The most powerful and modular stable diffusion GUI, api and backend with a graph/nodes interface.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

35 lines
1.4 KiB

2 years ago
import sd1_clip
import torch
import os
2 years ago
class SD2ClipModel(sd1_clip.SD1ClipModel):
def __init__(self, arch="ViT-H-14", device="cpu", max_length=77, freeze=True, layer="penultimate", layer_idx=None):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "sd2_clip_config.json")
super().__init__(device=device, freeze=freeze, textmodel_json_config=textmodel_json_config)
2 years ago
self.empty_tokens = [[49406] + [49407] + [0] * 75]
if layer == "last":
pass
elif layer == "penultimate":
layer_idx = -1
self.clip_layer(layer_idx)
2 years ago
elif self.layer == "hidden":
assert layer_idx is not None
assert abs(layer_idx) < 24
self.clip_layer(layer_idx)
2 years ago
else:
raise NotImplementedError()
def clip_layer(self, layer_idx):
if layer_idx < 0:
layer_idx -= 1 #The real last layer of SD2.x clip is the penultimate one. The last one might contain garbage.
if abs(layer_idx) >= 24:
self.layer = "hidden"
self.layer_idx = -2
else:
self.layer = "hidden"
self.layer_idx = layer_idx
2 years ago
class SD2Tokenizer(sd1_clip.SD1Tokenizer):
def __init__(self, tokenizer_path=None, embedding_directory=None):
super().__init__(tokenizer_path, pad_with_end=False, embedding_directory=embedding_directory)