dataset-construction/src/ndb_data/sample_questions_1000.py [117:251]: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - continue else: if ( len(question["answer"]) > 2 or len(question["derivations"]) == 1 ): if random.random() < 0.95: continue else: if random.random() < 0.3: continue # if question["type"] == "bool" and "TRUE" in question["answer"]: if random.random() < 0.3: continue # Less than 8 facts if q_bin == 5: # Drop half of the facts if random.random() < 0.5: continue if q_bin == 4: # Drop half of the facts if random.random() < 0.5: continue if q_bin == 3: # Drop half of the facts if random.random() < 0.6: continue if q_bin == 2: # Drop half of the facts if random.random() < 0.9: continue if q_bin == 1: # Drop 80% of the facts if question["type"] == "bool": if random.random() < 0: continue else: if random.random() < 0.5: continue if q_bin == 0: # Drop 90% of the facts if random.random() < 0.5: continue if q_bin < 4: if random.random() < 0.6: continue # if ( # question["type"] == "bool" # and "FALSE" in question["answer"] # and random.randint(0, 100) > 55 # ): # continue # # if ( # "complex" not in question["id"] and # question["type"] in {"argmax", "argmin"} # and random.randint(0, 100) > 30 # ): # continue # # # # if q_bin == 0 and random.randint(0, 100) > 30: # continue # if ( # question["type"] != "bool" # and q_bin == 1 # and random.randint(0, 100) > 40 # ): # continue # if ( # # question["type"] != "bool" # and q_bin == 2 # and random.randint(0, 100) > 30 # ): # continue # if q_bin == 3 and random.randint(0, 100) > 90: # continue # if q_bin == 4 and random.randint(0, 100) > 9: # continue # if q_bin == 5 and random.randint(0, 100) > 90: # continue # if q_bin == 6 and random.randint(0, 100) > 90: # continue # if q_bin == 7 and random.randint(0, 100) > 90: # continue # if q_bin == 8 and random.randint(0, 100) > 90: # continue counts_types[question["type"]] += 1 counts_facts[len(question["facts"])] += 1 if "complex" in question["id"] or "join" in question["id"]: complex_counts_types[question["type"]] += 1 complex_counts_facts[len(question["facts"])] += 1 instance["queries"].append(question) del instance["all_queries"] if len(instance["queries"]): of.write(json.dumps(instance) + "\n") db_sizes[len(instance["queries"])] += 1 for k, v in added_q_type_bin.items(): print(k, len(v)) print(added) # print(q_bin) # print(q_type) # print(q_type_bin) print("bins") for i in range(0, 9): print(i, counts_bins[i], complex_counts_bins[i]) print("lens") for i in range(0, 26): print(i, counts_facts[i], complex_counts_facts[i]) for k, v in counts_types.items(): print(k, v, complex_counts_types[k]) print(db_sizes) - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - dataset-construction/src/ndb_data/sample_questions_500.py [117:251]: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - continue else: if ( len(question["answer"]) > 2 or len(question["derivations"]) == 1 ): if random.random() < 0.95: continue else: if random.random() < 0.3: continue # if question["type"] == "bool" and "TRUE" in question["answer"]: if random.random() < 0.3: continue # Less than 8 facts if q_bin == 5: # Drop half of the facts if random.random() < 0.5: continue if q_bin == 4: # Drop half of the facts if random.random() < 0.5: continue if q_bin == 3: # Drop half of the facts if random.random() < 0.6: continue if q_bin == 2: # Drop half of the facts if random.random() < 0.9: continue if q_bin == 1: # Drop 80% of the facts if question["type"] == "bool": if random.random() < 0: continue else: if random.random() < 0.5: continue if q_bin == 0: # Drop 90% of the facts if random.random() < 0.5: continue if q_bin < 4: if random.random() < 0.6: continue # if ( # question["type"] == "bool" # and "FALSE" in question["answer"] # and random.randint(0, 100) > 55 # ): # continue # # if ( # "complex" not in question["id"] and # question["type"] in {"argmax", "argmin"} # and random.randint(0, 100) > 30 # ): # continue # # # # if q_bin == 0 and random.randint(0, 100) > 30: # continue # if ( # question["type"] != "bool" # and q_bin == 1 # and random.randint(0, 100) > 40 # ): # continue # if ( # # question["type"] != "bool" # and q_bin == 2 # and random.randint(0, 100) > 30 # ): # continue # if q_bin == 3 and random.randint(0, 100) > 90: # continue # if q_bin == 4 and random.randint(0, 100) > 9: # continue # if q_bin == 5 and random.randint(0, 100) > 90: # continue # if q_bin == 6 and random.randint(0, 100) > 90: # continue # if q_bin == 7 and random.randint(0, 100) > 90: # continue # if q_bin == 8 and random.randint(0, 100) > 90: # continue counts_types[question["type"]] += 1 counts_facts[len(question["facts"])] += 1 if "complex" in question["id"] or "join" in question["id"]: complex_counts_types[question["type"]] += 1 complex_counts_facts[len(question["facts"])] += 1 instance["queries"].append(question) del instance["all_queries"] if len(instance["queries"]): of.write(json.dumps(instance) + "\n") db_sizes[len(instance["queries"])] += 1 for k, v in added_q_type_bin.items(): print(k, len(v)) print(added) # print(q_bin) # print(q_type) # print(q_type_bin) print("bins") for i in range(0, 9): print(i, counts_bins[i], complex_counts_bins[i]) print("lens") for i in range(0, 26): print(i, counts_facts[i], complex_counts_facts[i]) for k, v in counts_types.items(): print(k, v, complex_counts_types[k]) print(db_sizes) - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -