netadmin/UI_latest
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Release on 09.11.24

Browse Source
This commit is contained in:
netadmin
2024-11-09 17:19:42 +03:00
commit 3c64fe0c3e
339 changed files with 456101 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
comfy/text_encoders/aura_t5.py Normal file
View File

@@ -0,0 +1,22 @@
from comfy import sd1_clip
from .spiece_tokenizer import SPieceTokenizer
import comfy.text_encoders.t5
import os
class PT5XlModel(sd1_clip.SDClipModel):
def __init__(self, device="cpu", layer="last", layer_idx=None, dtype=None, model_options={}):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "t5_pile_config_xl.json")
super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, dtype=dtype, special_tokens={"end": 2, "pad": 1}, model_class=comfy.text_encoders.t5.T5, enable_attention_masks=True, zero_out_masked=True, model_options=model_options)
class PT5XlTokenizer(sd1_clip.SDTokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
tokenizer_path = os.path.join(os.path.join(os.path.dirname(os.path.realpath(__file__)), "t5_pile_tokenizer"), "tokenizer.model")
super().__init__(tokenizer_path, pad_with_end=False, embedding_size=2048, embedding_key='pile_t5xl', tokenizer_class=SPieceTokenizer, has_start_token=False, pad_to_max_length=False, max_length=99999999, min_length=256, pad_token=1)
class AuraT5Tokenizer(sd1_clip.SD1Tokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
super().__init__(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data, clip_name="pile_t5xl", tokenizer=PT5XlTokenizer)
class AuraT5Model(sd1_clip.SD1ClipModel):
def __init__(self, device="cpu", dtype=None, model_options={}, **kwargs):
super().__init__(device=device, dtype=dtype, model_options=model_options, name="pile_t5xl", clip_model=PT5XlModel, **kwargs)

140
comfy/text_encoders/bert.py Normal file
View File

@@ -0,0 +1,140 @@
import torch
from comfy.ldm.modules.attention import optimized_attention_for_device
import comfy.ops
class BertAttention(torch.nn.Module):
def __init__(self, embed_dim, heads, dtype, device, operations):
super().__init__()
self.heads = heads
self.query = operations.Linear(embed_dim, embed_dim, bias=True, dtype=dtype, device=device)
self.key = operations.Linear(embed_dim, embed_dim, bias=True, dtype=dtype, device=device)
self.value = operations.Linear(embed_dim, embed_dim, bias=True, dtype=dtype, device=device)
def forward(self, x, mask=None, optimized_attention=None):
q = self.query(x)
k = self.key(x)
v = self.value(x)
out = optimized_attention(q, k, v, self.heads, mask)
return out
class BertOutput(torch.nn.Module):
def __init__(self, input_dim, output_dim, layer_norm_eps, dtype, device, operations):
super().__init__()
self.dense = operations.Linear(input_dim, output_dim, dtype=dtype, device=device)
self.LayerNorm = operations.LayerNorm(output_dim, eps=layer_norm_eps, dtype=dtype, device=device)
# self.dropout = nn.Dropout(0.0)
def forward(self, x, y):
x = self.dense(x)
# hidden_states = self.dropout(hidden_states)
x = self.LayerNorm(x + y)
return x
class BertAttentionBlock(torch.nn.Module):
def __init__(self, embed_dim, heads, layer_norm_eps, dtype, device, operations):
super().__init__()
self.self = BertAttention(embed_dim, heads, dtype, device, operations)
self.output = BertOutput(embed_dim, embed_dim, layer_norm_eps, dtype, device, operations)
def forward(self, x, mask, optimized_attention):
y = self.self(x, mask, optimized_attention)
return self.output(y, x)
class BertIntermediate(torch.nn.Module):
def __init__(self, embed_dim, intermediate_dim, dtype, device, operations):
super().__init__()
self.dense = operations.Linear(embed_dim, intermediate_dim, dtype=dtype, device=device)
def forward(self, x):
x = self.dense(x)
return torch.nn.functional.gelu(x)
class BertBlock(torch.nn.Module):
def __init__(self, embed_dim, intermediate_dim, heads, layer_norm_eps, dtype, device, operations):
super().__init__()
self.attention = BertAttentionBlock(embed_dim, heads, layer_norm_eps, dtype, device, operations)
self.intermediate = BertIntermediate(embed_dim, intermediate_dim, dtype, device, operations)
self.output = BertOutput(intermediate_dim, embed_dim, layer_norm_eps, dtype, device, operations)
def forward(self, x, mask, optimized_attention):
x = self.attention(x, mask, optimized_attention)
y = self.intermediate(x)
return self.output(y, x)
class BertEncoder(torch.nn.Module):
def __init__(self, num_layers, embed_dim, intermediate_dim, heads, layer_norm_eps, dtype, device, operations):
super().__init__()
self.layer = torch.nn.ModuleList([BertBlock(embed_dim, intermediate_dim, heads, layer_norm_eps, dtype, device, operations) for i in range(num_layers)])
def forward(self, x, mask=None, intermediate_output=None):
optimized_attention = optimized_attention_for_device(x.device, mask=mask is not None, small_input=True)
if intermediate_output is not None:
if intermediate_output < 0:
intermediate_output = len(self.layer) + intermediate_output
intermediate = None
for i, l in enumerate(self.layer):
x = l(x, mask, optimized_attention)
if i == intermediate_output:
intermediate = x.clone()
return x, intermediate
class BertEmbeddings(torch.nn.Module):
def __init__(self, vocab_size, max_position_embeddings, type_vocab_size, pad_token_id, embed_dim, layer_norm_eps, dtype, device, operations):
super().__init__()
self.word_embeddings = operations.Embedding(vocab_size, embed_dim, padding_idx=pad_token_id, dtype=dtype, device=device)
self.position_embeddings = operations.Embedding(max_position_embeddings, embed_dim, dtype=dtype, device=device)
self.token_type_embeddings = operations.Embedding(type_vocab_size, embed_dim, dtype=dtype, device=device)
self.LayerNorm = operations.LayerNorm(embed_dim, eps=layer_norm_eps, dtype=dtype, device=device)
def forward(self, input_tokens, token_type_ids=None, dtype=None):
x = self.word_embeddings(input_tokens, out_dtype=dtype)
x += comfy.ops.cast_to_input(self.position_embeddings.weight[:x.shape[1]], x)
if token_type_ids is not None:
x += self.token_type_embeddings(token_type_ids, out_dtype=x.dtype)
else:
x += comfy.ops.cast_to_input(self.token_type_embeddings.weight[0], x)
x = self.LayerNorm(x)
return x
class BertModel_(torch.nn.Module):
def __init__(self, config_dict, dtype, device, operations):
super().__init__()
embed_dim = config_dict["hidden_size"]
layer_norm_eps = config_dict["layer_norm_eps"]
self.embeddings = BertEmbeddings(config_dict["vocab_size"], config_dict["max_position_embeddings"], config_dict["type_vocab_size"], config_dict["pad_token_id"], embed_dim, layer_norm_eps, dtype, device, operations)
self.encoder = BertEncoder(config_dict["num_hidden_layers"], embed_dim, config_dict["intermediate_size"], config_dict["num_attention_heads"], layer_norm_eps, dtype, device, operations)
def forward(self, input_tokens, attention_mask=None, intermediate_output=None, final_layer_norm_intermediate=True, dtype=None):
x = self.embeddings(input_tokens, dtype=dtype)
mask = None
if attention_mask is not None:
mask = 1.0 - attention_mask.to(x.dtype).reshape((attention_mask.shape[0], 1, -1, attention_mask.shape[-1])).expand(attention_mask.shape[0], 1, attention_mask.shape[-1], attention_mask.shape[-1])
mask = mask.masked_fill(mask.to(torch.bool), float("-inf"))
x, i = self.encoder(x, mask, intermediate_output)
return x, i
class BertModel(torch.nn.Module):
def __init__(self, config_dict, dtype, device, operations):
super().__init__()
self.bert = BertModel_(config_dict, dtype, device, operations)
self.num_layers = config_dict["num_hidden_layers"]
def get_input_embeddings(self):
return self.bert.embeddings.word_embeddings
def set_input_embeddings(self, embeddings):
self.bert.embeddings.word_embeddings = embeddings
def forward(self, *args, **kwargs):
return self.bert(*args, **kwargs)

