def start_generate_jobs_for_train_val(generator: InstanceGenerator,
name_prefix):
# claim ids split to train/val
print("Loading data ....")
d_ids: List[int] = list(load_train_claim_ids())
claims = get_claims_from_ids(d_ids)
train, val = split_7_3(claims)
train_cids = {str(t['cId']) for t in train}
val_cids = {str(t['cId']) for t in val}
qk_candidate: List[QKUnit] = load_qk_candidate_train()
print("Generate instances : train")
qk_candidate_train: List[QKUnit] = list(
[qk for qk in qk_candidate if qk[0].query_id in train_cids])
qk_candidate_val = list(
[qk for qk in qk_candidate if qk[0].query_id in val_cids])
def worker_factory(out_dir):
return QCKWorker(qk_candidate_train, generator, out_dir)
runner = JobRunner(job_man_dir, 378, name_prefix + "_train",
worker_factory)
runner.start()
print("Generate instances : val")
def worker_factory(out_dir):
return QCKWorker(qk_candidate_val, generator, out_dir)
runner = JobRunner(job_man_dir, 162, name_prefix + "_val", worker_factory)
runner.start()
def start_generate_jobs_for_train_val(
generator_functor: Callable[[Dict[int, List[Tuple[List[str], float]]]],
CPPNCGeneratorInterface], writer,
name_prefix):
# claim ids split to train/val
d_ids: List[int] = list(load_train_claim_ids())
claims = get_claims_from_ids(d_ids)
train, val = split_7_3(claims)
data = load_from_pickle("pc_train_a_passages")
entries, all_passages = data
cid_to_passages: Dict[int, List[Tuple[List[str], float]]] = {
claim['cId']: p
for claim, p in entries
}
generator = generator_functor(cid_to_passages)
print("Generate instances : train")
def worker_factory(out_dir):
return CPPNCWorker(train, generator, writer, out_dir)
runner = JobRunner(job_man_dir, 378, name_prefix + "_train",
worker_factory)
runner.start()
print("Generate instances : val")
def worker_factory(out_dir):
return CPPNCWorker(val, generator, writer, out_dir)
runner = JobRunner(job_man_dir, 162, name_prefix + "_val", worker_factory)
runner.start()
def load_data_point_50_train_val(split):
d_ids = list(load_claim_ids_for_split("train"))
train_ids, val_ids = split_7_3(d_ids)
print(len(d_ids), len(train_ids), len(val_ids))
if split == "train":
sel_ids = train_ids
elif split == "val":
sel_ids = val_ids
else:
sel_ids = d_ids
print("Taking all claims in {}".format(split))
claims = get_claims_from_ids(sel_ids)
all_data_points = get_candidates(claims, False)
return all_data_points
def load_claims_for_sub_split(sub_split) -> List[Dict]:
if sub_split in ["train", "val"]:
split = "train"
d_ids: List[int] = list(load_train_claim_ids())
claims = get_claims_from_ids(d_ids)
train, val = split_7_3(claims)
if sub_split == "train":
return train
elif sub_split == "val":
return val
else:
assert False
else:
d_ids: Iterable[int] = load_claim_ids_for_split(sub_split)
claims = get_claims_from_ids(d_ids)
return claims
def main():
print("Loading data ....")
d_ids: List[int] = list(load_train_claim_ids())
claims = get_claims_from_ids(d_ids)
train, val = split_7_3(claims)
val_cids = {str(t['cId']) for t in val}
qk_candidate: List[QKUnit] = load_qk_candidate_train()
qk_candidate_val = list(
[qk for qk in qk_candidate if qk[0].query_id in val_cids])
print(qk_candidate_val[0][0])
for q, kdp_list in qk_candidate_val[1:9]:
job_id = request_kdp_eval(kdp_list)
print('qid:', q.query_id)
print('job_id', job_id)
def train_split():
d_ids: List[int] = list(load_train_claim_ids())
claims = get_claims_from_ids(d_ids)
train, val = split_7_3(claims)
return claims, val