Home
last modified time | relevance | path

Searched defs:max_seqlen_batch_q (Results 1 – 4 of 4) sorted by relevance

/aosp_15_r20/external/pytorch/aten/src/ATen/native/transformers/cuda/
H A Dattention_backward.cu70 int64_t max_seqlen_batch_q, in _flash_attention_backward()
198 const int64_t max_seqlen_batch_q = query.size(2); in _scaled_dot_product_cudnn_attention_backward_cuda() local
741 const int64_t max_seqlen_batch_q, in _scaled_dot_product_flash_attention_backward_cuda()
H A Dattention.cu698 const int64_t max_seqlen_batch_q = query.size(2); in _scaled_dot_product_flash_attention_cuda() local
753 const int64_t max_seqlen_batch_q = query.size(2); in _scaled_dot_product_cudnn_attention_cuda() local
849 int64_t max_seqlen_batch_q, in _flash_attention_forward()
/aosp_15_r20/external/pytorch/aten/src/ATen/native/nested/cuda/
H A DNestedTensorTransformerUtils.cpp250 int64_t max_seqlen_batch_q = 0, Nnz_q = 0; in sdpa_nested_preprocessing_with_broadcast() local
466 const int64_t max_seqlen_batch_q, in sdpa_nested_preprocessing_backward()
H A DNestedTensorTransformerFunctions.cpp335 const int64_t max_seqlen_batch_q, in _scaled_dot_product_flash_attention_backward_nested()