72
comfy/text_encoders/flux.py Normal file
View File

@@ -0,0 +1,72 @@
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_

38
comfy/text_encoders/genmo.py Normal file
View File

@@ -0,0 +1,38 @@
from comfy import sd1_clip
import comfy.text_encoders.sd3_clip
import os
from transformers import T5TokenizerFast
class T5XXLModel(comfy.text_encoders.sd3_clip.T5XXLModel):
def __init__(self, **kwargs):
kwargs["attention_mask"] = True
super().__init__(**kwargs)
class MochiT5XXL(sd1_clip.SD1ClipModel):
def __init__(self, device="cpu", dtype=None, model_options={}):
super().__init__(device=device, dtype=dtype, name="t5xxl", clip_model=T5XXLModel, model_options=model_options)
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 MochiT5Tokenizer(sd1_clip.SD1Tokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
super().__init__(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data, clip_name="t5xxl", tokenizer=T5XXLTokenizer)
def mochi_te(dtype_t5=None, t5xxl_scaled_fp8=None):
class MochiTEModel_(MochiT5XXL):
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
if dtype is None:
dtype = dtype_t5
super().__init__(device=device, dtype=dtype, model_options=model_options)
return MochiTEModel_

79
comfy/text_encoders/hydit.py Normal file
View File

@@ -0,0 +1,79 @@
from comfy import sd1_clip
from transformers import BertTokenizer
from .spiece_tokenizer import SPieceTokenizer
from .bert import BertModel
import comfy.text_encoders.t5
import os
import torch
class HyditBertModel(sd1_clip.SDClipModel):
def __init__(self, device="cpu", layer="last", layer_idx=None, dtype=None, model_options={}):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "hydit_clip.json")
super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, dtype=dtype, special_tokens={"start": 101, "end": 102, "pad": 0}, model_class=BertModel, enable_attention_masks=True, return_attention_masks=True, model_options=model_options)
class HyditBertTokenizer(sd1_clip.SDTokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
tokenizer_path = os.path.join(os.path.dirname(os.path.realpath(__file__)), "hydit_clip_tokenizer")
super().__init__(tokenizer_path, pad_with_end=False, embedding_size=1024, embedding_key='chinese_roberta', tokenizer_class=BertTokenizer, pad_to_max_length=False, max_length=512, min_length=77)
class MT5XLModel(sd1_clip.SDClipModel):
def __init__(self, device="cpu", layer="last", layer_idx=None, dtype=None, model_options={}):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "mt5_config_xl.json")
super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, dtype=dtype, special_tokens={"end": 1, "pad": 0}, model_class=comfy.text_encoders.t5.T5, enable_attention_masks=True, return_attention_masks=True, model_options=model_options)
class MT5XLTokenizer(sd1_clip.SDTokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
#tokenizer_path = os.path.join(os.path.join(os.path.dirname(os.path.realpath(__file__)), "mt5_tokenizer"), "spiece.model")
tokenizer = tokenizer_data.get("spiece_model", None)
super().__init__(tokenizer, pad_with_end=False, embedding_size=2048, embedding_key='mt5xl', tokenizer_class=SPieceTokenizer, has_start_token=False, pad_to_max_length=False, max_length=99999999, min_length=256)
def state_dict(self):
return {"spiece_model": self.tokenizer.serialize_model()}
class HyditTokenizer:
def __init__(self, embedding_directory=None, tokenizer_data={}):
mt5_tokenizer_data = tokenizer_data.get("mt5xl.spiece_model", None)
self.hydit_clip = HyditBertTokenizer(embedding_directory=embedding_directory)
self.mt5xl = MT5XLTokenizer(tokenizer_data={"spiece_model": mt5_tokenizer_data}, embedding_directory=embedding_directory)
def tokenize_with_weights(self, text:str, return_word_ids=False):
out = {}
out["hydit_clip"] = self.hydit_clip.tokenize_with_weights(text, return_word_ids)
out["mt5xl"] = self.mt5xl.tokenize_with_weights(text, return_word_ids)
return out
def untokenize(self, token_weight_pair):
return self.hydit_clip.untokenize(token_weight_pair)
def state_dict(self):
return {"mt5xl.spiece_model": self.mt5xl.state_dict()["spiece_model"]}
class HyditModel(torch.nn.Module):
def __init__(self, device="cpu", dtype=None, model_options={}):
super().__init__()
self.hydit_clip = HyditBertModel(dtype=dtype, model_options=model_options)
self.mt5xl = MT5XLModel(dtype=dtype, model_options=model_options)
self.dtypes = set()
if dtype is not None:
self.dtypes.add(dtype)
def encode_token_weights(self, token_weight_pairs):
hydit_out = self.hydit_clip.encode_token_weights(token_weight_pairs["hydit_clip"])
mt5_out = self.mt5xl.encode_token_weights(token_weight_pairs["mt5xl"])
return hydit_out[0], hydit_out[1], {"attention_mask": hydit_out[2]["attention_mask"], "conditioning_mt5xl": mt5_out[0], "attention_mask_mt5xl": mt5_out[2]["attention_mask"]}
def load_sd(self, sd):
if "bert.encoder.layer.0.attention.self.query.weight" in sd:
return self.hydit_clip.load_sd(sd)
else:
return self.mt5xl.load_sd(sd)
def set_clip_options(self, options):
self.hydit_clip.set_clip_options(options)
self.mt5xl.set_clip_options(options)
def reset_clip_options(self):
self.hydit_clip.reset_clip_options()
self.mt5xl.reset_clip_options()

