Add inference program for Transformer. #727
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
guoshengCS
force-pushed
the
add-transformer-infer
branch
from
38a8d43
to
60dba9f
Compare
guoshengCS
force-pushed
the
add-transformer-infer
branch
from
60dba9f
to
ff80721
Compare
lcy-seso
reviewed
Mar 20, 2018
| @@ -15,6 +15,23 @@ class TrainTaskConfig(object): | |||
| # the params for learning rate scheduling | |||
| warmup_steps = 4000 | |||
|
|
|||
| # the directory for saving inference models | |||
There was a problem hiding this comment.
for saving inference models --> for saving trained models
|
|
||
| class InferTaskConfig(object): | ||
| use_gpu = False | ||
| # number of sequences contained in a mini-batch |
There was a problem hiding this comment.
- number of sequences contained in a mini-batch --> the number of examples in one run for sequence generation.
- Please add a comment here to warn users currently the batch size can only be set to 1.
|
|
||
| class InferTaskConfig(object): | ||
| use_gpu = False | ||
| # number of sequences contained in a mini-batch |
There was a problem hiding this comment.
number of sequences contained in a mini-batch --> the number of examples in one run for sequence generation.
| # the params for beam search | ||
| beam_size = 5 | ||
| max_length = 30 | ||
| n_best = 1 |
There was a problem hiding this comment.
please comment n_best. It is confusing to me about what is the difference between beam_size and n_best.
| @@ -15,6 +15,23 @@ class TrainTaskConfig(object): | |||
| # the params for learning rate scheduling | |||
| warmup_steps = 4000 | |||
|
|
|||
| # the directory for saving inference models | |||
| model_dir = "transformer_model" | |||
There was a problem hiding this comment.
change the name to "trained_models"
| max_length, | ||
| slf_attn_bias_flag, | ||
| src_attn_bias_flag, | ||
| pos_flag=1): |
There was a problem hiding this comment.
change "pos_flag" into "is_pos=True"
| dtype="float32", | ||
| append_batch_size=False) | ||
| enc_input_layers = make_inputs(encoder_input_data_names, n_head, d_model, | ||
| batch_size, max_length, 1, 0) |
|
|
||
| dec_input_layers = make_inputs(decoder_input_data_names, n_head, d_model, | ||
| batch_size, max_length, 1, 1) | ||
|
|
| # Padding index do not contribute to the total loss. The weights is used to | ||
| # cancel padding index in calculating the loss. | ||
| gold, weights = make_inputs(label_data_names, n_head, d_model, batch_size, | ||
| max_length, 0, 0, 0) |
There was a problem hiding this comment.
make the last three parameters of make_inputs boolean parameters.
| name=input_data_names[2] | ||
| if slf_attn_bias_flag == 1 else input_data_names[-1], | ||
| shape=[batch_size, n_head, max_length, max_length] | ||
| if slf_attn_bias_flag == 1 else [batch_size, max_length, d_model], |
There was a problem hiding this comment.
make src_attn_bias_flag a boolean parameter.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Add inference program for Transformer.