Functions — github.com/PixArt-alpha/PixArt-sigma

hub / github.com/PixArt-alpha/PixArt-sigma / functions

Functions537 in github.com/PixArt-alpha/PixArt-sigma

⨍ Functions 537◇ Types & classes 71

↓ 1 callersMethodtoken_drop

Drops labels to enable classifier-free guidance.

diffusion/model/nets/PixArt_blocks.py:359

↓ 1 callersMethodtoken_drop

Drops labels to enable classifier-free guidance.

diffusion/model/nets/PixArt_blocks.py:389

↓ 1 callersMethodtoken_drop

Drops labels to enable classifier-free guidance.

diffusion/model/nets/PixArt_blocks.py:422

↓ 1 callersFunctiontokenize_captions

(examples, is_train=True, proportion_empty_prompts=0., max_length=120)

train_scripts/train_pixart_lora_hf.py:672

↓ 1 callersMethodtrace_frames

(self)

diffusion/utils/misc.py:219

↓ 1 callersFunctiontrain

()

train_scripts/train.py:125

↓ 1 callersFunctiontrain

()

train_scripts/train_pixart_lcm.py:176

↓ 1 callersFunctiontrain_dreambooth

(pretrained_model_path, images_folder, captions_folder,

train_scripts/train_dreambooth_lora.py:356

↓ 1 callersMethodunpatchify

x: (N, T, patch_size**2 * C) imgs: (N, H, W, C)

diffusion/model/nets/PixArtMS.py:236

↓ 1 callersMethodunpatchify

x: (N, T, patch_size**2 * C) imgs: (N, H, W, C)

diffusion/model/nets/PixArt.py:199

↓ 1 callersMethodupdate_fn

(p, grad, exp_avg, lr, wd, beta1, beta2)

diffusion/utils/optimizer.py:191

↓ 1 callersMethodupdate_with_all_losses

Update the reweighting using losses from a model. Sub-classes should override this method to update the reweighting using los

diffusion/model/timestep_sampler.py:106

↓ 1 callersFunctionvalidation_and_save

(pretrained_model_path : str, transformer, validation_prompt : str, val_embeds, generator, output_folder, epoc

train_scripts/train_dreambooth_lora.py:318

↓ 1 callersFunctionvisualize

(items, bs, sample_steps, cfg_scale)

scripts/inference.py:54

↓ 1 callersMethodweights

Get a numpy array of weights, one per diffusion step. The weights needn't be normalized, but must be positive.

diffusion/model/timestep_sampler.py:38

↓ 1 callersFunctionwrite_entry

(json_entries, folder, image_path, caption_path, image_filename, intern_imgs_path)

tools/convert_images_to_json.py:63

Method__call__

(self, x, timestep, **kwargs)

diffusion/model/respace.py:129

Method__call__

( self, attn: Attention, hidden_states: torch.FloatTensor, encoder_hidden_stat

scripts/DMD/transformer_train/attention_processor.py:9

Method__enter__

(self)

diffusion/model/utils.py:385

Method__exit__

(self, exc_type, exc_value, traceback)

diffusion/model/utils.py:388

Method__getattr__

(self, name)

diffusion/data/datasets/InternalData.py:156

Method__getattr__

(self, name)

diffusion/data/datasets/InternalData.py:326

Method__getitem__

(self, idx)

tools/extract_features.py:91

Method__getitem__

(self, idx)

diffusion/data/datasets/InternalData.py:114

Method__getitem__

(self, idx)

diffusion/data/datasets/InternalData.py:282

Method__getitem__

(self, idx)

diffusion/data/datasets/InternalData_ms.py:162

Method__getitem__

(self, idx)

diffusion/data/datasets/InternalData_ms.py:343

Method__getitem__

(self, idx)

diffusion/data/datasets/dmd.py:177

Method__getitem__

(self, idx)

train_scripts/train_dreambooth_lora.py:267

Method__init__

(self, root, # Notice: need absolute path here image_list_json=['data_info.