35
comfy/text_encoders/hydit_clip.json Normal file
View File

@@ -0,0 +1,35 @@
{
"_name_or_path": "hfl/chinese-roberta-wwm-ext-large",
"architectures": [
"BertModel"
],
"attention_probs_dropout_prob": 0.1,
"bos_token_id": 0,
"classifier_dropout": null,
"directionality": "bidi",
"eos_token_id": 2,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 1024,
"initializer_range": 0.02,
"intermediate_size": 4096,
"layer_norm_eps": 1e-12,
"max_position_embeddings": 512,
"model_type": "bert",
"num_attention_heads": 16,
"num_hidden_layers": 24,
"output_past": true,
"pad_token_id": 0,
"pooler_fc_size": 768,
"pooler_num_attention_heads": 12,
"pooler_num_fc_layers": 3,
"pooler_size_per_head": 128,
"pooler_type": "first_token_transform",
"position_embedding_type": "absolute",
"torch_dtype": "float32",
"transformers_version": "4.22.1",
"type_vocab_size": 2,
"use_cache": true,
"vocab_size": 47020
}

7
comfy/text_encoders/hydit_clip_tokenizer/special_tokens_map.json Normal file
View File

@@ -0,0 +1,7 @@
{
"cls_token": "[CLS]",
"mask_token": "[MASK]",
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"unk_token": "[UNK]"
}

16
comfy/text_encoders/hydit_clip_tokenizer/tokenizer_config.json Normal file
View File

@@ -0,0 +1,16 @@
{
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": true,
"mask_token": "[MASK]",
"name_or_path": "hfl/chinese-roberta-wwm-ext",
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"special_tokens_map_file": "/home/chenweifeng/.cache/huggingface/hub/models--hfl--chinese-roberta-wwm-ext/snapshots/5c58d0b8ec1d9014354d691c538661bf00bfdb44/special_tokens_map.json",
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"unk_token": "[UNK]",
"model_max_length": 77
}

47020
comfy/text_encoders/hydit_clip_tokenizer/vocab.txt Normal file
View File

File diff suppressed because it is too large Load Diff

25
comfy/text_encoders/long_clipl.json Normal file
View File

@@ -0,0 +1,25 @@
{
"_name_or_path": "openai/clip-vit-large-patch14",
"architectures": [
"CLIPTextModel"
],
"attention_dropout": 0.0,
"bos_token_id": 0,
"dropout": 0.0,
"eos_token_id": 49407,
"hidden_act": "quick_gelu",
"hidden_size": 768,
"initializer_factor": 1.0,
"initializer_range": 0.02,
"intermediate_size": 3072,
"layer_norm_eps": 1e-05,
"max_position_embeddings": 248,
"model_type": "clip_text_model",
"num_attention_heads": 12,
"num_hidden_layers": 12,
"pad_token_id": 1,
"projection_dim": 768,
"torch_dtype": "float32",
"transformers_version": "4.24.0",
"vocab_size": 49408
}

30
comfy/text_encoders/long_clipl.py Normal file
View File

@@ -0,0 +1,30 @@
from comfy import sd1_clip
import os
class LongClipTokenizer_(sd1_clip.SDTokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
super().__init__(max_length=248, embedding_directory=embedding_directory, tokenizer_data=tokenizer_data)
class LongClipModel_(sd1_clip.SDClipModel):
def __init__(self, *args, **kwargs):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "long_clipl.json")
super().__init__(*args, textmodel_json_config=textmodel_json_config, **kwargs)
class LongClipTokenizer(sd1_clip.SD1Tokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
super().__init__(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data, tokenizer=LongClipTokenizer_)
class LongClipModel(sd1_clip.SD1ClipModel):
def __init__(self, device="cpu", dtype=None, model_options={}, **kwargs):
super().__init__(device=device, dtype=dtype, model_options=model_options, clip_model=LongClipModel_, **kwargs)
def model_options_long_clip(sd, tokenizer_data, model_options):
w = sd.get("clip_l.text_model.embeddings.position_embedding.weight", None)
if w is None:
w = sd.get("text_model.embeddings.position_embedding.weight", None)
if w is not None and w.shape[0] == 248:
tokenizer_data = tokenizer_data.copy()
model_options = model_options.copy()
tokenizer_data["clip_l_tokenizer_class"] = LongClipTokenizer_
model_options["clip_l_class"] = LongClipModel_
return tokenizer_data, model_options

22
comfy/text_encoders/mt5_config_xl.json Normal file
View File

@@ -0,0 +1,22 @@
{
"d_ff": 5120,
"d_kv": 64,
"d_model": 2048,
"decoder_start_token_id": 0,
"dropout_rate": 0.1,
"eos_token_id": 1,
"dense_act_fn": "gelu_pytorch_tanh",
"initializer_factor": 1.0,
"is_encoder_decoder": true,
"is_gated_act": true,
"layer_norm_epsilon": 1e-06,
"model_type": "mt5",
"num_decoder_layers": 24,
"num_heads": 32,
"num_layers": 24,
"output_past": true,
"pad_token_id": 0,
"relative_attention_num_buckets": 32,
"tie_word_embeddings": false,
"vocab_size": 250112
}

22
comfy/text_encoders/sa_t5.py Normal file
View File

@@ -0,0 +1,22 @@
from comfy import sd1_clip
from transformers import T5TokenizerFast
import comfy.text_encoders.t5
import os
class T5BaseModel(sd1_clip.SDClipModel):
def __init__(self, device="cpu", layer="last", layer_idx=None, dtype=None, model_options={}):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "t5_config_base.json")
super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, dtype=dtype, model_options=model_options, special_tokens={"end": 1, "pad": 0}, model_class=comfy.text_encoders.t5.T5, enable_attention_masks=True, zero_out_masked=True)
class T5BaseTokenizer(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, pad_with_end=False, embedding_size=768, embedding_key='t5base', tokenizer_class=T5TokenizerFast, has_start_token=False, pad_to_max_length=False, max_length=99999999, min_length=128)
class SAT5Tokenizer(sd1_clip.SD1Tokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
super().__init__(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data, clip_name="t5base", tokenizer=T5BaseTokenizer)
class SAT5Model(sd1_clip.SD1ClipModel):
def __init__(self, device="cpu", dtype=None, model_options={}, **kwargs):
super().__init__(device=device, dtype=dtype, model_options=model_options, name="t5base", clip_model=T5BaseModel, **kwargs)

