Liang Su

liangsu9988

·

AI & ML interests

None yet

Recent Activity

new activity 15 days ago

flashrt/gated-delta-attention:support for GB10 (DGX Spark)

updated a Space 22 days ago

liangsu9988/fast-kernels-are-not-fast-pipelines

updated a Space about 1 month ago

liangsu9988/the-context-dividend

View all activity

Organizations

New activity in flashrt/gated-delta-attention 15 days ago

support for GB10 (DGX Spark)

#1 opened 16 days ago by

updated a Space 22 days ago

Fast Kernels Are Not Fast Pipelines

Explore interactive galaxy visualization with hover info

updated a Space about 1 month ago

The Context Dividend

Explore how hot‑plugging extends LLM context on a GPU

published a Space about 1 month ago

The Context Dividend

Explore how hot‑plugging extends LLM context on a GPU

updated 16 models about 1 month ago

flashrt/fp4-gemm

flashrt/weight-only-ffn

flashrt/flashrt-gemm-epilogues

flashrt/flashrt-fp8-ffn

flashrt/transformer-fused-ops

flashrt/fa2-seqused-runtime

flashrt/turboquant-kv

flashrt/speculative-draft-primitives

flashrt/small-matrix-cholesky

flashrt/padded-fp8-producers

flashrt/linear-attention-seq-state

flashrt/linear-attention-primitives

flashrt/fused-mlp-megakernels-blackwell

flashrt/fp8-prefill-attention-blackwell

Updated Aug 17 • 1

flashrt/flashrt-vla-video

flashrt/flashrt-flex-attention-train