Create a TIMM-based featurizer model with appropriate configurations. Args: model_id: The TIMM model ID to load

Download a file from the OpenPI S3 bucket using boto3. This is a more performant version of download but can only handle s3 urls. In openpi repo,

Since we are sharing cache directory with containerized runtime as well as training script, we need to ensure that the cache directory has the cor

Since we are sharing cache directory with containerized runtime as well as training script, we need to ensure that the cache directory has the cor

Applies gated residual connection with optional gate parameter. Args: x: Input tensor (residual) y: Output tensor to be

Generator of examples for each split.

Return an activation function given a string

Get bins per dimension based on codebook size, from the original FSQ paper.

Get the mtime of a given date at midnight UTC.

Merges the loaded parameters with the reference parameters. Args: loaded_params: The parameters to merge. params: The reference p

Merges the loaded parameters with the reference parameters. Args: loaded_params: The parameters to merge. params: The reference p

Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape `(batch_size, key_value_l

Preprocess an image or batch of images. Copy of the `preprocess` method from `CLIPImageProcessor`. Args: images (`ImageI

Helper method to preprocess observation.

Process proprioceptive features and append to vision features

Process vision features with optional FiLM conditioning

Replace embeddings in input_embeddings at positions where all_actions_mask is True with embeddings from noisy_action_features, using

Define filepaths for data splits.

Define filepaths for data splits.

Define filepaths for data splits.

Define filepaths for data splits.

Define filepaths for data splits.

Tokenize a list of strings.

transform target pose from world frame to base frame base_pose: np.array([x, y, z, qw, qx, qy, qz]) t

Creates wrapper around an individual vision transformer to allow for infusion of language inputs. Args: vit (VisionTrans

lang_tokens: (batch_size, lang_len, lang_token_dim) img_tokens: (batch_size, img_len, img_token_dim) state_tokens: (batch_siz

Allocates an integer number of threads across datasets based on weights. The final array sums to `n`, but each element is no less than 1. If

Applies common transforms that happen at a frame level. These transforms are usually more CPU-intensive, (e.g., decoding or resizing images).

Applies Rotary Position Embedding to the query and key tensors. Args: q (`torch.Tensor`): The query tensor. k (`torch.Tensor`): T

Applies common transforms that happen at a trajectory level. Such transforms are usually some sort of "relabeling" (e.g., filtering, chunking

Check and sync model logic files between current code and checkpoint. Handles the relationship between current and checkpoint versions of bo

Given state and delta action sequence generate equivalent goal position and orientation for each step keep the original gripp

Setup the optimizer. We provide a reasonable default that works well. If you want to use something else, you can pass a tuple in the

u: (..., 3) v: (..., 3) u x v: (..., 3)

Get the data config for this data loader.

Embed images with SigLIP and language tokens with embedding layer to prepare for PaliGemma transformer processing.