32 independent transformer blocks › Attention
Rotated keys
×tensorxk · 160
Key feature pairs after their position-dependent rotations.
- shape
- [B batch, 8 kv heads, 1 positions, 128 d_h]
Only 8 heads, already rotated: this is what the cache stores (model.py 165).
Codemodel.py 65–177
self.cache_k = torch.zeros(
(
args.max_batch_size,
args.max_seq_len,
self.n_local_kv_heads,
self.head_dim,
)
).cuda()
self.cache_v = torch.zeros(
(
args.max_batch_size,
args.max_seq_len,
self.n_local_kv_heads,
self.head_dim,
)
).cuda() self.cache_k[:bsz, start_pos : start_pos + seqlen] = xk
self.cache_v[:bsz, start_pos : start_pos + seqlen] = xv
keys = self.cache_k[:bsz, : start_pos + seqlen]
values = self.cache_v[:bsz, : start_pos + seqlen]
# repeat k/v heads if n_kv_heads < n_heads
keys = repeat_kv(
keys, self.n_rep
) # (bs, cache_len + seqlen, n_local_heads, head_dim)
values = repeat_kv(
values, self.n_rep
) # (bs, cache_len + seqlen, n_local_heads, head_dim)def apply_rotary_emb(
xq: torch.Tensor,
xk: torch.Tensor,
freqs_cis: torch.Tensor,
) -> Tuple[torch.Tensor, torch.Tensor]:
xq_ = torch.view_as_complex(xq.float().reshape(*xq.shape[:-1], -1, 2))
xk_ = torch.view_as_complex(xk.float().reshape(*xk.shape[:-1], -1, 2))
freqs_cis = reshape_for_broadcast(freqs_cis, xq_)
xq_out = torch.view_as_real(xq_ * freqs_cis).flatten(3)
xk_out = torch.view_as_real(xk_ * freqs_cis).flatten(3)
return xq_out.type_as(xq), xk_out.type_as(xk) xq = xq.view(bsz, seqlen, self.n_local_heads, self.head_dim)
xk = xk.view(bsz, seqlen, self.n_local_kv_heads, self.head_dim)
xv = xv.view(bsz, seqlen, self.n_local_kv_heads, self.head_dim)
xq, xk = apply_rotary_emb(xq, xk, freqs_cis=freqs_cis)Sources2
Contextual evidence
The cache is written with the rotated keys (model.py 165).
kv-cache · meta-llama/llama3 / llama/model.py
Pinned revision a0940f9cf706 ↗ · lines 129–177
Read code excerpt
self.cache_k = torch.zeros(
(
args.max_batch_size,
args.max_seq_len,
self.n_local_kv_heads,
self.head_dim,
)
).cuda()
self.cache_v = torch.zeros(
(
args.max_batch_size,
args.max_seq_len,
self.n_local_kv_heads,
self.head_dim,
)
).cuda()… lines 145–164 omitted …
self.cache_k[:bsz, start_pos : start_pos + seqlen] = xk
self.cache_v[:bsz, start_pos : start_pos + seqlen] = xv
keys = self.cache_k[:bsz, : start_pos + seqlen]
values = self.cache_v[:bsz, : start_pos + seqlen]
# repeat k/v heads if n_kv_heads < n_heads
keys = repeat_kv(
keys, self.n_rep
) # (bs, cache_len + seqlen, n_local_heads, head_dim)
values = repeat_kv(
values, self.n_rep
) # (bs, cache_len + seqlen, n_local_heads, head_dim)Contextual evidence
Complex-pair rotations apply to queries and keys, with their original dtype restored.
rope · meta-llama/llama3 / llama/model.py
Pinned revision a0940f9cf706 ↗ · lines 65–160
Read code excerpt
def apply_rotary_emb(
xq: torch.Tensor,
xk: torch.Tensor,
freqs_cis: torch.Tensor,
) -> Tuple[torch.Tensor, torch.Tensor]:
xq_ = torch.view_as_complex(xq.float().reshape(*xq.shape[:-1], -1, 2))
xk_ = torch.view_as_complex(xk.float().reshape(*xk.shape[:-1], -1, 2))
freqs_cis = reshape_for_broadcast(freqs_cis, xq_)
xq_out = torch.view_as_real(xq_ * freqs_cis).flatten(3)
xk_out = torch.view_as_real(xk_ * freqs_cis).flatten(3)
return xq_out.type_as(xq), xk_out.type_as(xk)… lines 76–155 omitted …
xq = xq.view(bsz, seqlen, self.n_local_heads, self.head_dim)
xk = xk.view(bsz, seqlen, self.n_local_kv_heads, self.head_dim)
xv = xv.view(bsz, seqlen, self.n_local_kv_heads, self.head_dim)
xq, xk = apply_rotary_emb(xq, xk, freqs_cis=freqs_cis)