23
comfy/text_encoders/sd2_clip.py Normal file
View File

@@ -0,0 +1,23 @@
from comfy import sd1_clip
import os
class SD2ClipHModel(sd1_clip.SDClipModel):
def __init__(self, arch="ViT-H-14", device="cpu", max_length=77, freeze=True, layer="penultimate", layer_idx=None, dtype=None, model_options={}):
if layer == "penultimate":
layer="hidden"
layer_idx=-2
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "sd2_clip_config.json")
super().__init__(device=device, freeze=freeze, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, dtype=dtype, special_tokens={"start": 49406, "end": 49407, "pad": 0}, return_projected_pooled=True, model_options=model_options)
class SD2ClipHTokenizer(sd1_clip.SDTokenizer):
def __init__(self, tokenizer_path=None, embedding_directory=None, tokenizer_data={}):
super().__init__(tokenizer_path, pad_with_end=False, embedding_directory=embedding_directory, embedding_size=1024)
class SD2Tokenizer(sd1_clip.SD1Tokenizer):
def __init__(self, embedding_directory=None, tokenizer_data={}):
super().__init__(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data, clip_name="h", tokenizer=SD2ClipHTokenizer)
class SD2ClipModel(sd1_clip.SD1ClipModel):
def __init__(self, device="cpu", dtype=None, model_options={}, **kwargs):
super().__init__(device=device, dtype=dtype, model_options=model_options, clip_name="h", clip_model=SD2ClipHModel, **kwargs)

23
comfy/text_encoders/sd2_clip_config.json Normal file
View File

@@ -0,0 +1,23 @@
{
"architectures": [
"CLIPTextModel"
],
"attention_dropout": 0.0,
"bos_token_id": 0,
"dropout": 0.0,
"eos_token_id": 49407,
"hidden_act": "gelu",
"hidden_size": 1024,
"initializer_factor": 1.0,
"initializer_range": 0.02,
"intermediate_size": 4096,
"layer_norm_eps": 1e-05,
"max_position_embeddings": 77,
"model_type": "clip_text_model",
"num_attention_heads": 16,
"num_hidden_layers": 24,
"pad_token_id": 1,
"projection_dim": 1024,
"torch_dtype": "float32",
"vocab_size": 49408
}

167
comfy/text_encoders/sd3_clip.py Normal file
View File

@@ -0,0 +1,167 @@
from comfy import sd1_clip
from comfy import sdxl_clip
from transformers import T5TokenizerFast
import comfy.text_encoders.t5
import torch
import os
import comfy.model_management
import logging
class T5XXLModel(sd1_clip.SDClipModel):
def __init__(self, device="cpu", layer="last", layer_idx=None, dtype=None, attention_mask=False, model_options={}):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "t5_config_xxl.json")
t5xxl_scaled_fp8 = model_options.get("t5xxl_scaled_fp8", None)
if t5xxl_scaled_fp8 is not None:
model_options = model_options.copy()
model_options["scaled_fp8"] = t5xxl_scaled_fp8
super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config=textmodel_json_config, dtype=dtype, special_tokens={"end": 1, "pad": 0}, model_class=comfy.text_encoders.t5.T5, enable_attention_masks=attention_mask, return_attention_masks=attention_mask, model_options=model_options)
def t5_xxl_detect(state_dict, prefix=""):
out = {}
t5_key = "{}encoder.final_layer_norm.weight".format(prefix)
if t5_key in state_dict:
out["dtype_t5"] = state_dict[t5_key].dtype
scaled_fp8_key = "{}scaled_fp8".format(prefix)
if scaled_fp8_key in state_dict:
out["t5xxl_scaled_fp8"] = state_dict[scaled_fp8_key].dtype
return out
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=77)
class SD3Tokenizer:
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.clip_g = sdxl_clip.SDXLClipGTokenizer(embedding_directory=embedding_directory)
self.t5xxl = T5XXLTokenizer(embedding_directory=embedding_directory)
def tokenize_with_weights(self, text:str, return_word_ids=False):
out = {}
out["g"] = self.clip_g.tokenize_with_weights(text, return_word_ids)
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_g.untokenize(token_weight_pair)
def state_dict(self):
return {}
class SD3ClipModel(torch.nn.Module):
def __init__(self, clip_l=True, clip_g=True, t5=True, dtype_t5=None, t5_attention_mask=False, device="cpu", dtype=None, model_options={}):
super().__init__()
self.dtypes = set()
if clip_l:
clip_l_class = model_options.get("clip_l_class", sd1_clip.SDClipModel)
self.clip_l = clip_l_class(layer="hidden", layer_idx=-2, device=device, dtype=dtype, layer_norm_hidden_state=False, return_projected_pooled=False, model_options=model_options)
self.dtypes.add(dtype)
else:
self.clip_l = None
if clip_g:
self.clip_g = sdxl_clip.SDXLClipG(device=device, dtype=dtype, model_options=model_options)
self.dtypes.add(dtype)
else:
self.clip_g = None
if t5:
dtype_t5 = comfy.model_management.pick_weight_dtype(dtype_t5, dtype, device)
self.t5_attention_mask = t5_attention_mask
self.t5xxl = T5XXLModel(device=device, dtype=dtype_t5, model_options=model_options, attention_mask=self.t5_attention_mask)
self.dtypes.add(dtype_t5)
else:
self.t5xxl = None
logging.debug("Created SD3 text encoder with: clip_l {}, clip_g {}, t5xxl {}:{}".format(clip_l, clip_g, t5, dtype_t5))
def set_clip_options(self, options):
if self.clip_l is not None:
self.clip_l.set_clip_options(options)
if self.clip_g is not None:
self.clip_g.set_clip_options(options)
if self.t5xxl is not None:
self.t5xxl.set_clip_options(options)
def reset_clip_options(self):
if self.clip_l is not None:
self.clip_l.reset_clip_options()
if self.clip_g is not None:
self.clip_g.reset_clip_options()
if self.t5xxl is not None:
self.t5xxl.reset_clip_options()
def encode_token_weights(self, token_weight_pairs):
token_weight_pairs_l = token_weight_pairs["l"]
token_weight_pairs_g = token_weight_pairs["g"]
token_weight_pairs_t5 = token_weight_pairs["t5xxl"]
lg_out = None
pooled = None
out = None
extra = {}
if len(token_weight_pairs_g) > 0 or len(token_weight_pairs_l) > 0:
if self.clip_l is not None:
lg_out, l_pooled = self.clip_l.encode_token_weights(token_weight_pairs_l)
else:
l_pooled = torch.zeros((1, 768), device=comfy.model_management.intermediate_device())
if self.clip_g is not None:
g_out, g_pooled = self.clip_g.encode_token_weights(token_weight_pairs_g)
if lg_out is not None:
cut_to = min(lg_out.shape[1], g_out.shape[1])
lg_out = torch.cat([lg_out[:,:cut_to], g_out[:,:cut_to]], dim=-1)
else:
lg_out = torch.nn.functional.pad(g_out, (768, 0))
else:
g_out = None
g_pooled = torch.zeros((1, 1280), device=comfy.model_management.intermediate_device())
if lg_out is not None:
lg_out = torch.nn.functional.pad(lg_out, (0, 4096 - lg_out.shape[-1]))
out = lg_out
pooled = torch.cat((l_pooled, g_pooled), dim=-1)
if self.t5xxl is not None:
t5_output = self.t5xxl.encode_token_weights(token_weight_pairs_t5)
t5_out, t5_pooled = t5_output[:2]
if self.t5_attention_mask:
extra["attention_mask"] = t5_output[2]["attention_mask"]
if lg_out is not None:
out = torch.cat([lg_out, t5_out], dim=-2)
else:
out = t5_out
if out is None:
out = torch.zeros((1, 77, 4096), device=comfy.model_management.intermediate_device())
if pooled is None:
pooled = torch.zeros((1, 768 + 1280), device=comfy.model_management.intermediate_device())
return out, pooled, extra
def load_sd(self, sd):
if "text_model.encoder.layers.30.mlp.fc1.weight" in sd:
return self.clip_g.load_sd(sd)
elif "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 sd3_clip(clip_l=True, clip_g=True, t5=True, dtype_t5=None, t5xxl_scaled_fp8=None, t5_attention_mask=False):
class SD3ClipModel_(SD3ClipModel):
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__(clip_l=clip_l, clip_g=clip_g, t5=t5, dtype_t5=dtype_t5, t5_attention_mask=t5_attention_mask, device=device, dtype=dtype, model_options=model_options)
return SD3ClipModel_

