ComfyUI/comfy/text_encoders/cosmos.py
comfyanonymous e1e50e017c Make old scaled fp8 format use the new mixed quant ops system.
Since model saving is not implemented for the mixed quant system this
breaks model saving for every scaled fp8 model which needs to be fixed
before this gets merged.
2025-11-30 00:33:01 -05:00

43 lines
2.5 KiB
Python

from comfy import sd1_clip
import comfy.text_encoders.t5
import os
from transformers import T5TokenizerFast
class T5XXLModel(sd1_clip.SDClipModel):
def __init__(self, device="cpu", layer="last", layer_idx=None, dtype=None, attention_mask=True, model_options={}):
textmodel_json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "t5_old_config_xxl.json")
t5xxl_quantization_metadata = model_options.get("t5xxl_quantization_metadata", None)
if t5xxl_quantization_metadata is not None:
model_options = model_options.copy()
model_options["quantization_metadata"] = t5xxl_quantization_metadata
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, zero_out_masked=attention_mask, model_options=model_options)
class CosmosT5XXL(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=1024, embedding_key='t5xxl', tokenizer_class=T5TokenizerFast, has_start_token=False, pad_to_max_length=False, max_length=99999999, min_length=512, tokenizer_data=tokenizer_data)
class CosmosT5Tokenizer(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 te(dtype_t5=None, t5_quantization_metadata=None):
class CosmosTEModel_(CosmosT5XXL):
def __init__(self, device="cpu", dtype=None, model_options={}):
if t5_quantization_metadata is not None:
model_options = model_options.copy()
model_options["t5xxl_quantization_metadata"] = t5_quantization_metadata
if dtype is None:
dtype = dtype_t5
super().__init__(device=device, dtype=dtype, model_options=model_options)
return CosmosTEModel_