(self, b: int, l: int)

( video: torch.Tensor, # B C T H W shape, [-1,1] values cotracker: torch.nn.Module, device: torch

( self, cond: Float[torch.Tensor, "... d_cond"], t_embs: list[Float[torch.Tensor, "...

( sample_dir: str, dataset: BilliardSimDataset, sim: Float[torch.Tensor, "t n c"], batch: dict

(q, k, scale, eps)

( num_frames, sequence_length, frame_skip=0, )

( frame_size: int, min_border_offset: int, max_border_offset: int, nr_balls: int, ball_rad

(input_path: Path, output_path: Path, max_size: int, prefix: str = "shard")

(model: MyriadStepByStep, batch: dict[str, torch.Tensor], num_traj: int, ensemble_size: int)

(c)

(x: torch.Tensor)

(x: torch.Tensor)

( num_test_rollouts: int, batch: dict[str, torch.Tensor], best_action: torch.Tensor, n_sim_ste

( input_filename: Path, output_dirname: Path, batch_size: int = 1, nr_workers: int = 1, sh

(e, data)

(e, data)

( input_filename: Path, output_filename: Path, device: str = "cuda:0", init_cotracker_kwargs:

(self)

( frames, input_args: list = ["pipe:"], input_kwargs: dict = { "format": "rawvideo",

(e, data, touch, distX, distY, fn)

(event, touch, timeStamp)

(a,b,c)

(self)

( video: np.ndarray, path:str, fps: int = 24, vcodec: str = "libx264", pix_fmt: str = "yuv

(c)

(a)

(a,b)

()

(slider)

(slider)

()

( event, params )

()

()

(a)

(transitioner, slider)

(a)

(slider)

()

(slider)

(a)

(slider)

(a)

(slider)

()

(c)

(slider)

(fn)

(slider)

(transitioner, slider)

()

(c)

(self, name: str)

(self, idx)

(self, idx)

(self, idx)

(self, idx)

(self, idx)

(self, idx)

(self, idx)

(self, idx)

(self, inner_dataloader: torch.utils.data.DataLoader, collation_factor: int, collate_fn)

( self, batch_size: int, train=None, validation=None, num_workers=4,

( self, tar_base: str | list | ListConfig, batch_size: int, train=None,

( self, image_size: int = 448, model_version: str = "dinov2_vitb14_reg", requi

(self, module: nn.Module)

(self, shape: tuple[int, ...], dtype: torch.dtype, device: torch.device)

(self, in_features, out_features, std=1.0)

( self, d_head: int, n_heads: int, theta_t: float = 100.0, )

(self, shape, eps=1e-6)

(self, features: int, eps: float=1e-6)

( self, out_features: int, input_scale: float = 0.5, aux_feature_dim: int = 0,

( self, width: int, depth: int = 2, )

( self, d_model: int, d_head: int, dropout: float = 0.0, ff_expand: in

( self, width: int, depth: int, aux_feat_dim: int, d_head: int,

( self, d_model: int, d_cond_norm: int, expansion_factor: float=1., re

( self, depth: int, d_model: int, d_cond: int, d_out: int, exp

( self, rf_head: RFHead, cond: Float[torch.Tensor, "... d_cond"], steps: int =

( self, depth: int, denoiser_width: int, d_cond: int, expansion_factor

(self, width: int, flow_dim: int)

(self, width: int, flow_dim: int, n_components: int = 4, dist_dtype: torch.dtype = torch.float64)

(self, x: torch.Tensor)

( self, width: int, flow_dim: int, depth: int = 2, expansion_factor: i

( self, # general parameters width: int, # transformer parameters dept

(self, img_path: str, background_queries: list, object_queries: list, object_pokes: list)

(self, in_features, hidden_features=None, out_features=None, bias=True, **kwargs)

(self, img_size=224, patch_size=16, in_chans=3, embed_dim=768)

(self, dim, num_heads=8, qkv_bias=False)

( self, dim, init_values=1e-5, inplace=False, )

( self, dim, num_heads, mlp_ratio=4.0, qkv_bias=False, ffn_bia

( self, img_size=224, patch_size=16, in_chans=3, embed_dim=768,

( self, image_size: int = 448, model_version: str = "dinov2_vitb14_reg", requi

( self, tar_base: str, batch_size: int, num_workers: int = 8, shuffle:

(self, features: int, cond_features: int, eps: float = 1e-6)

(self, d_head: int, n_heads: int)

( self, d_model: int, d_cross: int, d_head: int = 64, dropout: float =

(self, width: int, depth: int, d_cross: int, layer_params: dict[str, Any] = {})

(self, width: int, d_aux: int, flow_dim: int = 2, depth: int = 2, flow_prescale: float = 0.5)

(self, width: int, d_out: int, depth: int = 2)