ݘh ddlZddlmZddlmZ erddlmZn #e$rYnwxYw ddejj dej dej d ej d ej d ed ej fd Z dS)N)PagedAttentionCache)is_flash_attn_2_available)flash_attn_varlen_funcmoduleqkvattention_maskcachereturnc |j|||jfi| \}}t|ddsdn|jdf} | | j}d| vrd| dini}||ddd|ddd|ddd| tj | tj  || f|j d | d |}t|tr|d}|dfS) aPerform the forward pass of attention with paged key-value cache. This function handles the cache updates and performs the attention computation using the flash_attn_varlen_func for efficient processing. Args: q: (total_q, nheads, headdim), where total_q = total number of query tokens in the batch. k: (total_k, nheads_k, headdim), where total_k = total number of key tokens in the batch. but if there is a block table it can be the full k v: (total_k, nheads_k, headdim), where total_k = total number of key tokens in the batch. but if there is a block table it can be the full v cu_seq_lens_q: (batch_size + 1,), dtype torch.int32. The cumulative sequence lengths of the sequences in the batch, used to index into q. cu_seq_lens_k: (batch_size + 1,), dtype torch.int32. The cumulative sequence lengths of the sequences in the batch, used to index into kv. max_seqlen_q: int. Maximum query sequence length in the batch. max_seqlen_k: int. Maximum key sequence length in the batch. dropout_p: float. Dropout probability. softmax_scale: float. The scaling of QK^T before applying softmax. Default to 1 / sqrt(headdim). causal: bool. Whether to apply causal attention mask (e.g., for auto-regressive modeling). window_size: (left, right). If not (-1, -1), implements sliding window local attention. softcap: float. Anything > 0 activates softcapping attention. sliding_windowF)rrNs_auxrT) softmax_scalecausal window_size)update layer_idxgetattrrrget transposesqueeze contiguoustotorchint32clonescaling isinstancetuple)rrr r r r cu_seq_lens_q cu_seq_lens_k max_seqlen_q max_seqlen_k block_tablesimplementationkwargsrr custom_kwargs attn_outputs m/var/www/html/services/rag-service/venv/lib/python3.11/site-packages/transformers/integrations/flash_paged.pypaged_attention_forwardr.sJ 5<1f. 9 9& 9 9DAq%,V5Eu%M%MmXXTZTiklSmN!!/!F6=6G6GWfjj1122RM(( Aq!!!$$//11 Aq!!!$$//11 Aq!!!$$//11%%%%++-- n"     K+u%%%!!n  )NNNNNNNN) rgeneration.continuous_batchingrutilsr flash_attnr ExceptionnnModuleTensorr.r/r-r8s @@@@@@------   ""6555555   D $(!%;; HO; |; |; | ; L ;  ; \;;;;;;s #++