File size: 4,501 Bytes
9dd5ea5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 |
"""
Adds clue/cluewsc2020 samples to xwinograd
From: https://gist.github.com/jordiclive/26506ea7e897ad8270f9e793bdc285b5
"""
import json
import datasets
import pandas as pd
from datasets import load_dataset
def find_pronoun(x):
pronoun = x["target"]["span2_text"]
indices = [x["target"]["span2_index"], x["target"]["span2_index"] + len(pronoun)]
return [pronoun, indices, list(pronoun)]
def find_switch(x):
pronoun = x["target"]["span1_text"]
indices = [x["target"]["span1_index"], x["target"]["span1_index"] + len(pronoun)]
if x["label"] == 1:
label = False
else:
label = True
return [pronoun, indices, list(pronoun), label]
def convert_to_format(df):
df["pronoun"] = df.apply(find_pronoun, axis=1)
df["toks"] = df["text"].apply(lambda x: list(x))
df["switch"] = df.apply(find_switch, axis=1)
df.reset_index(inplace=True, drop=True)
lang = []
original = []
o_text = []
sent = []
toks = []
pronoun = []
switch = []
df["pronoun_to_replace"] = df["target"].apply(lambda x: x["span2_text"])
for i, df_text in df.groupby(["text", "pronoun_to_replace"]):
if len(df_text) == 1:
continue
df_text.reset_index(inplace=True, drop=True)
try:
if df_text["label"][0] != df_text["label"][1]:
df_text = df_text[:2]
elif df_text["label"][0] != df_text["label"][2] and len(df_text) > 2:
df_text = df_text.iloc[[0, 2], :]
df_text.reset_index(inplace=True, drop=True)
df_new = df_text[:1]
df_new["switch"] = df_new["switch"].apply(
lambda x: [df_text["switch"][0], df_text["switch"][1]]
)
lang.append("zh")
original.append("original")
o_text.append("?")
sent.append(df_new.iloc[0]["text"])
toks.append(df_new.iloc[0]["toks"])
pronoun.append(df_new.iloc[0]["pronoun"])
switch.append(df_new.iloc[0]["switch"])
except:
continue
total_df = pd.DataFrame(
{0: lang, 1: original, 2: o_text, 3: sent, 4: toks, 5: pronoun, 6: switch}
)
count = total_df[5].apply(lambda x: len(x[0]))
total_df[4] = total_df[4].apply(lambda x: json.dumps(x))
total_df[5] = total_df[5].apply(lambda x: json.dumps(x))
total_df[6] = total_df[6].apply(lambda x: json.dumps(x))
return total_df, count
def remove_at(i, s):
return s[:i] + s[i + 1 :]
def remove_to(x):
if x["count"] == 2:
return remove_at(x["sentence"].index("_") + 1, x["sentence"])
else:
return x["sentence"]
def get_original_splits():
# xwinograd directory is https://huggingface.co/datasets/Muennighoff/xwinograd/tree/9dbcc59f86f9e53e0b36480d806982499d877edc
for j, i in enumerate(["en", "jp", "ru", "pt", "fr", "zh"]):
if j == 0:
dfx = datasets.load_dataset("xwinograd", i)["test"].to_pandas()
dfx["lang"] = i
else:
df = datasets.load_dataset("xwinograd", i)["test"].to_pandas()
df["lang"] = i
dfx = pd.concat([dfx, df])
return dfx
def get_examples_from_clue():
# xwinograd directory is commit: https://huggingface.co/datasets/Muennighoff/xwinograd/tree/9dbcc59f86f9e53e0b36480d806982499d877edc
dataset = load_dataset("clue", "cluewsc2020")
df = dataset["train"].to_pandas()
df_val = dataset["validation"].to_pandas()
df = pd.concat([df, df_val])
new_examples, count = convert_to_format(df)
new_examples.reset_index(inplace=True, drop=True)
new_examples.to_csv(
"xwinograd/data/xwinograd.tsv", sep="\t", header=None, index=False
)
df_post = datasets.load_dataset("xwinograd", "zh")
df_post = df_post["test"].to_pandas()
df_post["count"] = count
df_post["sentence"] = df_post.apply(remove_to, axis=1)
df_post = df_post[["sentence", "option1", "option2", "answer"]]
return df_post
if __name__ == "__main__":
# xwinograd directory is commit: https://huggingface.co/datasets/Muennighoff/xwinograd/tree/9dbcc59f86f9e53e0b36480d806982499d877edc
dfx = get_original_splits()
df_post = get_examples_from_clue()
df_post.to_json("new_examples_updated.json", orient="split")
df_post["lang"] = "zh"
dfx = pd.concat([dfx, df_post])
dfx.drop_duplicates()
dfx.reset_index(inplace=True, drop=True)
dfx.to_json("all_examples_updated.json", orient="split") |