flashrt/gated-delta-attention
• Updated • 18 • 3
AI & ML interests
training & inference infra
Recent Activity
flashrt/fp8-gemm
• Updated • 12 • 2
flashrt/transformer-fused-ops
• Updated • 48 • 1
flashrt/fused-mlp-megakernels-blackwell
• Updated
flashrt/audio-codebook-primitives
• Updated
flashrt/fp8-cross-attention-blackwell
Updated
flashrt/fp4-fused-ops
• Updated • 12 • 1
flashrt/fp8-kv-attention
• Updated • 13 • 1
flashrt/transformer-layout-primitives
• Updated • 9 • 2
flashrt/fp4-gemm
• Updated • 19 • 1
flashrt/flashrt-vla-residual-gates
• Updated • 16 • 3
flashrt/adaptive-layernorm-producers
• Updated • 11 • 1
flashrt/blockwise-fp8-producers
• Updated • 1
flashrt/world-model-conv
• Updated • 8 • 1
flashrt/diffusion-step-ops
• Updated • 12 • 1
flashrt/flashrt-spatiotemporal-layout
• Updated • 18 • 2
flashrt/flashrt-qkv-cache-rope
• Updated • 28 • 2
flashrt/padded-fp8-producers
• Updated • 1
flashrt/grouped-moe-gemm
• Updated • 2 • 2
flashrt/fp8-prefill-attention-blackwell
• Updated • 2 • 1
flashrt/small-matrix-cholesky
• Updated • 1 • 3
flashrt/smallm-ffn-megakernels-blackwell
• Updated • 1 • 1
flashrt/fa2-seqused-runtime
• Updated • 12 • 1
flashrt/flashrt-fp8-ffn
• Updated • 30 • 5
flashrt/flashrt-fp8-swiglu-ffn
• Updated • 45 • 5
flashrt/weight-only-ffn
• Updated • 7 • 1
flashrt/int4-blackwell
• Updated • 26 • 5
flashrt/flashrt-flex-attention-train
• Updated • 2
flashrt/flashrt-adarms-train
• Updated • 16 • 2
flashrt/flashrt-vocab-ce-train
• Updated • 12 • 1