tools/extract_features.py:37

Method__init__

( self, num_train_timesteps: int = 1000, beta_start: float = 0.0001,

diffusion/lcm_scheduler.py:176

Method__init__

(self, model, noise_schedule="linear", diffusion_steps=1000,

diffusion/sa_sampler.py:11

Method__init__

( self, num_train_timesteps: int = 1000, beta_start: float = 0.0001,

diffusion/sa_solver_diffusers.py:139

Method__init__

(self, *args, **kwargs)

diffusion/utils/data_sampler.py:80

Method__init__

(self, num_tasks, log_interval=1, desc="Process")

diffusion/utils/misc.py:86

Method__init__

(self, model, max_frames_to_save=21, trace_batch_nums=[], abort_after_batch_num=None)

diffusion/utils/misc.py:193

Method__init__

(self, *args, **kwargs)

diffusion/utils/optimizer.py:244

Method__init__

Create a wrapper class for the forward SDE (VP type). *** Update: We support discrete-time diffusion models by implementing a picewis

diffusion/model/dpm_solver.py:6

Method__init__

Construct a DPM-Solver. We support both DPM-Solver (`algorithm_type="dpmsolver"`) and DPM-Solver++ (`algorithm_type="dpmsolver++"`).

diffusion/model/dpm_solver.py:340

Method__init__

(self, diffusion)

diffusion/model/timestep_sampler.py:63

Method__init__

(self, diffusion, history_per_term=10, uniform_prob=0.001)

diffusion/model/timestep_sampler.py:121

Method__init__

(self, device, seeds)

diffusion/model/utils.py:426

Method__init__

(self, device, dir_or_name='t5-v1_1-xxl', *, local_cache=False, cache_dir=None, hf_token=None, use_text_prepro

diffusion/model/t5.py:18

Method__init__

Thanks to DPM-Solver for their code base

diffusion/model/sa_solver.py:8

Method__init__

Construct a SA-Solver The default value for algorithm_type is "data_prediction" and we recommend not to change it to "noise_

diffusion/model/sa_solver.py:326

Method__init__

( self, *, betas, model_mean_type, model_var_type, loss_type,

diffusion/model/gaussian_diffusion.py:154

Method__init__

(self, use_timesteps, **kwargs)

diffusion/model/respace.py:73

Method__init__

( self, patch_size=16, in_chans=3, embed_dim=768,

diffusion/model/nets/PixArtMS.py:25

Method__init__

(self, hidden_size, num_heads, mlp_ratio=4.0, drop_path=0., input_size=None, sampling=None, s

diffusion/model/nets/PixArtMS.py:54

Method__init__

(self, hidden_size, num_heads, mlp_ratio=4.0, drop_path=0, input_size=None, sampling=None, sr

diffusion/model/nets/PixArt.py:30

Method__init__

(self, d_model, num_heads, attn_drop=0., proj_drop=0., **block_kwargs)

diffusion/model/nets/PixArt_blocks.py:29

Method__init__

Args: dim (int): Number of input channels. num_heads (int): Number of attention heads. qkv_bias (bool: I

diffusion/model/nets/PixArt_blocks.py:64

Method__init__

(self, hidden_size, patch_size, out_channels)

diffusion/model/nets/PixArt_blocks.py:210

Method__init__

(self, final_hidden_size, c_emb_size, patch_size, out_channels)

diffusion/model/nets/PixArt_blocks.py:229

Method__init__

(self, hidden_size, decoder_hidden_size)

diffusion/model/nets/PixArt_blocks.py:249

Method__init__

(self, hidden_size, frequency_embedding_size=256)

diffusion/model/nets/PixArt_blocks.py:272

Method__init__

(self, hidden_size, frequency_embedding_size=256)

diffusion/model/nets/PixArt_blocks.py:317

Method__init__

(self, num_classes, hidden_size, dropout_prob)

diffusion/model/nets/PixArt_blocks.py:352

Method__init__

(self, in_channels, hidden_size, uncond_prob, act_layer=nn.GELU(approximate='tanh'), token_num=120)

diffusion/model/nets/PixArt_blocks.py:383

Method__init__

(self, in_channels, hidden_size, uncond_prob, act_layer=nn.GELU(approximate='tanh'), token_num=120)

diffusion/model/nets/PixArt_blocks.py:415

Method__init__

(self, config)

diffusion/model/llava/llava_mpt.py:176

Method__init__

(self, d_model: int, n_heads: int, expansion_ratio: int, attn_config: Dict={'attn_type': 'multihead_attention'

diffusion/model/llava/mpt/blocks.py:22

Method__init__

(self, config: MPTConfig)

diffusion/model/llava/mpt/modeling_mpt.py:219

Method__init__

The MPT configuration class. Args: d_model (int): The size of the embedding dimension of the model. n_heads (int): Th

diffusion/model/llava/mpt/configuration_mpt.py:10

Method__init__

(self, normalized_shape, eps=1e-05, elementwise_affine=True, device=None, dtype=None)

