Alternative constructor. Args: opt (argparse.Namespace): The parsed arguments.

Alternate constructor. Subclasses should override this method.

Alternate constructor.

Alternate constructor.

Alternate constructor.

Creates an image-to-text nmtmodel with a custom opt function. Forwards a testbatch and checks output size. Args:

Sort using the size of the image: (width, height).

Init decoder state.

Initialize decoder state.

See :obj:`RNNDecoderBase.init_state()`

DecodeStrategy subclasses should override :func:`initialize()`. `initialize` should be called before all actions. used to prepare nec

Returns the current sequence length.

Returns unmodified scores.

GNMT length re-ranking score. See "Google's Neural Machine Translation System" :cite:`wu2016google`.

Pass in needed options only when modify function definition.

?

Convert a sparse index_tensor into a batch of alignment matrix, with row normalize to the sum of 1 if set normalize. Args: index

In token batching scheme, the number of sequences is limited such that the total number of src/tgt tokens (including padding) in a batch

In token batching scheme, the number of sequences is limited such that the total number of src/tgt tokens (including padding) in a batch

These options are passed to the construction of the model. Be careful with these as they will be used during translation.

Creates a nmtmodel with a custom opt function. Forwards a testbatch and checks output size. Args: opt: Namespace

Learning rate schedule described in https://arxiv.org/pdf/1706.03762.pdf.

Learning rate schedule optimized for huge batches

Pad a batch of examples to the length of the longest example. Args: minibatch (List[torch.FloatTensor]): A list of audio data,

Parse Pharaoh alignment into [[<src>, <tgt>], ...]

Filters batches that belong only to gpu_ranks of current node

Pre-procesing options

Read data from file system and yield as dicts.

Decay based on the reciprocal of the step square root.

Process a disjoint subset of the videos on each device.

run process

Translate `inputs` We keep the same format as the Lua version i.e. ``[{"id": model_id, "src": "sequence to translate"},{ ...}]``

Projection.

signal handler

?

?

?

Performs a single optimization step. Arguments: closure (callable, optional): A closure that reevaluates the model