| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566 |
- import torch
- from ..generation.continuous_batching.cache import PagedAttentionCache
- def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
- """
- This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
- num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
- """
- batch, num_key_value_heads, slen, head_dim = hidden_states.shape
- if n_rep == 1:
- return hidden_states
- hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
- return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
- def sdpa_attention_paged_forward(
- module: torch.nn.Module,
- query: torch.Tensor,
- key: torch.Tensor,
- value: torch.Tensor,
- attention_mask: torch.Tensor | None,
- dropout: float = 0.0,
- scaling: float | None = None,
- **kwargs,
- ) -> tuple[torch.Tensor, None]:
- # Add KV cache to the key and value tensors
- cache: PagedAttentionCache | None = kwargs.pop("cache", None)
- if cache is not None:
- # This changes the shape of k and v from [1, num_kv_heads, seqlen_kv, head_dim] to [-1, num_kv_heads, head_dim]
- key, value = cache.update(
- key_states=key,
- value_states=value,
- layer_idx=module.layer_idx,
- read_index=kwargs["read_index"],
- write_index=kwargs["write_index"],
- )
- key = key.transpose(0, 1).unsqueeze(0)
- value = value.transpose(0, 1).unsqueeze(0)
- # Repeat the key and value tensors for each group of key-value heads
- if hasattr(module, "num_key_value_groups"):
- key = repeat_kv(key, module.num_key_value_groups)
- value = repeat_kv(value, module.num_key_value_groups)
- # Get the right causal mask for the current layer
- causal_mask = attention_mask
- # Run the actual attention
- query = query.contiguous()
- key = key.contiguous()
- value = value.contiguous()
- attn_output = torch.nn.functional.scaled_dot_product_attention(
- query,
- key,
- value,
- attn_mask=causal_mask,
- dropout_p=dropout,
- scale=scaling,
- # Packed sequence format is used for input, so that it can never be causal.
- is_causal=False,
- )
- attn_output = attn_output.transpose(1, 2).contiguous()
- return attn_output, None
|
