Spaces:
Running
on
Zero
Running
on
Zero
File size: 2,441 Bytes
5f093a6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 |
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# 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.
import torch
from torch import nn
from transformers import CLIPPreTrainedModel, CLIPVisionModel
from ...models.attention import BasicTransformerBlock
from ...utils import logging
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
class PaintByExampleImageEncoder(CLIPPreTrainedModel):
def __init__(self, config, proj_size=768):
super().__init__(config)
self.proj_size = proj_size
self.model = CLIPVisionModel(config)
self.mapper = PaintByExampleMapper(config)
self.final_layer_norm = nn.LayerNorm(config.hidden_size)
self.proj_out = nn.Linear(config.hidden_size, self.proj_size)
# uncondition for scaling
self.uncond_vector = nn.Parameter(torch.randn((1, 1, self.proj_size)))
def forward(self, pixel_values, return_uncond_vector=False):
clip_output = self.model(pixel_values=pixel_values)
latent_states = clip_output.pooler_output
latent_states = self.mapper(latent_states[:, None])
latent_states = self.final_layer_norm(latent_states)
latent_states = self.proj_out(latent_states)
if return_uncond_vector:
return latent_states, self.uncond_vector
return latent_states
class PaintByExampleMapper(nn.Module):
def __init__(self, config):
super().__init__()
num_layers = (config.num_hidden_layers + 1) // 5
hid_size = config.hidden_size
num_heads = 1
self.blocks = nn.ModuleList(
[
BasicTransformerBlock(hid_size, num_heads, hid_size, activation_fn="gelu", attention_bias=True)
for _ in range(num_layers)
]
)
def forward(self, hidden_states):
for block in self.blocks:
hidden_states = block(hidden_states)
return hidden_states
|