Update the model parameters based on current gradients. Optionally, will employ gradient modification or update learning rate.

Update the model parameters based on current gradients. Optionally, will employ gradient modification or update learning rate.

Update the model parameters based on current gradients. Optionally, will employ gradient modification or update learning rate.

Update the model parameters based on current gradients. Optionally, will employ gradient modification or update learning rate.

Update the model parameters based on current gradients. Optionally, will employ gradient modification or update learning rate.

The main training loops. by iterating over training data (i.e. `train_iter_fct`) and running validation (i.e. iterating over `valid_iter_

The main training loops. by iterating over training data (i.e. `train_iter_fct`) and running validation (i.e. iterating over `valid_iter_

The main training loops. by iterating over training data (i.e. `train_iter_fct`) and running validation (i.e. iterating over `valid_iter_

The main training loops. by iterating over training data (i.e. `train_iter_fct`) and running validation (i.e. iterating over `valid_iter_

The main training loops. by iterating over training data (i.e. `train_iter_fct`) and running validation (i.e. iterating over `valid_iter_

Compute the loss. Subclass must define this method. Args: batch: the current batch. output: the predict out

Compute the loss. Subclass must define this method. Args: batch: the current batch. output: the predict out

Compute the loss. Subclass must define this method. Args: batch: the current batch. output: the predict out

Compute the loss. Subclass must define this method. Args: batch: the current batch. output: the predict out

Compute the loss. Subclass must define this method. Args: batch: the current batch. output: the predict out

Make shard state dictionary for shards() to return iterable shards for efficient loss computation. Subclass must define this

Make shard state dictionary for shards() to return iterable shards for efficient loss computation. Subclass must define this

Make shard state dictionary for shards() to return iterable shards for efficient loss computation. Subclass must define this

Make shard state dictionary for shards() to return iterable shards for efficient loss computation. Subclass must define this

Make shard state dictionary for shards() to return iterable shards for efficient loss computation. Subclass must define this

Gathers arbitrary data from all nodes into a list.

Gathers arbitrary data from all nodes into a list.

Gathers arbitrary data from all nodes into a list.

Gathers arbitrary data from all nodes into a list.

Gathers arbitrary data from all nodes into a list.

Gather a `Statistics` object accross multiple process/nodes Args: stat(:obj:Statistics): the statistics object to gather

Gather a `Statistics` object accross multiple process/nodes Args: stat(:obj:Statistics): the statistics object to gather

Gather a `Statistics` object accross multiple process/nodes Args: stat(:obj:Statistics): the statistics object to gather

Gather a `Statistics` object accross multiple process/nodes Args: stat(:obj:Statistics): the statistics object to gather

Gather a `Statistics` object accross multiple process/nodes Args: stat(:obj:Statistics): the statistics object to gather

All-reduce and rescale tensors in chunks of the specified size. Args: tensors: list of Tensors to all-reduce rescale_denom: denom

All-reduce and rescale tensors in chunks of the specified size. Args: tensors: list of Tensors to all-reduce rescale_denom: denom

All-reduce and rescale tensors in chunks of the specified size. Args: tensors: list of Tensors to all-reduce rescale_denom: denom

All-reduce and rescale tensors in chunks of the specified size. Args: tensors: list of Tensors to all-reduce rescale_denom: denom

All-reduce and rescale tensors in chunks of the specified size. Args: tensors: list of Tensors to all-reduce rescale_denom: denom

Args: model_opt: the option loaded from checkpoint. fields: `Field` objects for the model. gpu(bool): whether to use gpu. c

Args: model_opt: the option loaded from checkpoint. fields: `Field` objects for the model. gpu(bool): whether to use gpu. c

Args: model_opt: the option loaded from checkpoint. fields: `Field` objects for the model. gpu(bool): whether to use gpu. c

Args: model_opt: the option loaded from checkpoint. fields: `Field` objects for the model. gpu(bool): whether to use gpu. c

Args: model_opt: the option loaded from checkpoint. fields: `Field` objects for the model. gpu(bool): whether to use gpu. c

This returns user-defined train/validate data iterator for the trainer to iterate over. We implement simple ordered iterator strategy here, but

This returns user-defined train/validate data iterator for the trainer to iterate over. We implement simple ordered iterator strategy here, but

This returns user-defined train/validate data iterator for the trainer to iterate over. We implement simple ordered iterator strategy here, but

This returns user-defined train/validate data iterator for the trainer to iterate over. We implement simple ordered iterator strategy here, but

This returns user-defined train/validate data iterator for the trainer to iterate over. We implement simple ordered iterator strategy here, but

Build an Embeddings instance. Args: opt: the option in current environment. word_dict(Vocab): words dictionary. feature_dicts([

Returns a LossCompute subclass which wraps around an nn.Module subclass (such as nn.NLLLoss) which defines the loss criterion. The LossCompute

Returns a LossCompute subclass which wraps around an nn.Module subclass (such as nn.NLLLoss) which defines the loss criterion. The LossCompute

Returns a LossCompute subclass which wraps around an nn.Module subclass (such as nn.NLLLoss) which defines the loss criterion. The LossCompute

Returns a LossCompute subclass which wraps around an nn.Module subclass (such as nn.NLLLoss) which defines the loss criterion. The LossCompute

Returns a LossCompute subclass which wraps around an nn.Module subclass (such as nn.NLLLoss) which defines the loss criterion. The LossCompute

Building and saving the dataset

Building and saving the dataset

Building and saving the dataset

Building and saving the dataset

Building and saving the dataset

compute elapsed time

compute elapsed time

compute elapsed time

compute elapsed time

compute elapsed time