32
comfy/text_encoders/spiece_tokenizer.py Normal file
View File

@@ -0,0 +1,32 @@
import os
import torch
class SPieceTokenizer:
add_eos = True
@staticmethod
def from_pretrained(path):
return SPieceTokenizer(path)
def __init__(self, tokenizer_path):
import sentencepiece
if torch.is_tensor(tokenizer_path):
tokenizer_path = tokenizer_path.numpy().tobytes()
if isinstance(tokenizer_path, bytes):
self.tokenizer = sentencepiece.SentencePieceProcessor(model_proto=tokenizer_path, add_eos=self.add_eos)
else:
self.tokenizer = sentencepiece.SentencePieceProcessor(model_file=tokenizer_path, add_eos=self.add_eos)
def get_vocab(self):
out = {}
for i in range(self.tokenizer.get_piece_size()):
out[self.tokenizer.id_to_piece(i)] = i
return out
def __call__(self, string):
out = self.tokenizer.encode(string)
return {"input_ids": out}
def serialize_model(self):
return torch.ByteTensor(list(self.tokenizer.serialized_model_proto()))

241
comfy/text_encoders/t5.py Normal file
View File

@@ -0,0 +1,241 @@
import torch
import math
from comfy.ldm.modules.attention import optimized_attention_for_device
import comfy.ops
class T5LayerNorm(torch.nn.Module):
def __init__(self, hidden_size, eps=1e-6, dtype=None, device=None, operations=None):
super().__init__()
self.weight = torch.nn.Parameter(torch.empty(hidden_size, dtype=dtype, device=device))
self.variance_epsilon = eps
def forward(self, x):
variance = x.pow(2).mean(-1, keepdim=True)
x = x * torch.rsqrt(variance + self.variance_epsilon)
return comfy.ops.cast_to_input(self.weight, x) * x
activations = {
"gelu_pytorch_tanh": lambda a: torch.nn.functional.gelu(a, approximate="tanh"),
"relu": torch.nn.functional.relu,
}
class T5DenseActDense(torch.nn.Module):
def __init__(self, model_dim, ff_dim, ff_activation, dtype, device, operations):
super().__init__()
self.wi = operations.Linear(model_dim, ff_dim, bias=False, dtype=dtype, device=device)
self.wo = operations.Linear(ff_dim, model_dim, bias=False, dtype=dtype, device=device)
# self.dropout = nn.Dropout(config.dropout_rate)
self.act = activations[ff_activation]
def forward(self, x):
x = self.act(self.wi(x))
# x = self.dropout(x)
x = self.wo(x)
return x
class T5DenseGatedActDense(torch.nn.Module):
def __init__(self, model_dim, ff_dim, ff_activation, dtype, device, operations):
super().__init__()
self.wi_0 = operations.Linear(model_dim, ff_dim, bias=False, dtype=dtype, device=device)
self.wi_1 = operations.Linear(model_dim, ff_dim, bias=False, dtype=dtype, device=device)
self.wo = operations.Linear(ff_dim, model_dim, bias=False, dtype=dtype, device=device)
# self.dropout = nn.Dropout(config.dropout_rate)
self.act = activations[ff_activation]
def forward(self, x):
hidden_gelu = self.act(self.wi_0(x))
hidden_linear = self.wi_1(x)
x = hidden_gelu * hidden_linear
# x = self.dropout(x)
x = self.wo(x)
return x
class T5LayerFF(torch.nn.Module):
def __init__(self, model_dim, ff_dim, ff_activation, gated_act, dtype, device, operations):
super().__init__()
if gated_act:
self.DenseReluDense = T5DenseGatedActDense(model_dim, ff_dim, ff_activation, dtype, device, operations)
else:
self.DenseReluDense = T5DenseActDense(model_dim, ff_dim, ff_activation, dtype, device, operations)
self.layer_norm = T5LayerNorm(model_dim, dtype=dtype, device=device, operations=operations)
# self.dropout = nn.Dropout(config.dropout_rate)
def forward(self, x):
forwarded_states = self.layer_norm(x)
forwarded_states = self.DenseReluDense(forwarded_states)
# x = x + self.dropout(forwarded_states)
x += forwarded_states
return x
class T5Attention(torch.nn.Module):
def __init__(self, model_dim, inner_dim, num_heads, relative_attention_bias, dtype, device, operations):
super().__init__()
# Mesh TensorFlow initialization to avoid scaling before softmax
self.q = operations.Linear(model_dim, inner_dim, bias=False, dtype=dtype, device=device)
self.k = operations.Linear(model_dim, inner_dim, bias=False, dtype=dtype, device=device)
self.v = operations.Linear(model_dim, inner_dim, bias=False, dtype=dtype, device=device)
self.o = operations.Linear(inner_dim, model_dim, bias=False, dtype=dtype, device=device)
self.num_heads = num_heads
self.relative_attention_bias = None
if relative_attention_bias:
self.relative_attention_num_buckets = 32
self.relative_attention_max_distance = 128
self.relative_attention_bias = operations.Embedding(self.relative_attention_num_buckets, self.num_heads, device=device, dtype=dtype)
@staticmethod
def _relative_position_bucket(relative_position, bidirectional=True, num_buckets=32, max_distance=128):
"""
Adapted from Mesh Tensorflow:
https://github.com/tensorflow/mesh/blob/0cb87fe07da627bf0b7e60475d59f95ed6b5be3d/mesh_tensorflow/transformer/transformer_layers.py#L593
Translate relative position to a bucket number for relative attention. The relative position is defined as
memory_position - query_position, i.e. the distance in tokens from the attending position to the attended-to
position. If bidirectional=False, then positive relative positions are invalid. We use smaller buckets for
small absolute relative_position and larger buckets for larger absolute relative_positions. All relative
positions >=max_distance map to the same bucket. All relative positions <=-max_distance map to the same bucket.
This should allow for more graceful generalization to longer sequences than the model has been trained on
Args:
relative_position: an int32 Tensor
bidirectional: a boolean - whether the attention is bidirectional
num_buckets: an integer
max_distance: an integer
Returns:
a Tensor with the same shape as relative_position, containing int32 values in the range [0, num_buckets)
"""
relative_buckets = 0
if bidirectional:
num_buckets //= 2
relative_buckets += (relative_position > 0).to(torch.long) * num_buckets
relative_position = torch.abs(relative_position)
else:
relative_position = -torch.min(relative_position, torch.zeros_like(relative_position))
