MCPcopy Create free account

hub / github.com/andrewsonga/generative_view_stitching / functions

Functions993 in github.com/andrewsonga/generative_view_stitching

Method__init__
(self, in_channels: int, out_channels: int)
algorithms/dfot/backbones/u_vit/u_vit_blocks.py:310
Method__init__
(self, in_channels: int, out_channels: int)
algorithms/dfot/backbones/u_vit/u_vit_blocks.py:327
Method__init__
( self, cfg: DictConfig, x_shape: torch.Size, max_tokens: int, externa
algorithms/dfot/backbones/u_vit/u_vit3d.py:30
Method__init__
( self, cfg: DictConfig, x_shape: torch.Size, max_tokens: int, externa
algorithms/dfot/backbones/dit/dit3d_pose.py:15
Method__init__
(self, embed_dim: int, shape: Tuple[int, ...], learnable: bool = False)
algorithms/dfot/backbones/dit/dit_base.py:328
Method__init__
( self, cfg: DictConfig, x_shape: torch.Size, max_tokens: int, externa
algorithms/dfot/backbones/dit/dit3d.py:13
Method__init__
( self, dim: int, num_heads: int = 8, qkv_bias: bool = False, qk_norm:
algorithms/dfot/backbones/dit/dit_blocks.py:24
Method__init__
(self, hidden_size: int)
algorithms/dfot/backbones/dit/dit_blocks.py:91
Method__init__
(self, hidden_size: int)
algorithms/dfot/backbones/dit/dit_blocks.py:121
Method__init__
( self, hidden_size: int, out_channels: int, )
algorithms/dfot/backbones/dit/dit_blocks.py:217
Method__init__
( self, cfg: DictConfig, x_shape: torch.Size, max_tokens: int, externa
algorithms/dfot/backbones/u_net/u_net3d.py:26
Method__init__
( self, dim: int, dim_out: int, emb_dim: Optional[int] = None, dropout
algorithms/dfot/backbones/u_net/u_net_blocks.py:15
Method__init__
(self, dim: int)
algorithms/dfot/backbones/u_net/u_net_blocks.py:69
Method__init__
( self, query_dim: int, heads: int = 4, dim_head: int = 32, bias: bool
algorithms/dfot/backbones/u_net/u_net_blocks.py:90
Method__init__
( self, dim: int, heads: int = 4, dim_head: int = 32, use_linear: bool
algorithms/dfot/backbones/u_net/u_net_blocks.py:184
Method__init__
( self, query_dim: int, heads: int = 4, dim_head: int = 32, rotary_emb
algorithms/dfot/backbones/u_net/u_net_blocks.py:214
Method__init__
( self, dim: int, heads: int = 4, dim_head: int = 32, is_causal: bool
algorithms/dfot/backbones/u_net/u_net_blocks.py:261
Method__init__
(self, from_shape: str, to_shape: str, module: nn.Module)
algorithms/dfot/backbones/u_net/u_net_blocks.py:298
Method__init__
(self, *args, **kwargs)
algorithms/dfot/backbones/u_net/u_net_blocks.py:314
Method__len__
(self)
datasets/video/base_video.py:275
Method__new__
(cls, value, dim_weight, min_val, max_val)
algorithms/common/metrics/video/vbench/vbench.py:31
Method__post_init__
(self)
utils/retrieval_utils.py:277
Method_basic_init
(module: nn.Module)
algorithms/dfot/backbones/dit/dit_blocks.py:186
Method_build_buffer
(self)
algorithms/dfot/diffusion/continuous_diffusion.py:110
Method_build_dataset
(self, split: str)
experiments/data_modules/resumable_data_module.py:35
Method_build_external_cond_embedding
(self)
algorithms/dfot/backbones/u_vit/u_vit3d_pose.py:35
Method_build_external_cond_embedding
(self)
algorithms/dfot/backbones/dit/dit3d_pose.py:46
Method_build_model
(self)
algorithms/vae/image_vae/preprocessor.py:27
Method_build_model
(self)
algorithms/vae/video_vae/trainer.py:35
Method_build_model
(self)
algorithms/dfot/dfot_video.py:98
Method_check_input
(fake: Tensor, real: Tensor, skip_message: str = "FVD")
algorithms/common/metrics/video/fvd.py:25
Method_check_input
(self, fake: Tensor, real: Tensor)
algorithms/common/metrics/video/fvmd.py:36
Method_compute_fov_overlap
Compute the fraction of sampled points that is visible in the field-of-view (FOV) of both chunks, each of which are subsequences of chunk_siz
algorithms/dfot/dfot_video_pose.py:113
Method_compute_histogram
Args: vectors (torch.Tensor): (B, S, N, 2) - velocities or accelerations
algorithms/common/metrics/video/models/motion_extractor.py:107
Function_data_module_cls
(_, root_cfg: DictConfig, compatible_datasets: Dict)
experiments/data_modules/utils.py:7
Function_dl_wrap
(tarpath: str, videopath: str, line: str)
datasets/video/kinetics_600.py:39
Function_download_youtube_video
Downloads a YouTube video to the specified directory. Retries with different clients if the download fails, to guarantee that it does not mis
