File size: 10,369 Bytes
fc35a48 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 |
"""
This file come from: https://github.com/microsoft/ToRA/blob/main/src/utils/parser.py
"""
import re
from typing import Any, Dict
def _fix_fracs(string):
substrs = string.split("\\frac")
new_str = substrs[0]
if len(substrs) > 1:
substrs = substrs[1:]
for substr in substrs:
new_str += "\\frac"
if len(substr) > 0 and substr[0] == "{":
new_str += substr
else:
try:
assert len(substr) >= 2
except:
return string
a = substr[0]
b = substr[1]
if b != "{":
if len(substr) > 2:
post_substr = substr[2:]
new_str += "{" + a + "}{" + b + "}" + post_substr
else:
new_str += "{" + a + "}{" + b + "}"
else:
if len(substr) > 2:
post_substr = substr[2:]
new_str += "{" + a + "}" + b + post_substr
else:
new_str += "{" + a + "}" + b
string = new_str
return string
def _fix_a_slash_b(string):
if len(string.split("/")) != 2:
return string
a = string.split("/")[0]
b = string.split("/")[1]
try:
if "sqrt" not in a:
a = int(a)
if "sqrt" not in b:
b = int(b)
assert string == "{}/{}".format(a, b)
new_string = "\\frac{" + str(a) + "}{" + str(b) + "}"
return new_string
except:
return string
def _fix_sqrt(string):
_string = re.sub(r"\\sqrt(\w+)", r"\\sqrt{\1}", string)
return _string
def strip_string(string):
string = str(string).strip()
# linebreaks
string = string.replace("\n", "")
# right "."
string = string.rstrip(".")
# remove inverse spaces
string = string.replace("\\!", "")
string = string.replace("\\ ", "")
# replace \\ with \
string = string.replace("\\\\", "\\")
string = string.replace("\\\\", "\\")
# replace tfrac and dfrac with frac
string = string.replace("tfrac", "frac")
string = string.replace("dfrac", "frac")
# remove \left and \right
string = string.replace("\\left", "")
string = string.replace("\\right", "")
# Remove unit: miles, dollars if after is not none
_string = re.sub(r"\\text{.*?}$", "", string).strip()
if _string != "" and _string != string:
# print("Warning: unit not removed: '{}' -> '{}'".format(string, _string))
string = _string
# Remove circ (degrees)
string = string.replace("^{\\circ}", "")
string = string.replace("^\\circ", "")
# remove dollar signs
string = string.replace("\\$", "")
string = string.replace("$", "")
string = string.replace("\\text", "")
string = string.replace("x\\in", "")
# remove percentage
string = string.replace("\\%", "")
string = string.replace("\%", "")
string = string.replace("%", "")
# " 0." equivalent to " ." and "{0." equivalent to "{." Alternatively, add "0" if "." is the start of the string
string = string.replace(" .", " 0.")
string = string.replace("{.", "{0.")
# cdot
string = string.replace("\\cdot", "")
# inf
string = string.replace("infinity", "\\infty")
if "\\infty" not in string:
string = string.replace("inf", "\\infty")
string = string.replace("+\\inity", "\\infty")
# and
string = string.replace("and", "")
string = string.replace("\\mathbf", "")
# use regex to remove \mbox{...}
string = re.sub(r"\\mbox{.*?}", "", string)
# quote
string.replace("'", "")
string.replace("\"", "")
# i, j
if "j" in string and "i" not in string:
string = string.replace("j", "i")
# replace a.000b where b is not number or b is end, with ab, use regex
string = re.sub(r"(\d+)\.0+([^\d])", r"\1\2", string)
string = re.sub(r"(\d+)\.0+$", r"\1", string)
# if empty, return empty string
if len(string) == 0:
return string
if string[0] == ".":
string = "0" + string
# to consider: get rid of e.g. "k = " or "q = " at beginning
if len(string.split("=")) == 2:
if len(string.split("=")[0]) <= 2:
string = string.split("=")[1]
string = _fix_sqrt(string)
string = string.replace(" ", "")
# \frac1b or \frac12 --> \frac{1}{b} and \frac{1}{2}, etc. Even works with \frac1{72} (but not \frac{72}1). Also does a/b --> \\frac{a}{b}
string = _fix_fracs(string)
# NOTE: X/Y changed to \frac{X}{Y} in dataset, but in simple cases fix in case the model output is X/Y
string = _fix_a_slash_b(string)
return string
def extract_answer(pred_str):
if 'boxed' in pred_str:
ans = pred_str.split('boxed')[-1]
