Datasets:

humanfia-lab
/

inference-kernels

Size:

Tags:

Dataset card Files Files and versions

inference-kernels / kda

15.9 kB

Ctrl+K

Ctrl+K

2 contributors

History: 3 commits

ubospica's picture

Keep per-shape definitions and complete mask-free attention curation

f08a805 2 days ago

dense_gemm_bf16_nt_n12288_k4096.json

741 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
dense_gemm_bf16_nt_n4096_k12288.json

741 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
dense_gemm_bf16_nt_n4096_k4096.json

694 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
grouped_gemm_fp8_fp4_e384_n4608_k5120_g32.json

927 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
grouped_gemm_fp8_fp4_e384_n5120_k2304_g32.json

927 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
mixed_mla_decode_deepseek41_h64_d512_k8128_k4512.json

895 Bytes
Keep only kernels real inference uses; add SGLang baselines and KDA configs 3 days ago
moe_router_gemm_topk_deepseek41_d5120_e384_topk6.json

943 Bytes
Keep only kernels real inference uses; add SGLang baselines and KDA configs 3 days ago
neighborhood_attention_3d_h16_d64_kt3_ky7_kx7.json

793 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
neighborhood_attention_3d_h32_d64_kt3_ky7_kx7.json

793 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
neighborhood_attention_3d_h4_d64_kt11_ky11_kx11.json

755 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
neighborhood_attention_3d_h8_d64_kt3_ky5_kx5.json

966 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
noncausal_dense_attention_h32_d128.json

918 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
noncausal_dense_attention_h32_d64.json

915 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
prefix_cached_block_causal_h32_d128.json

720 Bytes
Keep only kernels real inference uses; add SGLang baselines and KDA configs 3 days ago
prefix_cached_block_causal_h32_d128_prefix0.json

700 Bytes
Keep only kernels real inference uses; add SGLang baselines and KDA configs 3 days ago
qsa_index_topk_decode_qwen38_h4_d128_topk512_c4.json

932 Bytes
Keep only kernels real inference uses; add SGLang baselines and KDA configs 3 days ago
qsa_sparse_gqa_decode_qwen38_h24_kv2_d256_topk512_c4.json

965 Bytes
Keep only kernels real inference uses; add SGLang baselines and KDA configs 3 days ago
sparse_mla_prefill_h64_d512_k128.json

718 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago
sparse_mla_prefill_h64_d512_k640.json

894 Bytes
Keep per-shape definitions and complete mask-free attention curation 2 days ago