datasets/video/realestate10k.py:1034
Method_generate_pyramid_scheduling_matrix_without_repeat
( horizon, uncertainty_scale, sampling_timesteps )
algorithms/dfot/dfot_video.py:1103
Method_generate_trapezoid_scheduling_matrix_without_repeat
( horizon, uncertainty_scale, sampling_timesteps )
algorithms/dfot/dfot_video.py:1132
Method_list_to_tuple
( freq_ranges: Optional[List[List[float] | ALLType]], )
algorithms/dfot/history_guidance.py:913
Method_load_ema_weights_to_state_dict
(self, checkpoint: dict)
algorithms/vae/video_vae/trainer.py:74
Method_load_ema_weights_to_state_dict
(self, checkpoint: Dict[str, Any])
algorithms/dfot/dfot_video.py:1726
Method_make_ds_structure
move videos from split_folder/ ├── clip1.avi ├── clip2.avi to the correct format as described below:
datasets/video/kinetics_600.py:130
Method_mlp_init
(module: nn.Module)
algorithms/dfot/backbones/dit/dit3d.py:81
Method_predict_sequence
Predict a sequence given context tokens at the beginning, using sliding window if necessary. Args ---- context: torch
algorithms/dfot/dfot_rag_video_pose.py:32
Method_predict_videos
( self, xs: Float[Tensor, "B T C H W"], conditions: Optional[Float[Tensor, "B T ..."]]
algorithms/dfot/gvs_video_pose.py:21
Method_predict_videos
( self, xs: Float[Tensor, "B T C H W"], conditions: Optional[Float[Tensor, "B T ..."]]
algorithms/dfot/stochsync_video_pose.py:21
Method_predict_videos
( self, xs: Float[Tensor, "B T C H W"], conditions: Optional[Float[Tensor, "B T ..."]]
algorithms/dfot/gvs_no_omniguide_video_pose.py:20
Method_predict_videos
( self, xs: Float[Tensor, "B T C H W"], conditions: Optional[Float[Tensor, "B T ..."]]
algorithms/dfot/gvs_scalable_video_pose.py:20
Method_predict_videos
Predict the videos with the given context, using sliding window rollouts if necessary. Optionally, if cfg.tasks.prediction.keyframe_d
algorithms/dfot/dfot_video.py:535
Function_preprocess_video
( video_path: Path, resolution: int, preprocessing_type: VideoPreprocessingType = "npz", )
datasets/video/kinetics_600.py:43
Function_preprocess_video
( info: Tuple[str, Path, List[float]], resolutions_to_preprocessing: Dict[int, VideoPreprocessingType]
datasets/video/realestate10k.py:1084
Method_process_conditions
Process conditions (raw camera poses) to desired format for the model Args: conditions (Tensor): raw camera poses (B, T,
algorithms/dfot/dfot_video_pose.py:69
Method_rearrange_and_add_pos_emb_if_transformer
( self, x: Tensor, emb: Tensor, i_level: int )
algorithms/dfot/backbones/u_vit/u_vit3d_pose.py:46
Method_rope_kwargs
(i_level: int)
algorithms/dfot/backbones/u_vit/u_vit3d.py:109
Method_scan_and_log_checkpoints
(self, checkpoint_callback: ModelCheckpoint)
utils/wandb_utils.py:101
Method_should_download
(self)
datasets/video/kinetics_600.py:83
Method_should_download
(self)
datasets/video/realestate10k_mini.py:30
Method_should_download
(self)
datasets/video/realestate10k.py:56
Method_training_frame_skip
(self)
datasets/video/realestate10k.py:290
Method_training_frame_skip
(self)
datasets/video/realestate10k.py:897
Method_unnormalize_x_cpu
(self, xs)
algorithms/dfot/dfot_video.py:1627
Methodadd_param_group
(self, param_group)
utils/lightning_utils.py:397
Methodadd_pos_emb
( x: torch.Tensor, _: torch.Tensor, batch_size: int )
algorithms/dfot/backbones/dit/dit_base.py:266
Functionalphas_cumprod_linear_schedule
linear schedule as proposed in https://arxiv.org/abs/2301.10972
algorithms/dfot/diffusion/noise_schedule.py:84
Methodbase_reconstruction_guidance_fn
( target: Float[Tensor, "B T C H W"], pred_x0: Float[Tensor, "
algorithms/dfot/dfot_rag_video_pose.py:528
Methodbase_reconstruction_guidance_fn
( target: Float[Tensor, "B T C H W"], pred_x0: Float[Tensor, "
algorithms/dfot/dfot_video.py:1478
Functionbeta_linear_schedule
linear schedule as proposed in https://arxiv.org/abs/2006.11239 (original DDPM paper)
algorithms/dfot/diffusion/noise_schedule.py:93
Methodbuild_metadata
(self, split: SPLIT)
datasets/video/realestate10k.py:197
Methodbuild_transform
(self)
datasets/video/minecraft.py:63
Methodcompose
( self, x: torch.Tensor )
algorithms/dfot/history_guidance.py:1022
Methodcompose
( self, x: torch.Tensor )
algorithms/dfot/history_guidance.py:1146
Methodcompute
Compute the Inception Score (IS). Returns: The computed IS.
