Spaces:
Running
on
L40S
Running
on
L40S
File size: 3,349 Bytes
4450790 |
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 69 70 71 72 73 |
from comfy import sd1_clip
import comfy.text_encoders.t5
import comfy.text_encoders.sd3_clip
import comfy.model_management
from transformers import T5TokenizerFast
import torch
import os
class T5XXLTokenizer(sd1_clip.SDTokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
tokenizer_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "t5_tokenizer")
super().__init__(tokenizer_path, embedding_directory=embedding_directory, pad_with_end=False, embedding_size=4096, embedding_key='t5xxl', tokenizer_class=T5TokenizerFast, has_start_token=False, pad_to_max_length=False, max_length=99999999, min_length=256)
class FluxTokenizer:
def __init__(self, embedding_directory=None, tokenizer_data={}):
clip_l_tokenizer_class = tokenizer_data.get("clip_l_tokenizer_class", sd1_clip.SDTokenizer)
self.clip_l = clip_l_tokenizer_class(embedding_directory=embedding_directory)
self.t5xxl = T5XXLTokenizer(embedding_directory=embedding_directory)
def tokenize_with_weights(self, text:str, return_word_ids=False):
out = {}
out["l"] = self.clip_l.tokenize_with_weights(text, return_word_ids)
out["t5xxl"] = self.t5xxl.tokenize_with_weights(text, return_word_ids)
return out
def untokenize(self, token_weight_pair):
return self.clip_l.untokenize(token_weight_pair)
def state_dict(self):
return {}
class FluxClipModel(torch.nn.Module):
def __init__(self, dtype_t5=None, device="cpu", dtype=None, model_options={}):
super().__init__()
dtype_t5 = comfy.model_management.pick_weight_dtype(dtype_t5, dtype, device)
clip_l_class = model_options.get("clip_l_class", sd1_clip.SDClipModel)
self.clip_l = clip_l_class(device=device, dtype=dtype, return_projected_pooled=False, model_options=model_options)
self.t5xxl = comfy.text_encoders.sd3_clip.T5XXLModel(device=device, dtype=dtype_t5, model_options=model_options)
self.dtypes = set([dtype, dtype_t5])
def set_clip_options(self, options):
self.clip_l.set_clip_options(options)
self.t5xxl.set_clip_options(options)
def reset_clip_options(self):
self.clip_l.reset_clip_options()
self.t5xxl.reset_clip_options()
def encode_token_weights(self, token_weight_pairs):
token_weight_pairs_l = token_weight_pairs["l"]
token_weight_pairs_t5 = token_weight_pairs["t5xxl"]
t5_out, t5_pooled = self.t5xxl.encode_token_weights(token_weight_pairs_t5)
l_out, l_pooled = self.clip_l.encode_token_weights(token_weight_pairs_l)
return t5_out, l_pooled
def load_sd(self, sd):
if "text_model.encoder.layers.1.mlp.fc1.weight" in sd:
return self.clip_l.load_sd(sd)
else:
return self.t5xxl.load_sd(sd)
def flux_clip(dtype_t5=None, t5xxl_scaled_fp8=None):
class FluxClipModel_(FluxClipModel):
def __init__(self, device="cpu", dtype=None, model_options={}):
if t5xxl_scaled_fp8 is not None and "t5xxl_scaled_fp8" not in model_options:
model_options = model_options.copy()
model_options["t5xxl_scaled_fp8"] = t5xxl_scaled_fp8
super().__init__(dtype_t5=dtype_t5, device=device, dtype=dtype, model_options=model_options)
return FluxClipModel_
|