diffusion/model/llava/mpt/norm.py:16

Method__init__

(self, normalized_shape, eps=1e-05, weight=True, dtype=None, device=None)

diffusion/model/llava/mpt/norm.py:48

Method__init__

(self, d_model: int, n_heads: int, attn_impl: str='triton', clip_qkv: Optional[float]=None, qk_ln: bool=False,

diffusion/model/llava/mpt/attention.py:181

Method__init__

(self, root, image_list_json='data_info.json', transform=No

diffusion/data/datasets/InternalData.py:23

Method__init__

(self, root, image_list_json='data_info.json', transform=No

diffusion/data/datasets/InternalData.py:163

Method__init__

(self, root, image_list_json='data_info.json', transform=No

diffusion/data/datasets/InternalData_ms.py:22

Method__init__

(self, root, image_list_json='data_info.json', transform=No

diffusion/data/datasets/InternalData_ms.py:174

Method__init__

:param root: the root of saving txt features ./data/data/ :param latent_root: the root of saving latent image pairs :param im

diffusion/data/datasets/dmd.py:22

Method__init__

(self, output_folder : str, class_id : str, unique_id : str, batch_size : int, num_repeats)

train_scripts/train_dreambooth_lora.py:249

Method__init__

(self, alpha_cumprods, timesteps=1000, ddim_timesteps=50)

train_scripts/train_pixart_lcm.py:78

Method__init__

( self, tokenizer: T5Tokenizer, text_encoder: T5EncoderModel,

scripts/diffusers_patches.py:484

Method__iter__

(self)

diffusion/utils/data_sampler.py:51

Method__iter__

(self)

diffusion/utils/data_sampler.py:98

Method__len__

(self)

diffusion/lcm_scheduler.py:457

Method__len__

(self)

diffusion/sa_solver_diffusers.py:855

Method__len__

(self)

diffusion/data/datasets/InternalData.py:153

Method__len__

(self)

diffusion/data/datasets/InternalData.py:323

Method__len__

(self)

diffusion/data/datasets/dmd.py:103

Method__len__

(self)

train_scripts/train_dreambooth_lora.py:264

Method_basic_init

(module)

diffusion/model/nets/PixArtMS.py:252

Method_basic_init

(module)

diffusion/model/nets/PixArt.py:216

Method_get_variance

(self, timestep, prev_timestep)

diffusion/lcm_scheduler.py:244

Function_inner

(folder)

scripts/DMD/transformer_train/utils.py:84

Method_register_forward_hook

(self, module)

diffusion/utils/misc.py:290

Method_reorder_cache

Used by HuggingFace generate when using beam search with kv-caching. See https://github.com/huggingface/transformers/blob/3ec7a47664ebe40c40f

diffusion/model/llava/mpt/modeling_mpt.py:299

Method_scale_timesteps

(self, t)

diffusion/model/respace.py:117

Method_set_gradient_checkpointing

(self, module, value=False)

diffusion/model/llava/llava_mpt.py:195

Method_threshold_sample

"Dynamic thresholding: At each sampling step we set s to a certain percentile absolute pixel value in xt0 (the prediction of x_0 at t

diffusion/lcm_scheduler.py:255

Functionablation_sampler

( net, latents, class_labels=None, cfg_scale=None, feat=None, randn_like=torch.randn_like, num

diffusion/model/edm_sample.py:54

Methodactivation_checkpointing_fn

(self, module)

diffusion/model/llava/mpt/modeling_mpt.py:214

Methodactivation_checkpointing_fn

(self, module)

diffusion/model/llava/mpt/modeling_mpt.py:275

Functionadd_decomposed_rel_pos

Calculate decomposed Relative Positional Embeddings from :paper:`mvitv2`. https://github.com/facebookresearch/mvit/blob/19786631e330df9f3622e

diffusion/model/utils.py:154

Methodadd_noise

( self, original_samples: torch.FloatTensor, noise: torch.FloatTensor,

diffusion/sa_solver_diffusers.py:832

Methodadd_noise

Compute the noised input xt = alpha_t * x + sigma_t * noise. Args: x: A `torch.Tensor` with shape `(batch_size, *shape)`

diffusion/model/dpm_solver.py:1033

Methodadd_params

Add all parameters of module to the params list. The parameters of the given module will be added to the list of param groups, with s

diffusion/utils/optimizer.py:34

Functionall_gather

Run all_gather on arbitrary picklable data (not necessarily tensors) Args: data: any picklable object Returns: list[data]

diffusion/utils/dist_utils.py:85

← previous next →301–400 of 537, ranked by callers