Merge pull request #32 from wellcometrust/new-release

Preparing for new release

Makefile

@@ -1,5 +1,14 @@
 .DEFAULT_GOAL := all
 
+# Determine OS (from https://gist.github.com/sighingnow/deee806603ec9274fd47)
+UNAME_S := $(shell uname -s)
+ifeq ($(UNAME_S),Linux)
+	OSFLAG := linux
+endif
+ifeq ($(UNAME_S),Darwin)
+	OSFLAG := macosx
+endif
+
 #
 # Set file and version for embeddings and model, plus local paths
 #
@@ -84,9 +93,9 @@ datasets = data/splitting/2019.12.0_splitting_train.tsv \
 		   data/parsing/2020.3.2_parsing_train.tsv \
            data/parsing/2020.3.2_parsing_test.tsv \
            data/parsing/2020.3.2_parsing_valid.tsv \
-           data/multitask/2020.3.19_multitask_train.tsv \
-           data/multitask/2020.3.19_multitask_test.tsv \
-           data/multitask/2020.3.19_multitask_valid.tsv
+           data/multitask/2020.3.18_multitask_train.tsv \
+           data/multitask/2020.3.18_multitask_test.tsv \
+           data/multitask/2020.3.18_multitask_valid.tsv
 
 
 rodrigues_datasets = data/rodrigues/clean_train.txt \
@@ -121,9 +130,10 @@ sync_model_to_s3:
 # artefacts otherwise they can make a mess of your build! Public access to
 # the wheel is granted with the --acl public-read flag.
 
+
 .PHONY: dist
 dist:
-	-rm build/bin build/bdist.linux-x86_64 -r
+	-rm build/lib build/bin build/bdist.$(OSFLAG)* -r
 	-rm deep_reference_parser-20* -r
 	-rm deep_reference_parser.egg-info -r
 	-rm dist/*

deep_reference_parser/__version__.py

@@ -1,10 +1,10 @@
 __name__ = "deep_reference_parser"
-__version__ = "2020.3.3"
+__version__ = "2020.4.5"
 __description__ = "Deep learning model for finding and parsing references"
 __url__ = "https://github.com/wellcometrust/deep_reference_parser"
 __author__ = "Wellcome Trust DataLabs Team"
 __author_email__ = "[email protected]"
 __license__ = "MIT"
 __splitter_model_version__ = "2020.3.6_splitting"
 __parser_model_version__ = "2020.3.8_parsing"
-__splitparser_model_version__ = "2020.3.19_multitask"
+__splitparser_model_version__ = "2020.4.5_multitask"

deep_reference_parser/configs/2020.4.5_multitask.ini

@@ -0,0 +1,36 @@
+[DEFAULT]
+version = 2020.4.5_multitask
+description = Uses 2020.3.18 data
+deep_reference_parser_version = 9432b6e
+
+[data]
+# Note that test and valid proportion are only used for data creation steps,
+# not when running the train command.
+test_proportion = 0.25
+valid_proportion = 0.25
+data_path = data/
+respect_line_endings = 0
+respect_doc_endings = 1
+line_limit = 150
+policy_train = data/processed/annotated/deep_reference_parser/multitask/2020.3.18_multitask_train.tsv
+policy_test = data/processed/annotated/deep_reference_parser/multitask/2020.3.18_multitask_test.tsv
+policy_valid = data/processed/annotated/deep_reference_parser/multitask/2020.3.18_multitask_valid.tsv
+s3_slug = https://datalabs-public.s3.eu-west-2.amazonaws.com/deep_reference_parser/
+
+[build]
+output_path = data/models/multitask/2020.4.5_multitask/
+output = crf
+word_embeddings = embeddings/2020.1.1-wellcome-embeddings-300.txt
+pretrained_embedding = 0
+dropout = 0.5
+lstm_hidden = 400
+word_embedding_size = 300
+char_embedding_size = 100
+char_embedding_type = BILSTM
+optimizer = adam
+
+[train]
+epochs = 60
+batch_size = 100
+early_stopping_patience = 5
+metric = val_f1