Functionflash_bwd(
Q: T.Tensor(q_shape, dtype),
K: T.Tensor(kv_shape, dtype),
V: T.Tensor(kv_shape, dty
examples/attention_sink/example_gqa_sink_bwd_varlen.py:330
Functionflash_bwd_kernel(
Q: T.Tensor(q_shape, dtype),
K: T.Tensor(kv_shape, dtype),
V: T.Tensor(kv_shape, dty
examples/amd/example_amd_flash_attn_bwd.py:331
Functionflash_bwd_prep(O: T.Tensor(shape, dtype), dO: T.Tensor(shape, dtype), Delta: T.Tensor([batch, heads, seq_len], accum_dtype))
examples/amd/example_amd_flash_attn_bwd.py:289
Functionflashattn_bwd_pipeline(batch, heads, seq_len, dim, is_causal, block_M, block_N, groups=1)
examples/flash_attention_sm100/gqa_bwd_bshd.py:237
Functionflashattn_bwd_warp(batch, heads, seq_len, dim, is_causal, block_M, block_N, groups=1)
examples/flash_attention_sm100/gqa_bwd_bshd.py:252
Functionflashattn_gqa_decode_no_split(
Q: T.Tensor(shape_q, dtype),
K: T.Tensor(shape_k, dtype),
V: T.Tensor(shape_v, dtype
examples/flash_decoding/example_gqa_decode_varlen_logits.py:54
Functionflashattn_gqa_decode_no_split(
Q: T.Tensor(shape_q, dtype),
K: T.Tensor(shape_k, dtype),
V: T.Tensor(shape_v, dtype
examples/flash_decoding/example_gqa_decode.py:191
Functionflashattn_gqa_decode_split(
Q: T.Tensor(shape_q, dtype),
K: T.Tensor(shape_k, dtype),
V: T.Tensor(shape_v, dtype
examples/flash_decoding/example_gqa_decode.py:65
Functionflashattn_mha_inference(
Q: T.Tensor(shape_q, dtype),
K: T.Tensor(shape_kv, dtype),
V: T.Tensor(shape_kv, dty
examples/flash_decoding/example_mha_inference.py:21
Methodforward(ctx, q, k, v, block_sparse_dense, sm_scale)
benchmark/blocksparse_attention/benchmark_triton_block_sparse_fmha.py:240
Methodforward(ctx, q, k, v, block_indices, block_counts, block_size, window_size, scale, offsets)
examples/deepseek_nsa/example_triton_nsa_fwd_varlen.py:222
Methodforward(ctx, q, k, v, block_indices, block_counts, block_size, window_size, scale, offsets)
examples/deepseek_nsa/example_tilelang_nsa_fwd_varlen.py:203
Methodforward(
ctx,
q,
k,
v,
block_indices,
block_counts,
block_siz
examples/deepseek_nsa/example_tilelang_nsa_bwd.py:616
Methodforward(ctx, q, k, v, block_indices, block_size, scale, offsets)
examples/deepseek_nsa/example_triton_nsa_bwd.py:120
Methodforward(ctx, q, k, v, block_indices, block_size, scale, offsets)
examples/deepseek_nsa/example_triton_nsa_fwd.py:120
Methodforward(ctx, q, k, v, block_indices, block_size, scale, offsets)
examples/deepseek_nsa/benchmark/benchmark_nsa_fwd.py:114
Methodforward(self, x: torch.Tensor, qr: torch.Tensor, start_pos: int, freqs_cis: torch.Tensor,
mask: Optio
examples/deepseek_v32/inference/model.py:478
Methodforward(ctx, a, b, batch_sizes, block_M, block_N, block_K, num_stages, threads)
examples/grouped_gemm/example_grouped_gemm_bwd.py:59
Methodforward(ctx, q, k, v, causal, groups=1, use_atomic=True)
examples/flash_attention/example_gqa_bwd_tma_reduce.py:351
Methodforward(ctx, q, k, v, causal, groups=1, use_atomic=True)
examples/flash_attention/example_gqa_bwd.py:341
Methodforward(
ctx, q, k, v, seqlens_q, seqlens_k, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, causal,
examples/flash_attention/example_gqa_bwd_tma_reduce_varlen.py:489