flashrt/grouped-moe-gemv
• Updated • 13 • 2
AI & ML interests
training & inference infra
Recent Activity
flashrt/gated-delta-attention
• Updated • 33 • 3
flashrt/fp4-gemm
• Updated • 19 • 1
flashrt/world-model-conv
• Updated • 16 • 1
flashrt/smallm-ffn-megakernels-blackwell
• Updated • 19 • 1
flashrt/flashrt-nvfp4
• Updated • 22 • 3
flashrt/flashrt-fused-quant
• Updated • 20 • 3
flashrt/adaptive-layernorm-producers
• Updated • 39 • 1
flashrt/weight-only-ffn
• Updated • 15 • 1
flashrt/fp4-fused-ops
• Updated • 17 • 1
flashrt/fp8-gemm
• Updated • 14 • 2
flashrt/flashrt-qkv-cache-rope
• Updated • 31 • 2
flashrt/bf16-linear-gemv
• Updated • 23 • 1
flashrt/fp8-kv-attention
• Updated • 20 • 1
flashrt/transformer-layout-primitives
• Updated • 13 • 2
flashrt/fp8-cross-attention-blackwell
• Updated • 4 • 1
flashrt/transformer-fused-ops
• Updated • 52 • 1
flashrt/fused-mlp-megakernels-blackwell
• Updated • 5 • 2
flashrt/audio-codebook-primitives
• Updated • 1 • 2
flashrt/flashrt-vla-residual-gates
• Updated • 20 • 3
flashrt/blockwise-fp8-producers
• Updated • 3 • 1
flashrt/diffusion-step-ops
• Updated • 16 • 1
flashrt/flashrt-spatiotemporal-layout
• Updated • 21 • 2
flashrt/padded-fp8-producers
• Updated • 3 • 1
flashrt/grouped-moe-gemm
• Updated • 2 • 2
flashrt/fp8-prefill-attention-blackwell
• Updated • 3 • 1
flashrt/small-matrix-cholesky
• Updated • 1 • 3
flashrt/fa2-seqused-runtime
• Updated • 13 • 1
flashrt/flashrt-fp8-ffn
• Updated • 36 • 5
flashrt/flashrt-fp8-swiglu-ffn
• Updated • 49 • 5