name="davidson_train")
davidson_tr.read()
davidson_dv = DataSet(os.path.join("data", "davidson.dv.csv"),
reader=csvreader,
formatter=df,
name="davidson_dev")
davidson_dv.read()
davidson_te = DataSet(os.path.join("data", "davidson.te.csv"),
reader=csvreader,
formatter=df,
name="davidson_test")
davidson_te.read()
features = Features(get_feature_functions(mname))
primary_train_fs, aux_train_fs, dev_fs, test_fs_primary, test_fs_aux = features.load(
waseem_tr_composite, davidson_tr, waseem_de_composite,
waseem_te_composite, davidson_te)
print("Number of features in primary: {0}".format(
primary_train_fs[0].shape[1]))
print("Number of features aux (=): {0}".format(aux_train_fs[0].shape[1]))
model = MTMLP(primary_train_fs[0].shape[1], get_model_shape(), 3, 3)
if gpu():
model.cuda()
if model_exists(mname) and os.getenv("TRAIN").lower() not in [
"y", "1", "t", "yes"
logger.info("Loading DB {0}".format(args.db))
db = FeverDocDB(args.db)
mname = args.model
logger.info("Model name is {0}".format(mname))
ffns = []
if args.sentence:
logger.info("Model is Sentence level")
ffns.append(SentenceLevelTermFrequencyFeatureFunction(db, naming=mname))
else:
logger.info("Model is Document level")
ffns.append(TermFrequencyFeatureFunction(db,naming=mname))
f = Features(mname,ffns)
jlr = JSONLineReader()
formatter = FEVERGoldFormatter(None, FEVERLabelSchema(),filtering=args.filtering)
train_ds = DataSet(file=args.train, reader=jlr, formatter=formatter)
dev_ds = DataSet(file=args.dev, reader=jlr, formatter=formatter)
train_ds.read()
dev_ds.read()
test_ds = None
if args.test is not None:
test_ds = DataSet(file=args.test, reader=jlr, formatter=formatter)
test_ds.read()
def format_line(self, line):
annotation = self.label_schema.get_id(
line["Stance"]) if "Stance" in line else None
return {
"claim": line["Headline"],
"evidence": line["Body ID"],
"label": annotation
}
if __name__ == "__main__":
bodies = Bodies("data/fnc-1/train_bodies.csv",
"data/fnc-1/competition_test_bodies.csv")
f = Features([FNCTermFrequencyFeatureFunction(bodies)])
csvr = CSVReader()
formatter = FNCFormatter(FNCLabelSchema())
train_ds = DataSet(file="data/fnc-1/train_stances.csv",
reader=csvr,
formatter=formatter)
test_ds = DataSet(file="data/fnc-1/competition_test_stances.csv",
reader=csvr,
formatter=formatter)
train_ds.read()
test_ds.read()
train_feats, _, test_feats = f.load(train_ds, None, test_ds)
dataset.read()
waseem_dv_composite.add(dataset)
davidson_te = DataSet(os.path.join("data", "davidson.te.csv"),
reader=csvreader,
formatter=df,
name="davidson_test")
davidson_te.read()
davidson_dv = DataSet(os.path.join("data", "davidson.dv.csv"),
reader=csvreader,
formatter=df,
name="davidson_dev")
davidson_dv.read()
features = Features(get_feature_functions(mname))
train_fs, dev_fs, test_fs = features.load(waseem_tr_composite,
waseem_dv_composite, davidson_te)
print("Number of features: {0}".format(train_fs[0].shape[1]))
model = MLP(train_fs[0].shape[1], get_model_shape(), 3)
if gpu():
model.cuda()
if model_exists(mname) and os.getenv("TRAIN").lower() not in [
"y", "1", "t", "yes"
]:
model.load_state_dict(torch.load("models/{0}.model".format(mname)))
else:
train(model,
db = FeverDocDB(args.db)
mname = args.model
logger.info("Model name is {0}".format(mname))
ffns = []
if args.sentence:
logger.info("Model is Sentence level")
ffns.append(SentenceLevelTermFrequencyFeatureFunction(db,
naming=mname))
else:
logger.info("Model is Document level")
ffns.append(TermFrequencyFeatureFunction(db, naming=mname))
