DeepSeek FlashAttention2.forward: attn output

Read the equation

In DeepseekV3FlashAttention2.forward, this statement defines or updates attn_output. The displayed multiplication combines the source factors. For tensors this is elementwise unless the called operation explicitly performs a matrix contraction.

Terms & definitions

The recovered passage does not supply a separate symbol table. Read the source context for the notation used in this version; familiar letters may carry different meanings in other builds.

How it works

1. Read attn_output as the quantity being defined or updated by this relation.

2. Substitute the source-defined variables or recorded values and follow the displayed arithmetic and indexing exactly.

3. Use this version’s variable definitions and update order. The same symbol may be assigned a different meaning in another build.

Assumptions & limits

Source recovery establishes that the expression appears in this record. Its mathematical validity, implementation behavior, and interpretation must be assessed separately. See the linked source and research notes for the checks available for this version.

Original context

if self.q_head_dim != self.v_head_dim:
            attn_output = attn_output[:, :, :, :self.v_head_dim]

        attn_output = attn_output.reshape(bsz, q_len, self.num_heads *
                                          self.v_head_dim).contiguous()
        attn_output = self.o_proj(attn_output)

        if not output_attentions:

attn_output = attn_output.reshape(bsz, q_len, self.num_heads *
                                          self.v_head_dim).contiguous()

Related builds