Add files using upload-large-folder tool

9832690 verified 4 months ago

2.68 kB

# Adapted from https://github.com/mlcommons/training_results_v1.1/blob/main/NVIDIA/benchmarks/bert/implementations/pytorch/padding.py

import torch
import torch.nn.functional as F
from einops import rearrange


def unpad_input(hidden_states, attention_mask, unused_mask=None):
    """
    Arguments:
        hidden_states: (batch, seqlen, ...)
        attention_mask: (batch, seqlen), bool / int, 1 means valid and 0 means not valid.
        unused_mask: (batch, seqlen), bool / int, 1 means the element is allocated but unused.
    Return:
        hidden_states: (total_nnz, ...), where total_nnz = number of tokens selected in attention_mask + unused_mask.
        indices: (total_nnz), the indices of masked tokens from the flattened input sequence.
        cu_seqlens: (batch + 1), the cumulative sequence lengths, used to index into hidden_states.
        max_seqlen_in_batch: int
        seqused: (batch), returns the number of tokens selected in attention_mask + unused_mask.
    """
    all_masks = (attention_mask + unused_mask) if unused_mask is not None else attention_mask
    seqlens_in_batch = all_masks.sum(dim=-1, dtype=torch.int32)
    used_seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32)
    indices = torch.nonzero(all_masks.flatten(), as_tuple=False).flatten()
    max_seqlen_in_batch = seqlens_in_batch.max().item()
    cu_seqlens = F.pad(torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.int32), (1, 0))
    # TD [2022-03-04] We don't want to index with a bool mask, because Pytorch will expand the
    # bool mask, then call nonzero to get the indices, then index with those. The indices is @dim
    # times larger than it needs to be, wasting memory. It's faster and more memory-efficient to
    # index with integer indices.
    return (
        rearrange(hidden_states, "b s ... -> (b s) ...")[indices],
        indices,
        cu_seqlens,
        max_seqlen_in_batch,
        used_seqlens_in_batch,
    )


def pad_input(hidden_states, indices, batch, seqlen):
    """
    Arguments:
        hidden_states: (total_nnz, ...), where total_nnz = number of tokens in selected in attention_mask.
        indices: (total_nnz), the indices that represent the non-masked tokens of the original padded input sequence.
        batch: int, batch size for the padded sequence.
        seqlen: int, maximum sequence length for the padded sequence.
    Return:
        hidden_states: (batch, seqlen, ...)
    """
    dim = hidden_states.shape[1:]
    output = torch.zeros((batch * seqlen), *dim, device=hidden_states.device, dtype=hidden_states.dtype)
    output[indices] = hidden_states
    return rearrange(output, "(b s) ... -> b s ...", b=batch)

	# Adapted from https://github.com/mlcommons/training_results_v1.1/blob/main/NVIDIA/benchmarks/bert/implementations/pytorch/padding.py

	import torch
	import torch.nn.functional as F
	from einops import rearrange


	def unpad_input(hidden_states, attention_mask, unused_mask=None):
	"""
	Arguments:
	hidden_states: (batch, seqlen, ...)
	attention_mask: (batch, seqlen), bool / int, 1 means valid and 0 means not valid.
	unused_mask: (batch, seqlen), bool / int, 1 means the element is allocated but unused.
	Return:
	hidden_states: (total_nnz, ...), where total_nnz = number of tokens selected in attention_mask + unused_mask.
	indices: (total_nnz), the indices of masked tokens from the flattened input sequence.
	cu_seqlens: (batch + 1), the cumulative sequence lengths, used to index into hidden_states.
	max_seqlen_in_batch: int
	seqused: (batch), returns the number of tokens selected in attention_mask + unused_mask.
	"""
	all_masks = (attention_mask + unused_mask) if unused_mask is not None else attention_mask
	seqlens_in_batch = all_masks.sum(dim=-1, dtype=torch.int32)
	used_seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32)
	indices = torch.nonzero(all_masks.flatten(), as_tuple=False).flatten()
	max_seqlen_in_batch = seqlens_in_batch.max().item()
	cu_seqlens = F.pad(torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.int32), (1, 0))
	# TD [2022-03-04] We don't want to index with a bool mask, because Pytorch will expand the
	# bool mask, then call nonzero to get the indices, then index with those. The indices is @dim
	# times larger than it needs to be, wasting memory. It's faster and more memory-efficient to
	# index with integer indices.
	return (
	rearrange(hidden_states, "b s ... -> (b s) ...")[indices],
	indices,
	cu_seqlens,
	max_seqlen_in_batch,
	used_seqlens_in_batch,
	)


	def pad_input(hidden_states, indices, batch, seqlen):
	"""
	Arguments:
	hidden_states: (total_nnz, ...), where total_nnz = number of tokens in selected in attention_mask.
	indices: (total_nnz), the indices that represent the non-masked tokens of the original padded input sequence.
	batch: int, batch size for the padded sequence.
	seqlen: int, maximum sequence length for the padded sequence.
	Return:
	hidden_states: (batch, seqlen, ...)
	"""
	dim = hidden_states.shape[1:]
	output = torch.zeros((batch * seqlen), *dim, device=hidden_states.device, dtype=hidden_states.dtype)
	output[indices] = hidden_states
	return rearrange(output, "(b s) ... -> b s ...", b=batch)