Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

Modifying code from https://github.com/tencent-ailab/IP-Adapter. Also adding license notice at top.

Browse Source
This commit is contained in:
user1
2023-08-29 06:29:05 -07:00
parent 1ad98ce999
commit 8c1390166f
4 changed files with 94 additions and 68 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
invokeai/backend/ip_adapter/resampler.py
View File

@ -1,4 +1,6 @@
# modified from https://github.com/mlfoundations/open_flamingo/blob/main/open_flamingo/src/helpers.py
# copied from https://github.com/tencent-ailab/IP-Adapter (Apache License 2.0)
# tencent ailab comment: modified from https://github.com/mlfoundations/open_flamingo/blob/main/open_flamingo/src/helpers.py
import math
import torch
@ -14,8 +16,8 @@ def FeedForward(dim, mult=4):
nn.GELU(),
nn.Linear(inner_dim, dim, bias=False),
)
def reshape_tensor(x, heads):
bs, length, width = x.shape
#(bs, length, width) --> (bs, length, n_heads, dim_per_head)
@ -53,13 +55,13 @@ class PerceiverAttention(nn.Module):
"""
x = self.norm1(x)
latents = self.norm2(latents)
b, l, _ = latents.shape
q = self.to_q(latents)
kv_input = torch.cat((x, latents), dim=-2)
k, v = self.to_kv(kv_input).chunk(2, dim=-1)
q = reshape_tensor(q, self.heads)
k = reshape_tensor(k, self.heads)
v = reshape_tensor(v, self.heads)
@ -69,7 +71,7 @@ class PerceiverAttention(nn.Module):
weight = (q * scale) @ (k * scale).transpose(-2, -1) # More stable with f16 than dividing afterwards
weight = torch.softmax(weight.float(), dim=-1).type(weight.dtype)
out = weight @ v
out = out.permute(0, 2, 1, 3).reshape(b, l, -1)
return self.to_out(out)
@ -88,14 +90,14 @@ class Resampler(nn.Module):
ff_mult=4,
):
super().__init__()
self.latents = nn.Parameter(torch.randn(1, num_queries, dim) / dim**0.5)
self.proj_in = nn.Linear(embedding_dim, dim)
self.proj_out = nn.Linear(dim, output_dim)
self.norm_out = nn.LayerNorm(output_dim)
self.layers = nn.ModuleList([])
for _ in range(depth):
self.layers.append(
@ -108,14 +110,14 @@ class Resampler(nn.Module):
)
def forward(self, x):
latents = self.latents.repeat(x.size(0), 1, 1)
x = self.proj_in(x)
for attn, ff in self.layers:
latents = attn(x, latents) + latents
latents = ff(latents) + latents
latents = self.proj_out(latents)
return self.norm_out(latents)
return self.norm_out(latents)