algorithms/common/metrics/video/inception_score.py:49
Methodcompute
Compute the final metric value. Returns: Primary metric value (video collision rate) as a single tensor
algorithms/common/metrics/video/depth_collision.py:314
Methodcompute
Compute the final metric value.
algorithms/common/metrics/video/loop_consistency.py:149
Methodcompute
Compute the final metric value.
algorithms/common/metrics/video/temporal_consistency.py:102
Methodcompute
Compute the VBench score. Returns: - The computed VBench scores for each dimension. - The normalized VBench s
algorithms/common/metrics/video/vbench/vbench.py:114
Methodcompute_cum_snr
(reverse: bool = False)
algorithms/dfot/diffusion/discrete_diffusion.py:286
Methodcompute_detailed
Compute detailed metric values for analysis. Returns: Dictionary with all metric results
algorithms/common/metrics/video/depth_collision.py:327
Functioncompute_interpolant_chunks
( conditions: Float[torch.Tensor, "B N 2 chunk_size ..."], orbit_radius: float = None, start_rotat
utils/geometry_utils.py:707
Methodconfigure_optimizers
(self)
algorithms/vae/image_vae/trainer.py:262
Methodconfigure_optimizers
(self)
algorithms/vae/video_vae/trainer.py:89
Methodconfigure_optimizers
Return an optimizer. If you need to use more than one optimizer, refer to pytorch lightning documentation: https://lightning.ai/docs/
algorithms/common/base_pytorch_algo.py:92
Methodconfigure_optimizers
(self)
algorithms/dfot/dfot_video.py:163
Functioncoords_grid
(batch, ht, wd)
algorithms/common/metrics/video/models/pips2.py:410
Functioncosine_schedule
cosine schedule as proposed in https://openreview.net/forum?id=-NEXDKk8gZ
algorithms/dfot/diffusion/noise_schedule.py:36
Functioncosine_simple_diffusion_schedule
cosine schedule with different parameterization following Simple Diffusion - https://arxiv.org/abs/2301.11093 Supports "shifted cosine sc
algorithms/dfot/diffusion/noise_schedule.py:48
Methodcustom
The most flexible way to initialize history guidance.
algorithms/dfot/history_guidance.py:899
Functiondecorator
(fn)
utils/print_utils.py:61
Methoddownload_dataset
(self)
datasets/video/kinetics_600.py:92
Methoddownload_dataset
(self)
datasets/video/realestate10k_mini.py:33
Methoddownload_dataset
(self)
datasets/video/minecraft.py:20
Methoddownload_dataset
(self)
datasets/video/realestate10k.py:67
Methodestimate_noise_level
(self, x, mu=None)
algorithms/dfot/diffusion/discrete_diffusion.py:829
Methodexclude_short_videos
( self, metadata: List[Dict[str, Any]], min_frames: int )
datasets/video/realestate10k.py:987
Methodexclude_videos_with_latents
( self, metadata: List[Dict[str, Any]] )
datasets/video/realestate10k.py:256
Methodexclude_videos_without_latents
( self, metadata: List[Dict[str, Any]] )
datasets/video/realestate10k.py:300
Methodexclude_videos_without_latents
( self, metadata: List[Dict[str, Any]] )
datasets/video/realestate10k.py:907
Methodexecute_in_parallel
execute a function in parallel on the sequence and image tensors
algorithms/dfot/backbones/dit/dit_base.py:241
Methodexternal_cond_emb_dim
(self)
algorithms/dfot/backbones/base_backbone.py:58
← previousnext →601–700 of 993, ranked by callers