ComfyUI/comfy_extras/nodes_tomesd.py

178 lines
6.5 KiB
Python
Raw Permalink Normal View History

2023-04-03 19:49:28 +00:00
#Taken from: https://github.com/dbolya/tomesd
import torch
from typing import Tuple, Callable
import math
def do_nothing(x: torch.Tensor, mode:str=None):
return x
2023-04-03 19:49:28 +00:00
def mps_gather_workaround(input, dim, index):
if input.shape[-1] == 1:
return torch.gather(
input.unsqueeze(-1),
dim - 1 if dim < 0 else dim,
index.unsqueeze(-1)
).squeeze(-1)
else:
return torch.gather(input, dim, index)
def bipartite_soft_matching_random2d(metric: torch.Tensor,
w: int, h: int, sx: int, sy: int, r: int,
no_rand: bool = False) -> Tuple[Callable, Callable]:
"""
Partitions the tokens into src and dst and merges r tokens from src to dst.
Dst tokens are partitioned by choosing one randomy in each (sx, sy) region.
Args:
- metric [B, N, C]: metric to use for similarity
- w: image width in tokens
- h: image height in tokens
- sx: stride in the x dimension for dst, must divide w
- sy: stride in the y dimension for dst, must divide h
- r: number of tokens to remove (by merging)
- no_rand: if true, disable randomness (use top left corner only)
"""
B, N, _ = metric.shape
2023-05-12 21:49:09 +00:00
if r <= 0 or w == 1 or h == 1:
return do_nothing, do_nothing
2023-04-03 19:49:28 +00:00
gather = mps_gather_workaround if metric.device.type == "mps" else torch.gather
with torch.no_grad():
hsy, wsx = h // sy, w // sx
# For each sy by sx kernel, randomly assign one token to be dst and the rest src
if no_rand:
2023-04-03 19:49:28 +00:00
rand_idx = torch.zeros(hsy, wsx, 1, device=metric.device, dtype=torch.int64)
else:
2023-04-03 19:49:28 +00:00
rand_idx = torch.randint(sy*sx, size=(hsy, wsx, 1), device=metric.device)
2023-04-03 19:49:28 +00:00
# The image might not divide sx and sy, so we need to work on a view of the top left if the idx buffer instead
idx_buffer_view = torch.zeros(hsy, wsx, sy*sx, device=metric.device, dtype=torch.int64)
idx_buffer_view.scatter_(dim=2, index=rand_idx, src=-torch.ones_like(rand_idx, dtype=rand_idx.dtype))
idx_buffer_view = idx_buffer_view.view(hsy, wsx, sy, sx).transpose(1, 2).reshape(hsy * sy, wsx * sx)
# Image is not divisible by sx or sy so we need to move it into a new buffer
if (hsy * sy) < h or (wsx * sx) < w:
idx_buffer = torch.zeros(h, w, device=metric.device, dtype=torch.int64)
idx_buffer[:(hsy * sy), :(wsx * sx)] = idx_buffer_view
else:
idx_buffer = idx_buffer_view
# We set dst tokens to be -1 and src to be 0, so an argsort gives us dst|src indices
rand_idx = idx_buffer.reshape(1, -1, 1).argsort(dim=1)
# We're finished with these
del idx_buffer, idx_buffer_view
2023-04-03 19:49:28 +00:00
# rand_idx is currently dst|src, so split them
num_dst = hsy * wsx
a_idx = rand_idx[:, num_dst:, :] # src
b_idx = rand_idx[:, :num_dst, :] # dst
def split(x):
C = x.shape[-1]
2023-04-03 19:49:28 +00:00
src = gather(x, dim=1, index=a_idx.expand(B, N - num_dst, C))
dst = gather(x, dim=1, index=b_idx.expand(B, num_dst, C))
return src, dst
2023-04-03 19:49:28 +00:00
