IceClear
upload files
42f2c22
# // Copyright (c) 2025 Bytedance Ltd. and/or its affiliates
# //
# // Licensed under the Apache License, Version 2.0 (the "License");
# // you may not use this file except in compliance with the License.
# // You may obtain a copy of the License at
# //
# // http://www.apache.org/licenses/LICENSE-2.0
# //
# // Unless required by applicable law or agreed to in writing, software
# // distributed under the License is distributed on an "AS IS" BASIS,
# // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# // See the License for the specific language governing permissions and
# // limitations under the License.
from typing import Optional, Tuple, Union
import torch
from einops import rearrange
from torch import nn
from torch.nn import functional as F
from torch.nn.modules.utils import _triple
from common.cache import Cache
from common.distributed.ops import gather_heads_scatter_seq, gather_seq_scatter_heads_qkv
from ... import na
from ...attention import FlashAttentionVarlen
from ...mm import MMArg, MMModule
from ...normalization import norm_layer_type
from ...rope import get_na_rope
from ...window import get_window_op
from itertools import chain
class NaMMAttention(nn.Module):
def __init__(
self,
vid_dim: int,
txt_dim: int,
heads: int,
head_dim: int,
qk_bias: bool,
qk_norm: norm_layer_type,
qk_norm_eps: float,
rope_type: Optional[str],
rope_dim: int,
shared_weights: bool,
**kwargs,
):
super().__init__()
dim = MMArg(vid_dim, txt_dim)
inner_dim = heads * head_dim
qkv_dim = inner_dim * 3
self.head_dim = head_dim
self.proj_qkv = MMModule(
nn.Linear, dim, qkv_dim, bias=qk_bias, shared_weights=shared_weights
)
self.proj_out = MMModule(nn.Linear, inner_dim, dim, shared_weights=shared_weights)
self.norm_q = MMModule(
qk_norm,
dim=head_dim,
eps=qk_norm_eps,
elementwise_affine=True,
shared_weights=shared_weights,
)
self.norm_k = MMModule(
qk_norm,
dim=head_dim,
eps=qk_norm_eps,
elementwise_affine=True,
shared_weights=shared_weights,
)
self.rope = get_na_rope(rope_type=rope_type, dim=rope_dim)
self.attn = FlashAttentionVarlen()
def forward(
self,
vid: torch.FloatTensor, # l c
txt: torch.FloatTensor, # l c
vid_shape: torch.LongTensor, # b 3
txt_shape: torch.LongTensor, # b 1
cache: Cache,
) -> Tuple[
torch.FloatTensor,
torch.FloatTensor,
]:
vid_qkv, txt_qkv = self.proj_qkv(vid, txt)
vid_qkv = gather_seq_scatter_heads_qkv(
vid_qkv,
seq_dim=0,
qkv_shape=vid_shape,
cache=cache.namespace("vid"),
)
txt_qkv = gather_seq_scatter_heads_qkv(
txt_qkv,
seq_dim=0,
qkv_shape=txt_shape,
cache=cache.namespace("txt"),
)
vid_qkv = rearrange(vid_qkv, "l (o h d) -> l o h d", o=3, d=self.head_dim)
txt_qkv = rearrange(txt_qkv, "l (o h d) -> l o h d", o=3, d=self.head_dim)
vid_q, vid_k, vid_v = vid_qkv.unbind(1)
txt_q, txt_k, txt_v = txt_qkv.unbind(1)
vid_q, txt_q = self.norm_q(vid_q, txt_q)
vid_k, txt_k = self.norm_k(vid_k, txt_k)
if self.rope:
if self.rope.mm:
vid_q, vid_k, txt_q, txt_k = self.rope(
vid_q, vid_k, vid_shape, txt_q, txt_k, txt_shape, cache
)
else:
vid_q, vid_k = self.rope(vid_q, vid_k, vid_shape, cache)
vid_len = cache("vid_len", lambda: vid_shape.prod(-1))
txt_len = cache("txt_len", lambda: txt_shape.prod(-1))
all_len = cache("all_len", lambda: vid_len + txt_len)
concat, unconcat = cache("mm_pnp", lambda: na.concat_idx(vid_len, txt_len))
attn = self.attn(
q=concat(vid_q, txt_q).bfloat16(),
k=concat(vid_k, txt_k).bfloat16(),
v=concat(vid_v, txt_v).bfloat16(),
cu_seqlens_q=cache("mm_seqlens", lambda: F.pad(all_len.cumsum(0), (1, 0)).int()),
cu_seqlens_k=cache("mm_seqlens", lambda: F.pad(all_len.cumsum(0), (1, 0)).int()),
max_seqlen_q=cache("mm_maxlen", lambda: all_len.max().item()),
max_seqlen_k=cache("mm_maxlen", lambda: all_len.max().item()),
).type_as(vid_q)
attn = rearrange(attn, "l h d -> l (h d)")
vid_out, txt_out = unconcat(attn)
vid_out = gather_heads_scatter_seq(vid_out, head_dim=1, seq_dim=0)
txt_out = gather_heads_scatter_seq(txt_out, head_dim=1, seq_dim=0)