# now relative_position is in the range [0, inf)
# half of the buckets are for exact increments in positions
max_exact = num_buckets // 2
is_small = relative_position < max_exact
# The other half of the buckets are for logarithmically bigger bins in positions up to max_distance
relative_position_if_large = max_exact + (
torch.log(relative_position.float() / max_exact)
/ math.log(max_distance / max_exact)
* (num_buckets - max_exact)
).to(torch.long)
relative_position_if_large = torch.min(
relative_position_if_large, torch.full_like(relative_position_if_large, num_buckets - 1)
)
relative_buckets += torch.where(is_small, relative_position, relative_position_if_large)
return relative_buckets
def compute_bias(self, query_length, key_length, device, dtype):
"""Compute binned relative position bias"""
context_position = torch.arange(query_length, dtype=torch.long, device=device)[:, None]
memory_position = torch.arange(key_length, dtype=torch.long, device=device)[None, :]
relative_position = memory_position - context_position # shape (query_length, key_length)
relative_position_bucket = self._relative_position_bucket(
relative_position, # shape (query_length, key_length)
bidirectional=True,
num_buckets=self.relative_attention_num_buckets,
max_distance=self.relative_attention_max_distance,
)
values = self.relative_attention_bias(relative_position_bucket, out_dtype=dtype) # shape (query_length, key_length, num_heads)
values = values.permute([2, 0, 1]).unsqueeze(0) # shape (1, num_heads, query_length, key_length)
return values
def forward(self, x, mask=None, past_bias=None, optimized_attention=None):
q = self.q(x)
k = self.k(x)
v = self.v(x)
if self.relative_attention_bias is not None:
past_bias = self.compute_bias(x.shape[1], x.shape[1], x.device, x.dtype)
if past_bias is not None:
if mask is not None:
mask = mask + past_bias
else:
mask = past_bias
out = optimized_attention(q, k * ((k.shape[-1] / self.num_heads) ** 0.5), v, self.num_heads, mask)
return self.o(out), past_bias
class T5LayerSelfAttention(torch.nn.Module):
def __init__(self, model_dim, inner_dim, ff_dim, num_heads, relative_attention_bias, dtype, device, operations):
super().__init__()
self.SelfAttention = T5Attention(model_dim, inner_dim, num_heads, relative_attention_bias, dtype, device, operations)
self.layer_norm = T5LayerNorm(model_dim, dtype=dtype, device=device, operations=operations)
# self.dropout = nn.Dropout(config.dropout_rate)
def forward(self, x, mask=None, past_bias=None, optimized_attention=None):
normed_hidden_states = self.layer_norm(x)
output, past_bias = self.SelfAttention(self.layer_norm(x), mask=mask, past_bias=past_bias, optimized_attention=optimized_attention)
# x = x + self.dropout(attention_output)
x += output
return x, past_bias
class T5Block(torch.nn.Module):
def __init__(self, model_dim, inner_dim, ff_dim, ff_activation, gated_act, num_heads, relative_attention_bias, dtype, device, operations):
super().__init__()
self.layer = torch.nn.ModuleList()
self.layer.append(T5LayerSelfAttention(model_dim, inner_dim, ff_dim, num_heads, relative_attention_bias, dtype, device, operations))
self.layer.append(T5LayerFF(model_dim, ff_dim, ff_activation, gated_act, dtype, device, operations))
def forward(self, x, mask=None, past_bias=None, optimized_attention=None):
x, past_bias = self.layer[0](x, mask, past_bias, optimized_attention)
x = self.layer[-1](x)
return x, past_bias
class T5Stack(torch.nn.Module):
def __init__(self, num_layers, model_dim, inner_dim, ff_dim, ff_activation, gated_act, num_heads, relative_attention, dtype, device, operations):
super().__init__()
self.block = torch.nn.ModuleList(
[T5Block(model_dim, inner_dim, ff_dim, ff_activation, gated_act, num_heads, relative_attention_bias=((not relative_attention) or (i == 0)), dtype=dtype, device=device, operations=operations) for i in range(num_layers)]
)
self.final_layer_norm = T5LayerNorm(model_dim, dtype=dtype, device=device, operations=operations)
# self.dropout = nn.Dropout(config.dropout_rate)
def forward(self, x, attention_mask=None, intermediate_output=None, final_layer_norm_intermediate=True, dtype=None):
mask = None
if attention_mask is not None:
mask = 1.0 - attention_mask.to(x.dtype).reshape((attention_mask.shape[0], 1, -1, attention_mask.shape[-1])).expand(attention_mask.shape[0], 1, attention_mask.shape[-1], attention_mask.shape[-1])
mask = mask.masked_fill(mask.to(torch.bool), float("-inf"))
intermediate = None
optimized_attention = optimized_attention_for_device(x.device, mask=attention_mask is not None, small_input=True)
past_bias = None
for i, l in enumerate(self.block):
x, past_bias = l(x, mask, past_bias, optimized_attention)
if i == intermediate_output:
intermediate = x.clone()
x = self.final_layer_norm(x)
if intermediate is not None and final_layer_norm_intermediate:
intermediate = self.final_layer_norm(intermediate)
return x, intermediate
class T5(torch.nn.Module):
def __init__(self, config_dict, dtype, device, operations):
super().__init__()
self.num_layers = config_dict["num_layers"]
model_dim = config_dict["d_model"]
self.encoder = T5Stack(self.num_layers, model_dim, model_dim, config_dict["d_ff"], config_dict["dense_act_fn"], config_dict["is_gated_act"], config_dict["num_heads"], config_dict["model_type"] != "umt5", dtype, device, operations)
self.dtype = dtype
self.shared = operations.Embedding(config_dict["vocab_size"], model_dim, device=device, dtype=dtype)
def get_input_embeddings(self):
return self.shared
def set_input_embeddings(self, embeddings):
self.shared = embeddings
def forward(self, input_ids, *args, **kwargs):
x = self.shared(input_ids, out_dtype=kwargs.get("dtype", torch.float32))
if self.dtype not in [torch.float32, torch.float16, torch.bfloat16]:
x = torch.nan_to_num(x) #Fix for fp8 T5 base
return self.encoder(x, *args, **kwargs)

