Index

B

batch_deepgemm_fp8_nt_groupwise() (in module flashinfer.gemm) bmm() (in module mate.gemm) bmm_fp8() (in module flashinfer.gemm) built-in function mate.deep_gemm.fp8_fp8_mega_moe() mate.deep_gemm.get_symm_buffer_for_mega_moe() mate.deep_gemm.get_token_alignment_for_mega_moe() mate.deep_gemm.transform_weights_for_mega_moe()

F

flash_attn_combine() (in module mate.mha_interface) flash_attn_varlen_func() (in module mate.mha_interface) flash_attn_with_kvcache() (in module mate.mha_interface) flash_mla_sparse_fwd() (in module mate.flashmla) flash_mla_sparse_fwd_pack8() (in module mate.flashmla)

flash_mla_with_kvcache() (in module mate.flashmla) fp8_einsum() (in module mate.deep_gemm) fp8_gemm_nt_skip_head_mid() (in module mate.deep_gemm) fp8_mqa_logits() (in module mate.deep_gemm) fp8_paged_mqa_logits() (in module mate.deep_gemm)

G

gemm_fp8_nt_groupwise() (in module flashinfer.gemm) get_batch_decode_metadata_mla() (in module flashinfer.decode) get_mla_metadata() (in module mate.flashmla)

get_paged_mqa_logits_metadata() (in module mate.deep_gemm) get_scheduler_metadata() (in module mate.mha_interface) group_deepgemm_fp8_nt_groupwise() (in module flashinfer.gemm)

M

masked_moe_gemm_16bit() (in module mate.gemm) masked_moe_gemm_8bit() (in module mate.gemm) masked_moe_gemm_mixed_dtype() (in module mate.gemm) mate.deep_gemm.fp8_fp8_mega_moe() built-in function mate.deep_gemm.get_symm_buffer_for_mega_moe() built-in function mate.deep_gemm.get_token_alignment_for_mega_moe() built-in function mate.deep_gemm.transform_weights_for_mega_moe() built-in function

mhc_pre() (in module mate.hyperconnection) mhc_pre_big_fuse() (in module mate.hyperconnection) mhc_prenorm_gemm_sqrsum() (in module mate.hyperconnection) mla_rope_quantize_fp8() (in module mate.sparse_mla_interface) msa() (in module mate.msa_interface) msa_plan() (in module mate.msa_interface) MsaDecodePlan (class in mate.msa_interface) MsaPlan (class in mate.msa_interface) MsaPlanInfo (in module mate.msa_interface) MsaPrefillPlan (class in mate.msa_interface) MsaRuntimeMetadata (class in mate.msa_interface)

P

prewarm_selective_state_update() (in module mate.mamba)

prewarm_ssd_combined_fwd_varlen() (in module mate.mamba)

Q

qsa_indexer_mqa_score() (in module mate) qsa_mqa() (in module mate)

qsa_mqa_decode() (in module mate) qsa_mqa_prefill() (in module mate) qsa_sparse_attention() (in module mate)

R

ragged_k_moe_gemm_16bit() (in module mate.gemm) ragged_k_moe_gemm_8bit() (in module mate.gemm)

ragged_m_moe_gemm_16bit() (in module mate.gemm) ragged_m_moe_gemm_8bit() (in module mate.gemm) ragged_moe_gemm_mixed_dtype() (in module mate.gemm)

S

selective_state_update() (in module mate.mamba) sparse_decode_atten_func() (in module mate.msa_interface) sparse_mla_fp8_decode() (in module mate.sparse_mla_interface) sparse_mla_nope_decode() (in module mate.sparse_mla.glm_nope)

sparse_msa() (in module mate.msa_interface) sparse_msa_plan() (in module mate.msa_interface) sparse_topk_select() (in module mate.msa_interface) ssd_combined_fwd_varlen() (in module mate.mamba)

T

tf32_hc_prenorm_gemm() (in module mate.deep_gemm)

trtllm_batch_decode_with_kv_cache_mla() (in module flashinfer.decode)