f = Features(mname, ffns)
f.load_vocab(mname)
jlr = JSONLineReader()
formatter = FEVERGoldFormatter(None, FEVERLabelSchema())
test_ds = DataSet(file=args.test, reader=jlr, formatter=formatter)
test_ds.read()
feats = f.lookup(test_ds)
input_shape = feats[0].shape[1]
model = SimpleMLP(input_shape, 100, 3)
if gpu():
model.cuda()
print('Db is ', db)
mname = args.model
logger.info("Model name is {0}".format(mname))
ffns = []
if args.sentence:
logger.info("Model is Sentence level")
ffns.append(SentenceLevelTermFrequencyFeatureFunction(db,
naming=mname))
else:
logger.info("Model is Document level")
ffns.append(TermFrequencyFeatureFunction(db, naming=mname))
f = Features(mname, ffns)
jlr = JSONLineReader()
formatter = FEVERGoldFormatter(None,
FEVERLabelSchema(),
filtering=args.filtering)
train_ds = DataSet(file=args.train, reader=jlr, formatter=formatter)
dev_ds = DataSet(file=args.dev, reader=jlr, formatter=formatter)
train_ds.read()
dev_ds.read()
test_ds = None
if args.test is not None:
test_ds = DataSet(file=args.test, reader=jlr, formatter=formatter)
def __init__(self):
super().__init__(["related", "unrelated"])
if __name__ == "__main__":
SimpleRandom.set_seeds()
maxdoc = sys.argv[1]
ns_docsize = sys.argv[2]
db = FeverDocDB("data/fever/fever.db")
idx = set(db.get_doc_ids())
mname = "2way-p{0}-{1}".format(maxdoc, ns_docsize)
f = Features([SentenceTermFrequencyFeatureFunction(db, naming=mname)])
jlr = JSONLineReader()
formatter = FEVERSentenceFormatter(idx, db, RelatedLabelSchema())
train_ds = DataSet(
file="data/fever/train.ns.pages.p{0}.jsonl".format(ns_docsize),
reader=jlr,
formatter=formatter)
dev_ds = DataSet(file="data/fever/dev.pages.p{0}.jsonl".format(maxdoc),
reader=jlr,
formatter=formatter)
test_ds = DataSet(file="data/fever/test.pages.p{0}.jsonl".format(maxdoc),
reader=jlr,
formatter=formatter)
reader=csvreader,
name="davidson_train")
davidson_dv_dataset = DataSet(os.path.join("data", "davidson.dv.csv"),
formatter=df,
reader=csvreader,
name="davidson_dev")
davidson_te_dataset = DataSet(os.path.join("data", "davidson.te.csv"),
formatter=df,
reader=csvreader,
name="davidson_test")
davidson_tr_dataset.read()
davidson_dv_dataset.read()
davidson_te_dataset.read()
features = Features(get_feature_functions(mname))
train_fs, dev_fs, test_fs = features.load(davidson_tr_dataset,
davidson_dv_dataset,
davidson_te_dataset)
print("Number of features: {0}".format(train_fs[0].shape[1]))
model = MLP(train_fs[0].shape[1], get_model_shape(), 3)
if gpu():
model.cuda()
if model_exists(mname) and os.getenv("TRAIN").lower() not in [
"y", "1", "t", "yes"
]:
model.load_state_dict(torch.load("models/{0}.model".format(mname)))
else:
"claim": line["Headline"],
"evidence": line["Body ID"],
"label": annotation
}
if __name__ == "__main__":
db = FeverDocDB("data/fever/drqa.db")
idx = set(db.get_doc_ids())
fnc_bodies = Bodies("data/fnc-1/train_bodies.csv",
"data/fnc-1/competition_test_bodies.csv")
fever_bodies = db
f = Features(
[FeverOrFNCTermFrequencyFeatureFunction(fever_bodies, fnc_bodies)])
csvr = CSVReader()
jlr = JSONLineReader()
fnc_formatter = FNCFormatter2(FNCSimpleLabelSchema())
fever_formatter = FEVERPredictionsFormatter(idx, FEVERLabelSchema())
train_ds = DataSet(file="data/fnc-1/train_stances.csv",
reader=csvr,
formatter=fnc_formatter)
dev_ds = DataSet(file="data/fnc-1/competition_test_stances.csv",
reader=csvr,
formatter=fnc_formatter)
test_ds = DataSet(file="data/fever/fever.dev.pages.p5.jsonl",
reader=jlr,
formatter=fever_formatter)