Functions — github.com/tdrussell/diffusion-pipe

hub / github.com/tdrussell/diffusion-pipe / functions

Functions1,157 in github.com/tdrussell/diffusion-pipe

⨍ Functions 1,157◇ Types & classes 222

Methodsave_model

(self, save_dir, state_dict)

models/hunyuan_image.py:136

Methodsave_model

(self, save_dir, diffusers_sd)

models/flux.py:255

Methodsave_model

(self, save_dir, state_dict)

models/qwen_image.py:290

Methodsave_model

(self, save_dir, state_dict)

models/wan/wan.py:264

Methodseq_dim

(self)

models/cosmos_predict2_modeling.py:469

Methodseq_dim

(self)

models/cosmos_predict2_modeling.py:617

Methodset_eval_quantile

(self, quantile)

utils/dataset.py:1316

Functionsingle_stream_forward

(self, x: Tensor, vec: Tensor, pe: Tensor, attn_mask=None, modulation_dims=None, transformer_options={})

utils/patches.py:366

Methodstate_dict

Returns a dictionary containing the current state of the tensor.

optimizers/optimizer_utils.py:254

Methodstate_dict

(self, *args, **kwargs)

optimizers/automagic.py:359

Methodstate_dict

(self)

optimizers/gradient_release.py:16

Methodstep

Performs a single optimization step. Arguments: closure (`Callable`, *optional*): A closure that reevaluates the model a

optimizers/generic_optim.py:364

Methodstep

Performs a single optimization step Arguments: closure (callable, optional): A closure that reevaluates the model

optimizers/automagic.py:178

Methodstep

(self)

optimizers/gradient_release.py:23

Functionstochastic_grad_accummulation

(param)

optimizers/optimizer_utils.py:272

Methodswap_scale_shift

(weight)

models/flux.py:278

Functiontext_embedding_map_fn

(example, rank)

utils/dataset.py:1114

Functionto

(self, *args, **kwargs)

train.py:571

Methodto_layers

(self)

models/hidream.py:261

models/ernie_image.py:25

Methodto_layers

(self)

models/hunyuan_video_15.py:100

models/auraflow.py:148

Methodto_layers

(self)

models/lumina_2.py:182

models/ltx_video.py:195

Methodto_layers

(self)

models/cosmos_predict2.py:409

Methodto_layers

(self)

models/hunyuan_video.py:481

models/hunyuan_image.py:323

Methodto_layers

(self)

models/ideogram4.py:35

models/omnigen2.py:181

models/qwen_image.py:479

Methodto_layers

(self)

models/wan/wan.py:376

Functiontokenize_with_weights

(self, text:str, return_word_ids=False, **kwargs)

models/base.py:467

Functiontorch_attention_op

Computes multi-head attention using PyTorch's native implementation. This function provides a PyTorch backend alternative to Transformer Engine's

models/cosmos_predict2_modeling.py:275

Functiontrain_schedule_steps

Functionunsloth_checkpoint

(function, *args)

utils/unsloth_utils.py:70

Methodupdate_step

(self, group, p, gindex, pindex)

optimizers/adamw_8bit.py:17

Methodvae_encode_crop_pixels

(self, pixels)

models/base.py:490

Methodvae_encode_crop_pixels

Functionxlm_roberta_large

XLMRobertaLarge adapted from Huggingface.

models/wan/xlm_roberta.py:144

(self, set_to_none=True)

optimizers/gradient_release.py:26

Functionzeropower_via_newtonschulz5

Newton-Schulz iteration to compute the zeroth power / orthogonalization of G. We opt to use a quintic iteration whose coefficients are select

optimizers/generic_optim.py:153

← previous1,101–1,157 of 1,157, ranked by callers