vid_out, txt_out = self.proj_out(vid_out, txt_out)
return vid_out, txt_out
class NaSwinAttention(NaMMAttention):
def __init__(
self,
*args,
window: Union[int, Tuple[int, int, int]],
window_method: str,
**kwargs,
):
super().__init__(*args, **kwargs)
self.window = _triple(window)
self.window_method = window_method
assert all(map(lambda v: isinstance(v, int) and v >= 0, self.window))
self.window_op = get_window_op(window_method)
def forward(
self,
vid: torch.FloatTensor, # l c
txt: torch.FloatTensor, # l c
vid_shape: torch.LongTensor, # b 3
txt_shape: torch.LongTensor, # b 1
cache: Cache,
) -> Tuple[
torch.FloatTensor,
torch.FloatTensor,
]:
vid_qkv, txt_qkv = self.proj_qkv(vid, txt)
vid_qkv = gather_seq_scatter_heads_qkv(
vid_qkv,
seq_dim=0,
qkv_shape=vid_shape,
cache=cache.namespace("vid"),
)
txt_qkv = gather_seq_scatter_heads_qkv(
txt_qkv,
seq_dim=0,
qkv_shape=txt_shape,
cache=cache.namespace("txt"),
)
# re-org the input seq for window attn
cache_win = cache.namespace(f"{self.window_method}_{self.window}_sd3")
def make_window(x: torch.Tensor):
t, h, w, _ = x.shape
window_slices = self.window_op((t, h, w), self.window)
return [x[st, sh, sw] for (st, sh, sw) in window_slices]
window_partition, window_reverse, window_shape, window_count = cache_win(
"win_transform",
lambda: na.window_idx(vid_shape, make_window),
)
vid_qkv_win = window_partition(vid_qkv)
vid_qkv_win = rearrange(vid_qkv_win, "l (o h d) -> l o h d", o=3, d=self.head_dim)
txt_qkv = rearrange(txt_qkv, "l (o h d) -> l o h d", o=3, d=self.head_dim)
vid_q, vid_k, vid_v = vid_qkv_win.unbind(1)
txt_q, txt_k, txt_v = txt_qkv.unbind(1)
vid_q, txt_q = self.norm_q(vid_q, txt_q)
vid_k, txt_k = self.norm_k(vid_k, txt_k)
txt_len = cache("txt_len", lambda: txt_shape.prod(-1))
vid_len_win = cache_win("vid_len", lambda: window_shape.prod(-1))
txt_len_win = cache_win("txt_len", lambda: txt_len.repeat_interleave(window_count))
all_len_win = cache_win("all_len", lambda: vid_len_win + txt_len_win)
concat_win, unconcat_win = cache_win(
"mm_pnp", lambda: na.repeat_concat_idx(vid_len_win, txt_len, window_count)
)
# window rope
if self.rope:
if self.rope.mm:
# repeat text q and k for window mmrope
_, num_h, _ = txt_q.shape
txt_q_repeat = rearrange(txt_q, "l h d -> l (h d)")
txt_q_repeat = na.unflatten(txt_q_repeat, txt_shape)
txt_q_repeat = [[x] * n for x, n in zip(txt_q_repeat, window_count)]
txt_q_repeat = list(chain(*txt_q_repeat))
txt_q_repeat, txt_shape_repeat = na.flatten(txt_q_repeat)
txt_q_repeat = rearrange(txt_q_repeat, "l (h d) -> l h d", h=num_h)
txt_k_repeat = rearrange(txt_k, "l h d -> l (h d)")
txt_k_repeat = na.unflatten(txt_k_repeat, txt_shape)
txt_k_repeat = [[x] * n for x, n in zip(txt_k_repeat, window_count)]
txt_k_repeat = list(chain(*txt_k_repeat))
txt_k_repeat, _ = na.flatten(txt_k_repeat)
txt_k_repeat = rearrange(txt_k_repeat, "l (h d) -> l h d", h=num_h)
vid_q, vid_k, txt_q, txt_k = self.rope(
vid_q, vid_k, window_shape, txt_q_repeat, txt_k_repeat, txt_shape_repeat, cache_win
)
else:
vid_q, vid_k = self.rope(vid_q, vid_k, window_shape, cache_win)
out = self.attn(
q=concat_win(vid_q, txt_q).bfloat16(),
k=concat_win(vid_k, txt_k).bfloat16(),
v=concat_win(vid_v, txt_v).bfloat16(),
cu_seqlens_q=cache_win(
"vid_seqlens_q", lambda: F.pad(all_len_win.cumsum(0), (1, 0)).int()
),
cu_seqlens_k=cache_win(
"vid_seqlens_k", lambda: F.pad(all_len_win.cumsum(0), (1, 0)).int()
),
max_seqlen_q=cache_win("vid_max_seqlen_q", lambda: all_len_win.max().item()),
max_seqlen_k=cache_win("vid_max_seqlen_k", lambda: all_len_win.max().item()),
).type_as(vid_q)
# text pooling
vid_out, txt_out = unconcat_win(out)
vid_out = rearrange(vid_out, "l h d -> l (h d)")
txt_out = rearrange(txt_out, "l h d -> l (h d)")
vid_out = window_reverse(vid_out)
vid_out = gather_heads_scatter_seq(vid_out, head_dim=1, seq_dim=0)
txt_out = gather_heads_scatter_seq(txt_out, head_dim=1, seq_dim=0)
vid_out, txt_out = self.proj_out(vid_out, txt_out)
return vid_out, txt_out