updated data to combined hexa dependencies

clulab · Jul 4, 2024 · a300b8e · a300b8e
1 parent 5bf353b
commit a300b8e
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 4 deletions.
diff --git a/encoder/src/main/python/averaging_trainer.py b/encoder/src/main/python/averaging_trainer.py
@@ -124,9 +124,13 @@ def print_some_params(self, model: TokenClassificationModel, msg: str) -> None:
         ShortTaskDef("NER",        "conll-ner/", "train.txt",    "dev.txt",    "test.txt"),
         ShortTaskDef("POS",        "pos/",       "train.txt",    "dev.txt",    "test.txt"),
         ShortTaskDef("Chunking",   "chunking/",  "train.txt",    "test.txt",   "test.txt"),
+        #ShortTaskDef("Hexa Term", "deps-wsj/", "train.labels.hexaterms", "dev.labels.hexaterms", "test.labels.hexaterms"),
+        #ShortTaskDef("Hexa NonTerm", "deps-wsj/", "train.labels.hexanonterms", "dev.labels.hexanonterms", "test.labels.hexanonterms")
+        ShortTaskDef("Hexa Term",  "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels.hexaterms",  "test.labels.hexaterms",  "test.labels.hexaterms"), # dev is included in train
+        ShortTaskDef("Hexa NonTerm", "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels.hexanonterms", "test.labels.hexanonterms", "test.labels.hexanonterms") # dev is included in train
         #ShortTaskDef("Deps Head",  "deps-wsj/",  "train.heads",  "dev.heads",  "test.heads"),
         #ShortTaskDef("Deps Label", "deps-wsj/",  "train.labels", "dev.labels", "test.labels", dual_mode=True)
-        ShortTaskDef("Deps Head",  "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.heads",  "test.heads",  "test.heads"),
-        ShortTaskDef("Deps Label", "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels", "test.labels", "test.labels", dual_mode=True)
+        #ShortTaskDef("Deps Head",  "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.heads",  "test.heads",  "test.heads"),
+        #ShortTaskDef("Deps Label", "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels", "test.labels", "test.labels", dual_mode=True)
     ])
     AveragingTrainer(tokenizer).train(tasks)
diff --git a/encoder/src/main/python/clu_trainer.py b/encoder/src/main/python/clu_trainer.py
@@ -90,8 +90,12 @@ def compute_metrics(self, eval_pred: EvalPrediction) -> Dict[str, float]:
         ShortTaskDef("NER",       "conll-ner/", "train.txt",    "dev.txt",    "test.txt"),
         ShortTaskDef("POS",             "pos/", "train.txt",    "dev.txt",    "test.txt"),
         ShortTaskDef("Chunking",   "chunking/", "train.txt",    "test.txt",   "test.txt"), # this dataset has no dev
-        ShortTaskDef("Deps Head",  "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.heads",  "test.heads",  "test.heads"), # dev is included in train
-        ShortTaskDef("Deps Label", "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels", "test.labels", "test.labels", dual_mode=True) # dev is included in train
+        #ShortTaskDef("Hexa Term", "deps-wsj/", "train.labels.hexaterms", "dev.labels.hexaterms", "test.labels.hexaterms"),
+        #ShortTaskDef("Hexa NonTerm", "deps-wsj/", "train.labels.hexanonterms", "dev.labels.hexanonterms", "test.labels.hexanonterms"),
+        ShortTaskDef("Hexa Term",  "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels.hexaterms",  "test.labels.hexaterms",  "test.labels.hexaterms"), # dev is included in train
+        ShortTaskDef("Hexa NonTerm", "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels.hexanonterms", "test.labels.hexanonterms", "test.labels.hexanonterms") # dev is included in train
+        #ShortTaskDef("Deps Head",  "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.heads",  "test.heads",  "test.heads"), # dev is included in train
+        #ShortTaskDef("Deps Label", "deps-combined/", "wsjtrain-wsjdev-geniatrain-geniadev.labels", "test.labels", "test.labels", dual_mode=True) # dev is included in train
         #ShortTaskDef("Deps Head",  "deps-wsj/", "train.heads",  "dev.heads",  "test.heads"),
         #ShortTaskDef("Deps Label", "deps-wsj/", "train.labels", "dev.labels", "test.labels", dual_mode=True)
     ])