if len(ans) == 0:
return ""
elif (ans[0] == '{'):
stack = 1
a = ''
for c in ans[1:]:
if (c == '{'):
stack += 1
a += c
elif (c == '}'):
stack -= 1
if (stack == 0): break
a += c
else:
a += c
else:
a = ans.split('$')[0].strip()
pred=a
elif ('he answer is' in pred_str):
pred = pred_str.split('he answer is')[-1].strip()
elif extract_program_output(pred_str) != "":
# fall back to program
pred = extract_program_output(pred_str)
else: # use the last number
pattern = '-?\d*\.?\d+'
pred = re.findall(pattern, pred_str.replace(",", ""))
if(len(pred) >= 1):
pred = pred[-1]
else: pred = ''
# multiple line
pred = pred.split("\n")[0]
if pred != "" and pred[0] == ":":
pred = pred[1:]
if pred != "" and pred[-1] == ".":
pred = pred[:-1]
if pred != "" and pred[-1] == "/":
pred = pred[:-1]
pred = strip_string(pred)
return pred
def extract_program(result: str, last_only=True):
"""
extract the program after "```python", and before "```"
"""
program = ""
start = False
for line in result.split("\n"):
if line.startswith("```python"):
if last_only:
program = "" # only extract the last program
else:
program += "\n# ========\n"
start = True
elif line.startswith("```"):
start = False
elif start:
program += line + "\n"
return program
def extract_program_output(pred_str):
"""
extract output between the last ```output\n...\n```
"""
if "```output" not in pred_str:
return ""
if '```output' in pred_str:
pred_str = pred_str.split('```output')[-1]
if '```' in pred_str:
pred_str = pred_str.split('```')[0]
output = pred_str.strip()
return output
def parse_ground_truth(example: Dict[str, Any], data_name):
if 'gt_cot' in example:
return example['gt_cot'], strip_string(example['gt'])
# parse ground truth
if data_name in ["math", 'ocw']:
gt_cot = example['solution']
gt_ans = extract_answer(gt_cot)
elif data_name == "gsm8k":
gt_cot, gt_ans = example['answer'].split("####")
elif data_name == "gsm-hard":
gt_cot, gt_ans = example['code'], example['target']
elif data_name == "svamp":
gt_cot, gt_ans = example['Equation'], example['Answer']
elif data_name == "asdiv":
gt_cot = example['formula']
gt_ans = re.sub(r"\(.*?\)", "", example['answer'])
elif data_name == "mawps":
gt_cot, gt_ans = None, example['target']
elif data_name == "tabmwp":
gt_cot = example['solution']
gt_ans = example['answer']
if example['ans_type'] in ['integer_number', 'decimal_number']:
if '/' in gt_ans:
gt_ans = int(gt_ans.split('/')[0]) / int(gt_ans.split('/')[1])
elif ',' in gt_ans:
gt_ans = float(gt_ans.replace(',', ''))
elif '%' in gt_ans:
gt_ans = float(gt_ans.split('%')[0]) / 100
else:
gt_ans = float(gt_ans)
elif data_name == "bbh":
gt_cot, gt_ans = None, example['target']
else:
raise NotImplementedError(data_name)
# post process
gt_cot = str(gt_cot).strip()
gt_ans = strip_string(gt_ans)
return gt_cot, gt_ans
def parse_question(example, data_name):
question = ""
if data_name == "asdiv":
question = f"{example['body'].strip()} {example['question'].strip()}"
elif data_name == "svamp":
body = example["Body"].strip()
if not body.endswith("."):
body = body + "."
question = f'{body} {example["Question"].strip()}'
elif data_name == "tabmwp":
title_str = f'regarding "{example["table_title"]}" ' if example['table_title'] else ""
question = f'Read the following table {title_str}and answer a question:\n'
question += f'{example["table"]}\n{example["question"]}'
if example['choices']:
question += f' Please select from the following options: {example["choices"]}'
else:
for key in ['question', 'problem', 'Question', 'input']:
if key in example:
question = example[key]
break
assert question != ""
return question.strip()
def run_execute(executor, result, prompt_type, execute=False):
if not result or result == 'error':
return None, None
report = None
if "program_only" in prompt_type:
prediction = extract_program_output(result)
elif prompt_type in ["pot", "pal"] and execute:
code = extract_program(result)
prediction, report = executor.apply(code)
else:
prediction = extract_answer(result)
prediction = strip_string(prediction)
return prediction, report
|