def mk_test_graphs(hconf, dconf):
"Generate graphs for test data"
econf = hconf.test_evaluation
if econf is None:
return
with Torpor('creating test graphs'):
edus = concat_l(dpack.edus for dpack in dconf.pack.values())
gold = to_predictions(dconf.pack)
_mk_econf_graphs(hconf, edus, gold, econf, None)
def mk_test_graphs(hconf, dconf):
"Generate graphs for test data"
econf = hconf.test_evaluation
if econf is None:
return
with Torpor('creating test graphs'):
edus = concat_l(dpack.edus for dpack in dconf.pack.values())
gold = to_predictions(dconf.pack)
_mk_econf_graphs(hconf, edus, gold, econf, None)
def mk_graphs(hconf, dconf):
"Generate graphs for the gold data and for one of the folds"
with Torpor('creating gold graphs'):
_mk_gold_graphs(hconf, dconf)
fold = sorted(set(dconf.folds.values()))[0]
with Torpor('creating graphs for fold {}'.format(fold), sameline=False):
test_pack = select_testing(dconf.pack, dconf.folds, fold)
edus = concat_l(dpack.edus for dpack in test_pack.values())
gold = to_predictions(test_pack)
jobs = []
for econf in hconf.detailed_evaluations:
jobs.extend(_mk_econf_graphs(hconf, edus, gold, econf, fold))
Parallel(n_jobs=hconf.runcfg.n_jobs, verbose=True)(jobs)
def _mk_gold_graphs(hconf, dconf):
"Generate graphs for a single configuration"
# output path
output_dir = fp.join(hconf.report_dir_path(None), 'graphs-gold')
settings =\
GraphSettings(hide=None,
select=hconf.graph_docs,
unrelated=False,
timeout=15,
quiet=True)
predictions = to_predictions(dconf.pack)
edus = concat_l(dpack.edus for dpack in dconf.pack.values())
graph_all(edus, predictions, settings, output_dir)
def mk_graphs(hconf, dconf):
"Generate graphs for the gold data and for one of the folds"
with Torpor('creating gold graphs'):
_mk_gold_graphs(hconf, dconf)
fold = sorted(set(dconf.folds.values()))[0]
with Torpor('creating graphs for fold {}'.format(fold),
sameline=False):
test_pack = select_testing(dconf.pack, dconf.folds, fold)
edus = concat_l(dpack.edus for dpack in test_pack.values())
gold = to_predictions(test_pack)
jobs = []
for econf in hconf.detailed_evaluations:
jobs.extend(_mk_econf_graphs(hconf, edus, gold, econf, fold))
hconf.parallel(jobs)
def mk_graphs(lconf, dconf):
"Generate graphs for the gold data and for one of the folds"
with Torpor('creating gold graphs'):
_mk_gold_graphs(lconf, dconf)
fold = sorted(set(dconf.folds.values()))[0]
with Torpor('creating graphs for fold {}'.format(fold),
sameline=False):
test_pack = select_testing(dconf.pack, dconf.folds, fold)
edus = concat_l(dpack.edus for dpack in test_pack.values())
gold = to_predictions(test_pack)
jobs = []
for econf in DETAILED_EVALUATIONS:
jobs.extend(_mk_econf_graphs(lconf, edus, gold, econf, fold))
Parallel(n_jobs=-1)(jobs)
def _mk_gold_graphs(hconf, dconf):
"Generate graphs for a single configuration"
# output path
output_dir = fp.join(hconf.report_dir_path(None),
'graphs-gold')
settings =\
GraphSettings(hide=None,
select=hconf.graph_docs,
unrelated=False,
timeout=15,
quiet=True)
predictions = to_predictions(dconf.pack)
edus = concat_l(dpack.edus for dpack in dconf.pack.values())
graph_all(edus, predictions, settings, output_dir)
def _evaluations():
"the evaluations we want to run"
# non-prob mst decoder (dp learners don't do probs)
nonprob_mst = MstDecoder(MstRootStrategy.fake_root, False)
#
learners = []
learners.extend(_LOCAL_LEARNERS)
learners.extend(l(nonprob_mst) for l in _STRUCTURED_LEARNERS)
ipairs = list(itr.product(learners, _INTRA_INTER_CONFIGS))
res = concat_l([
concat_l(_core_parsers(l) for l in learners),
concat_l(_mk_basic_intras(l, x) for l, x in ipairs),
concat_l(_mk_sorc_intras(l, x) for l, x in ipairs),
concat_l(_mk_dorc_intras(l, x) for l, x in ipairs),
concat_l(_mk_last_intras(l, x) for l, x in ipairs),
])
return [x for x in res if not _is_junk(x)]
def _evaluations():
"the evaluations we want to run"
# non-prob mst decoder (dp learners don't do probs)
nonprob_mst = Keyed('', MstDecoder(MstRootStrategy.fake_root, False))
nonprob_mst = tc_decoder(nonprob_mst)
nonprob_mst = nonprob_mst.payload
#
learners = []
learners.extend(_LOCAL_LEARNERS)
learners.extend(l(nonprob_mst) for l in _STRUCTURED_LEARNERS)
ipairs = list(itr.product(learners, _INTRA_INTER_CONFIGS))
res = concat_l([
concat_l(_core_parsers(l) for l in learners),
concat_l(_mk_basic_intras(l, x) for l, x in ipairs),
concat_l(_mk_sorc_intras(l, x) for l, x in ipairs),
concat_l(_mk_dorc_intras(l, x) for l, x in ipairs),
concat_l(_mk_last_intras(l, x) for l, x in ipairs),
])
return [x for x in res if not _is_junk(x)]