22
comfy/text_encoders/t5_config_base.json Normal file
View File

@@ -0,0 +1,22 @@
{
"d_ff": 3072,
"d_kv": 64,
"d_model": 768,
"decoder_start_token_id": 0,
"dropout_rate": 0.1,
"eos_token_id": 1,
"dense_act_fn": "relu",
"initializer_factor": 1.0,
"is_encoder_decoder": true,
"is_gated_act": false,
"layer_norm_epsilon": 1e-06,
"model_type": "t5",
"num_decoder_layers": 12,
"num_heads": 12,
"num_layers": 12,
"output_past": true,
"pad_token_id": 0,
"relative_attention_num_buckets": 32,
"tie_word_embeddings": false,
"vocab_size": 32128
}

22
comfy/text_encoders/t5_config_xxl.json Normal file
View File

@@ -0,0 +1,22 @@
{
"d_ff": 10240,
"d_kv": 64,
"d_model": 4096,
"decoder_start_token_id": 0,
"dropout_rate": 0.1,
"eos_token_id": 1,
"dense_act_fn": "gelu_pytorch_tanh",
"initializer_factor": 1.0,
"is_encoder_decoder": true,
"is_gated_act": true,
"layer_norm_epsilon": 1e-06,
"model_type": "t5",
"num_decoder_layers": 24,
"num_heads": 64,
"num_layers": 24,
"output_past": true,
"pad_token_id": 0,
"relative_attention_num_buckets": 32,
"tie_word_embeddings": false,
"vocab_size": 32128
}

22
comfy/text_encoders/t5_pile_config_xl.json Normal file
View File

@@ -0,0 +1,22 @@
{
"d_ff": 5120,
"d_kv": 64,
"d_model": 2048,
"decoder_start_token_id": 0,
"dropout_rate": 0.1,
"eos_token_id": 2,
"dense_act_fn": "gelu_pytorch_tanh",
"initializer_factor": 1.0,
"is_encoder_decoder": true,
"is_gated_act": true,
"layer_norm_epsilon": 1e-06,
"model_type": "umt5",
"num_decoder_layers": 24,
"num_heads": 32,
"num_layers": 24,
"output_past": true,
"pad_token_id": 1,
"relative_attention_num_buckets": 32,
"tie_word_embeddings": false,
"vocab_size": 32128
}

BIN
comfy/text_encoders/t5_pile_tokenizer/tokenizer.model Normal file
View File

Binary file not shown.

125
comfy/text_encoders/t5_tokenizer/special_tokens_map.json Normal file
View File

@@ -0,0 +1,125 @@
{
"additional_special_tokens": [
"<extra_id_0>",
"<extra_id_1>",
"<extra_id_2>",
"<extra_id_3>",
"<extra_id_4>",
"<extra_id_5>",
"<extra_id_6>",
"<extra_id_7>",
"<extra_id_8>",
"<extra_id_9>",
"<extra_id_10>",
"<extra_id_11>",
"<extra_id_12>",
"<extra_id_13>",
"<extra_id_14>",
"<extra_id_15>",
"<extra_id_16>",
"<extra_id_17>",
"<extra_id_18>",
"<extra_id_19>",
"<extra_id_20>",
"<extra_id_21>",
"<extra_id_22>",
"<extra_id_23>",
"<extra_id_24>",
"<extra_id_25>",
"<extra_id_26>",
"<extra_id_27>",
"<extra_id_28>",
"<extra_id_29>",
"<extra_id_30>",
"<extra_id_31>",
"<extra_id_32>",
"<extra_id_33>",
"<extra_id_34>",
"<extra_id_35>",
"<extra_id_36>",
"<extra_id_37>",
"<extra_id_38>",
"<extra_id_39>",
"<extra_id_40>",
"<extra_id_41>",
"<extra_id_42>",
"<extra_id_43>",
"<extra_id_44>",
"<extra_id_45>",
"<extra_id_46>",
"<extra_id_47>",
"<extra_id_48>",
"<extra_id_49>",
"<extra_id_50>",
"<extra_id_51>",
"<extra_id_52>",
"<extra_id_53>",
"<extra_id_54>",
"<extra_id_55>",
"<extra_id_56>",
"<extra_id_57>",
"<extra_id_58>",
"<extra_id_59>",
"<extra_id_60>",
"<extra_id_61>",
"<extra_id_62>",
"<extra_id_63>",
"<extra_id_64>",
"<extra_id_65>",
"<extra_id_66>",
"<extra_id_67>",
"<extra_id_68>",
"<extra_id_69>",
"<extra_id_70>",
"<extra_id_71>",
"<extra_id_72>",
"<extra_id_73>",
"<extra_id_74>",
"<extra_id_75>",
"<extra_id_76>",
"<extra_id_77>",
"<extra_id_78>",
"<extra_id_79>",
"<extra_id_80>",
"<extra_id_81>",
"<extra_id_82>",
"<extra_id_83>",
"<extra_id_84>",
"<extra_id_85>",
"<extra_id_86>",
"<extra_id_87>",
"<extra_id_88>",
"<extra_id_89>",
"<extra_id_90>",
"<extra_id_91>",
"<extra_id_92>",
"<extra_id_93>",
"<extra_id_94>",
"<extra_id_95>",
"<extra_id_96>",
"<extra_id_97>",
"<extra_id_98>",
"<extra_id_99>"
],
"eos_token": {
"content": "</s>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false
},
"pad_token": {
"content": "<pad>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false
},
"unk_token": {
"content": "<unk>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false
}
}

