Generate maps

_elmo.map.dvc

-md5: 41a1504193d7a3c24342bd32f6cabc34
+md5: 3fd7ed4807e41c8215e0bade52ab2fda
 cmd: CUDA_VISIBLE_DEVICES=0 ./semrel/data/scripts/cli/make_elmo_map.py --input-path
   ./semrel/data/data/relations.files.list --model ./semrel/data/data/elmo/options.json
   ./semrel/data/data/elmo/weights.hdf5 --output-paths ./semrel/data/data/maps/elmo.map.keys
   ./semrel/data/data/maps/elmo.map.pt
 deps:
-- md5: 157a217de2adc49fcc72e3ed8b5d86bb
+- md5: e34115c7c699baaf15c97ca5364ad554
   path: semrel/data/data/relations.files.list
 - md5: 1111682d038ff9be1f622b0ec771e7d0
   path: semrel/data/scripts/cli/make_elmo_map.py
 - md5: 14fa344f624b3105e02e46e2567cb3ae
   path: semrel/data/scripts/maps.py
 outs:
-- md5: d8f7a05782631e72e328a4ec303ea770
+- md5: cd32cc2bc03aaf30e93665bae80e7d72
   path: semrel/data/data/maps/elmo.map.keys
   cache: true
   metric: false
   persist: false
-- md5: d29aea6be2de86643f75ff0e8b1d4a42
+- md5: 9987878c3752d38c5ecf287b95a75700
   path: semrel/data/data/maps/elmo.map.pt
   cache: true
   metric: false

_fasttext.map.dvc

-md5: d6ac7b87c9e7c87e41ae45620c06cb40
+md5: 218b1553a127004a3693c93f1cde6c13
 cmd: CUDA_VISIBLE_DEVICES=0 ./semrel/data/scripts/cli/make_fasttext_map.py --input-path
   ./semrel/data/data/relations.files.list --model ./semrel/data/data/fasttext/kgr10.plain.skipgram.dim300.neg10.bin
   --output-paths ./semrel/data/data/maps/fasttext.map.keys ./semrel/data/data/maps/fasttext.map.pt
 deps:
-- md5: 157a217de2adc49fcc72e3ed8b5d86bb
+- md5: e34115c7c699baaf15c97ca5364ad554
   path: semrel/data/data/relations.files.list
 - md5: 9d988a887cba28b9774b3171f9c511f0
   path: semrel/data/scripts/cli/make_fasttext_map.py
 - md5: 14fa344f624b3105e02e46e2567cb3ae
   path: semrel/data/scripts/maps.py
 outs:
-- md5: d8f7a05782631e72e328a4ec303ea770
+- md5: cd32cc2bc03aaf30e93665bae80e7d72
   path: semrel/data/data/maps/fasttext.map.keys
   cache: true
   metric: false
   persist: false
-- md5: ddaccfb76b6176bf54f66eadb4104125
+- md5: 8335f6b5fdc048f306ff4a3b659960cc
   path: semrel/data/data/maps/fasttext.map.pt
   cache: true
   metric: false

_relations.files.list.dvc

-md5: 4c32e1b2010e8f694eca90a6e1ca0c28
-cmd: find ./semrel/data/data/corpora/000 -type f -name *.rel.xml > ./semrel/data/data/relations.files.list
+md5: 4dc591a5158e89ab53a1c29a06561cbc
+cmd: find ./semrel/data/data/corpora -type f -name *.rel.xml > ./semrel/data/data/relations.files.list
 deps:
-- md5: d5067d94218270bfb54237e0109e24ab.dir
-  path: semrel/data/data/corpora/000
+- md5: d171bbc05d28098ebd033f5241cf6929.dir
+  path: semrel/data/data/corpora
 outs:
-- md5: 157a217de2adc49fcc72e3ed8b5d86bb
+- md5: e34115c7c699baaf15c97ca5364ad554
   path: semrel/data/data/relations.files.list
   cache: true
   metric: false

_retrofit.map.dvc

-md5: 4aa5d14bd1e3ced4fa6dd7e358f6d8b9
+md5: 85c576f8927d68435e5157edb6984ab8
 cmd: CUDA_VISIBLE_DEVICES=0 ./semrel/data/scripts/cli/make_retrofit_map.py --input-path
   ./semrel/data/data/relations.files.list --model-retrofit ./semrel/data/data/fasttext/kgr10.plain.skipgram.dim300.neg10.retrofit.vec
   --model-fasttext ./semrel/data/data/fasttext/kgr10.plain.skipgram.dim300.neg10.bin
   --output-paths ./semrel/data/data/maps/retrofit.map.keys ./semrel/data/data/maps/retrofit.map.pt
 deps:
-- md5: 157a217de2adc49fcc72e3ed8b5d86bb
+- md5: e34115c7c699baaf15c97ca5364ad554
   path: semrel/data/data/relations.files.list
 - md5: c42c5efba4f6aec66a615f974013d3cb
   path: semrel/data/scripts/cli/make_retrofit_map.py
 - md5: 14fa344f624b3105e02e46e2567cb3ae
   path: semrel/data/scripts/maps.py
 outs:
-- md5: d8f7a05782631e72e328a4ec303ea770
+- md5: cd32cc2bc03aaf30e93665bae80e7d72
   path: semrel/data/data/maps/retrofit.map.keys
   cache: true
   metric: false
   persist: false
-- md5: 333ac2d2ede42d07329f990502b27b4d
+- md5: c5c1c7acb41606f7d2abb3fc263697a3
   path: semrel/data/data/maps/retrofit.map.pt
   cache: true
   metric: false

semrel/data/0_list_corpora_files.sh

@@ -2,7 +2,7 @@
 
 pushd "$(git rev-parse --show-toplevel)"
 
-CORPORA_DIR="./semrel/data/data/corpora/000"
+CORPORA_DIR="./semrel/data/data/corpora"
 OUTPUT_FILE="./semrel/data/data/relations.files.list"
 
 dvc run \