# Cosine similarity between A and B
metric = metric / metric.norm(dim=-1, keepdim=True)
a, b = split(metric)
scores = a @ b.transpose(-1, -2)
# Can't reduce more than the # tokens in src
r = min(a.shape[1], r)
2023-04-03 19:49:28 +00:00
# Find the most similar greedily
node_max, node_idx = scores.max(dim=-1)
edge_idx = node_max.argsort(dim=-1, descending=True)[..., None]
unm_idx = edge_idx[..., r:, :] # Unmerged Tokens
src_idx = edge_idx[..., :r, :] # Merged Tokens
2023-04-03 19:49:28 +00:00
dst_idx = gather(node_idx[..., None], dim=-2, index=src_idx)
def merge(x: torch.Tensor, mode="mean") -> torch.Tensor:
src, dst = split(x)
n, t1, c = src.shape
2023-04-03 19:49:28 +00:00
unm = gather(src, dim=-2, index=unm_idx.expand(n, t1 - r, c))
src = gather(src, dim=-2, index=src_idx.expand(n, r, c))
dst = dst.scatter_reduce(-2, dst_idx.expand(n, r, c), src, reduce=mode)
return torch.cat([unm, dst], dim=1)
def unmerge(x: torch.Tensor) -> torch.Tensor:
unm_len = unm_idx.shape[1]
unm, dst = x[..., :unm_len, :], x[..., unm_len:, :]
_, _, c = unm.shape
2023-04-03 19:49:28 +00:00
src = gather(dst, dim=-2, index=dst_idx.expand(B, r, c))
# Combine back to the original shape
out = torch.zeros(B, N, c, device=x.device, dtype=x.dtype)
out.scatter_(dim=-2, index=b_idx.expand(B, num_dst, c), src=dst)
2023-04-03 19:49:28 +00:00
out.scatter_(dim=-2, index=gather(a_idx.expand(B, a_idx.shape[1], 1), dim=1, index=unm_idx).expand(B, unm_len, c), src=unm)
out.scatter_(dim=-2, index=gather(a_idx.expand(B, a_idx.shape[1], 1), dim=1, index=src_idx).expand(B, r, c), src=src)
return out
return merge, unmerge
def get_functions(x, ratio, original_shape):
b, c, original_h, original_w = original_shape
original_tokens = original_h * original_w
2023-04-03 19:49:28 +00:00
downsample = int(math.ceil(math.sqrt(original_tokens // x.shape[1])))
stride_x = 2
stride_y = 2
max_downsample = 1
if downsample <= max_downsample:
2023-04-03 19:49:28 +00:00
w = int(math.ceil(original_w / downsample))
h = int(math.ceil(original_h / downsample))
r = int(x.shape[1] * ratio)
no_rand = False
m, u = bipartite_soft_matching_random2d(x, w, h, stride_x, stride_y, r, no_rand)
return m, u
nothing = lambda y: y
return nothing, nothing
class TomePatchModel:
@classmethod
def INPUT_TYPES(s):
return {"required": { "model": ("MODEL",),
"ratio": ("FLOAT", {"default": 0.3, "min": 0.0, "max": 1.0, "step": 0.01}),
}}
RETURN_TYPES = ("MODEL",)
FUNCTION = "patch"
2024-09-25 07:24:13 +00:00
CATEGORY = "model_patches/unet"
def patch(self, model, ratio):
self.u = None
def tomesd_m(q, k, v, extra_options):
#NOTE: In the reference code get_functions takes x (input of the transformer block) as the argument instead of q
#however from my basic testing it seems that using q instead gives better results
m, self.u = get_functions(q, ratio, extra_options["original_shape"])
return m(q), k, v
def tomesd_u(n, extra_options):
return self.u(n)
m = model.clone()
m.set_model_attn1_patch(tomesd_m)
m.set_model_attn1_output_patch(tomesd_u)
return (m, )
NODE_CLASS_MAPPINGS = {
"TomePatchModel": TomePatchModel,
}