129428
comfy/text_encoders/t5_tokenizer/tokenizer.json Normal file
View File

File diff suppressed because one or more lines are too long

939
comfy/text_encoders/t5_tokenizer/tokenizer_config.json Normal file
View File

@@ -0,0 +1,939 @@
{
"added_tokens_decoder": {
"0": {
"content": "<pad>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"1": {
"content": "</s>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"2": {
"content": "<unk>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32000": {
"content": "<extra_id_99>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32001": {
"content": "<extra_id_98>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32002": {
"content": "<extra_id_97>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32003": {
"content": "<extra_id_96>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32004": {
"content": "<extra_id_95>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32005": {
"content": "<extra_id_94>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32006": {
"content": "<extra_id_93>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32007": {
"content": "<extra_id_92>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32008": {
"content": "<extra_id_91>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32009": {
"content": "<extra_id_90>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32010": {
"content": "<extra_id_89>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32011": {
"content": "<extra_id_88>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32012": {
"content": "<extra_id_87>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32013": {
"content": "<extra_id_86>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32014": {
"content": "<extra_id_85>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32015": {
"content": "<extra_id_84>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32016": {
"content": "<extra_id_83>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32017": {
"content": "<extra_id_82>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32018": {
"content": "<extra_id_81>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32019": {
"content": "<extra_id_80>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32020": {
"content": "<extra_id_79>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32021": {
"content": "<extra_id_78>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32022": {
"content": "<extra_id_77>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32023": {
"content": "<extra_id_76>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32024": {
"content": "<extra_id_75>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32025": {
"content": "<extra_id_74>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32026": {
"content": "<extra_id_73>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32027": {
"content": "<extra_id_72>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32028": {
"content": "<extra_id_71>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32029": {
"content": "<extra_id_70>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32030": {
"content": "<extra_id_69>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32031": {
"content": "<extra_id_68>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32032": {
"content": "<extra_id_67>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32033": {
"content": "<extra_id_66>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32034": {
"content": "<extra_id_65>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32035": {
"content": "<extra_id_64>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32036": {
"content": "<extra_id_63>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32037": {
"content": "<extra_id_62>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32038": {
"content": "<extra_id_61>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32039": {
"content": "<extra_id_60>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32040": {
"content": "<extra_id_59>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32041": {
"content": "<extra_id_58>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32042": {
"content": "<extra_id_57>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32043": {
"content": "<extra_id_56>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32044": {
"content": "<extra_id_55>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32045": {
"content": "<extra_id_54>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32046": {
"content": "<extra_id_53>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32047": {
"content": "<extra_id_52>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32048": {
"content": "<extra_id_51>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32049": {
"content": "<extra_id_50>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32050": {
"content": "<extra_id_49>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32051": {
"content": "<extra_id_48>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32052": {
"content": "<extra_id_47>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32053": {
"content": "<extra_id_46>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32054": {
"content": "<extra_id_45>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32055": {
"content": "<extra_id_44>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32056": {
"content": "<extra_id_43>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32057": {
"content": "<extra_id_42>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32058": {
"content": "<extra_id_41>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32059": {
"content": "<extra_id_40>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32060": {
"content": "<extra_id_39>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32061": {
"content": "<extra_id_38>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32062": {
"content": "<extra_id_37>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32063": {
"content": "<extra_id_36>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32064": {
"content": "<extra_id_35>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32065": {
"content": "<extra_id_34>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32066": {
"content": "<extra_id_33>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32067": {
"content": "<extra_id_32>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32068": {
"content": "<extra_id_31>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32069": {
"content": "<extra_id_30>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32070": {
"content": "<extra_id_29>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32071": {
"content": "<extra_id_28>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32072": {
"content": "<extra_id_27>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32073": {
"content": "<extra_id_26>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32074": {
"content": "<extra_id_25>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32075": {
"content": "<extra_id_24>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32076": {
"content": "<extra_id_23>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32077": {
"content": "<extra_id_22>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32078": {
"content": "<extra_id_21>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32079": {
"content": "<extra_id_20>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32080": {
"content": "<extra_id_19>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32081": {
"content": "<extra_id_18>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32082": {
"content": "<extra_id_17>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32083": {
"content": "<extra_id_16>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32084": {
"content": "<extra_id_15>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32085": {
"content": "<extra_id_14>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32086": {
"content": "<extra_id_13>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32087": {
"content": "<extra_id_12>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32088": {
"content": "<extra_id_11>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32089": {
"content": "<extra_id_10>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32090": {
"content": "<extra_id_9>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32091": {
"content": "<extra_id_8>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32092": {
"content": "<extra_id_7>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32093": {
"content": "<extra_id_6>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32094": {
"content": "<extra_id_5>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32095": {
"content": "<extra_id_4>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32096": {
"content": "<extra_id_3>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32097": {
"content": "<extra_id_2>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32098": {
"content": "<extra_id_1>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"32099": {
"content": "<extra_id_0>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
}
},
"additional_special_tokens": [
"<extra_id_0>",
"<extra_id_1>",
"<extra_id_2>",
"<extra_id_3>",
"<extra_id_4>",
"<extra_id_5>",
"<extra_id_6>",
"<extra_id_7>",
"<extra_id_8>",
"<extra_id_9>",
"<extra_id_10>",
"<extra_id_11>",
"<extra_id_12>",
"<extra_id_13>",
"<extra_id_14>",
"<extra_id_15>",
"<extra_id_16>",
"<extra_id_17>",
"<extra_id_18>",
"<extra_id_19>",
"<extra_id_20>",
"<extra_id_21>",
"<extra_id_22>",
"<extra_id_23>",
"<extra_id_24>",
"<extra_id_25>",
"<extra_id_26>",
"<extra_id_27>",
"<extra_id_28>",
"<extra_id_29>",
"<extra_id_30>",
"<extra_id_31>",
"<extra_id_32>",
"<extra_id_33>",
"<extra_id_34>",
"<extra_id_35>",
"<extra_id_36>",
"<extra_id_37>",
"<extra_id_38>",
"<extra_id_39>",
"<extra_id_40>",
"<extra_id_41>",
"<extra_id_42>",
"<extra_id_43>",
"<extra_id_44>",
"<extra_id_45>",
"<extra_id_46>",
"<extra_id_47>",
"<extra_id_48>",
"<extra_id_49>",
"<extra_id_50>",
"<extra_id_51>",
"<extra_id_52>",
"<extra_id_53>",
"<extra_id_54>",
"<extra_id_55>",
"<extra_id_56>",
"<extra_id_57>",
"<extra_id_58>",
"<extra_id_59>",
"<extra_id_60>",
"<extra_id_61>",
"<extra_id_62>",
"<extra_id_63>",
"<extra_id_64>",
"<extra_id_65>",
"<extra_id_66>",
"<extra_id_67>",
"<extra_id_68>",
"<extra_id_69>",
"<extra_id_70>",
"<extra_id_71>",
"<extra_id_72>",
"<extra_id_73>",
"<extra_id_74>",
"<extra_id_75>",
"<extra_id_76>",
"<extra_id_77>",
"<extra_id_78>",
"<extra_id_79>",
"<extra_id_80>",
"<extra_id_81>",
"<extra_id_82>",
"<extra_id_83>",
"<extra_id_84>",
"<extra_id_85>",
"<extra_id_86>",
"<extra_id_87>",
"<extra_id_88>",
"<extra_id_89>",
"<extra_id_90>",
"<extra_id_91>",
"<extra_id_92>",
"<extra_id_93>",
"<extra_id_94>",
"<extra_id_95>",
"<extra_id_96>",
"<extra_id_97>",
"<extra_id_98>",
"<extra_id_99>"
],
"clean_up_tokenization_spaces": true,
"eos_token": "</s>",
"extra_ids": 100,
"legacy": false,
"model_max_length": 512,
"pad_token": "<pad>",
"sp_model_kwargs": {},
"tokenizer_class": "T5Tokenizer",
"unk_token": "<unk>"
}