
flashrt/flashrt-residual-norm-quant
• Updated • 731 • 2
flashrt/fa4-cute-runtime
• Updated • 1
flashrt/grouped-moe-gemm
• Updated • 151 • 2
flashrt/fp8-kv-attention
• Updated • 4 • 1
flashrt/grouped-moe-gemv
• Updated • 2 • 3
flashrt/world-model-conv
• Updated • 2 • 1
flashrt/smallm-ffn-megakernels-blackwell
• Updated • 288 • 1
flashrt/flashrt-nvfp4
• Updated • 209 • 3
flashrt/flashrt-fused-quant
• Updated • 265 • 3
flashrt/adaptive-layernorm-producers
• Updated • 734 • 1
flashrt/bf16-linear-gemv
• Updated • 181 • 1
flashrt/transformer-layout-primitives
• Updated • 332 • 2
flashrt/fp8-cross-attention-blackwell
• Updated • 149 • 1
flashrt/fused-mlp-megakernels-blackwell
• Updated • 96 • 2
flashrt/audio-codebook-primitives
• Updated • 31 • 2
flashrt/flashrt-vla-residual-gates
• Updated • 10 • 4
flashrt/blockwise-fp8-producers
• Updated • 1 • 1
flashrt/diffusion-step-ops
• Updated • 329 • 1