Load the rank result from TREC format: <query_id> Q0 <doc_id> <rank> <score> <run_id> or <query_id> <doc_id> <score>

Normalize a final answer to a quantitative reasoning question. Args: final_answer: The answer string to normalize Returns: N

Normalize a final answer to a quantitative reasoning question. Args: final_answer: The answer string to normalize Returns: N

cfg/argparse to kwargs helper Convert optimizer args in argparse args or cfg like object to keyword args for updated create fn.

Pad a DataProto to size divisible by size_divisor Args: data (DataProto): the unpadded DataProto size_divisor (int): size divisor

Any module not under timm.models.layers should get this decorator if we don't want to trace through it.

Compute relative logits along one dimension As per: https://gist.github.com/aravindsrinivas/56359b79f0ce4449bcb04ab4b56a57a2 Originally from

Compute relative logits along one dimension As per: https://gist.github.com/aravindsrinivas/56359b79f0ce4449bcb04ab4b56a57a2 Originally from

Remove the obsolete checkpoints that exceed the save limit.

Note that this operation is in-place

Rescale the grid of position embeddings when loading from state_dict Expected shape of position embeddings is (1, T, N, C), and considers onl

Constructs a ResNet-26-D model.

Constructs a ResNet-50-D model.

Rotates half the hidden dims of the input.

Rotates half the hidden dims of the input.

cfg/argparse to kwargs helper Convert scheduler args in argparse args or cfg (.dot) like object to keyword args.

Set ulysses sequence parallel process group.

Recursively move tensors in a nested list, tuple, or dictionary to the specified device.

Method updates the image resolution to be processed and window size and so the pair-wise relative positions. Args: new_window_siz

Partition into non-overlapping windows with padding if needed. Args: x (tensor): input tokens with [B, H, W, C]. window_size

Args: x: (B, H, W, C) window_size (int): window size Returns: windows: (num_windows*B, window_size, window_size, C)

Args: x: (B, H, W, C) window_size (int): window size Returns: windows: (num_windows*B, window_size, window_size, C)

Args: in_features: The number of input features. num_classes: The number of classes for the final classifier layer (

Selective Kernel Convolution Module As described in Selective Kernel Networks (https://arxiv.org/abs/1903.06586) with some modifications.

Args: block_args: Arguments for blocks of the network. num_classes: Number of classes for classification head.