Datasets:

Modalities:
Text
Formats:
text
Size:
< 1K
ArXiv:
Libraries:
Datasets
License:
zdy023 commited on
Commit
12d83cf
1 Parent(s): a5a597d

ver Dec18th

Browse files

updated WikiHow task set with new subsets

README.md CHANGED
@@ -16,16 +16,12 @@ task set, 107,448 pages are crawled, and the dumped website data occupy about
16
  88 GiB totally.
17
 
18
  Several task definition templates are designed according to the functions of
19
- WikiHow app and 5,522 task definitions are instantiated through the template
20
- toolkit in Mobile-Env. This task set is named the *extended set*
21
- (`wikihow-extended.tar.xz`). There may be several faults that may make the
22
- system or the task fail in the auto-generated tasks. Therefore, 178 tasks are
23
- sampled from the extended set and have been verified by human beings to ensure
24
- correctness and stability, which is named the *canonical set*
25
- (`wikihow-canonical.tar.xz`). Owing to the limit of the budgets, only 70 tasks
26
- are tested using the proposed LLM-based agent in the corresponding pager.
27
- These 70 tasks are given in `wikihow-microcanon.tar.xz`. We call it the
28
- *canonical subset* or the *micro canonical set*.
29
 
30
  ### Website Data Replay
31
 
 
16
  88 GiB totally.
17
 
18
  Several task definition templates are designed according to the functions of
19
+ WikiHow app and task definitions are instantiated through the template toolkit
20
+ in Mobile-Env. 577 tasks are sampled from the extended set, which is named the
21
+ *canonical set* (`wikihow-canonical.tar.xz`). Owing to the limit of the
22
+ budgets, only 150 tasks are tested using the proposed LLM-based agent. These
23
+ 150 tasks are given in `wikihow-microcanon.tar.xz`. We call it the *canonical
24
+ subset* or the *micro canonical set*.
 
 
 
 
25
 
26
  ### Website Data Replay
27
 
instruction_rewriting/annotations/merged_doccano.json CHANGED
@@ -2310,5 +2310,87 @@
2310
  "Open the article \"Buy Stocks (for Beginners)\".",
2311
  "The target article is \"Buy Stocks (for Beginners)\"."
2312
  ]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2313
  }
2314
  ]
 
2310
  "Open the article \"Buy Stocks (for Beginners)\".",
2311
  "The target article is \"Buy Stocks (for Beginners)\"."
2312
  ]
2313
+ },
2314
+ {
2315
+ "id": 1302,
2316
+ "text": "Read the article and summarize the needed ingredients.",
2317
+ "task": "constant",
2318
+ "index": "0",
2319
+ "label": [
2320
+ "Read the article and give a summary of the needed ingredients.",
2321
+ "Give a summary of the needed ingredients according to the article.",
2322
+ "After reading the article, list the necessary ingredients.",
2323
+ "Read through the article and encapsulate the needed ingredients.",
2324
+ "Upon reading the article, note down the essential ingredients."
2325
+ ]
2326
+ },
2327
+ {
2328
+ "id": 1303,
2329
+ "text": "Read the article and summarize the needed items for the task.",
2330
+ "task": "constant",
2331
+ "index": "0",
2332
+ "label": [
2333
+ "Read the article and give a summary of the needed items.",
2334
+ "Give a summary of the needed items according to the article.",
2335
+ "After reading the article, list all the matierials required for this job.",
2336
+ "Read through the article and encapsulate the needed items.",
2337
+ "Upon reading the article, note down the necessary matierials."
2338
+ ]
2339
+ },
2340
+ {
2341
+ "id": 1304,
2342
+ "text": "Read the article and give a summary of each steps.",
2343
+ "task": "constant",
2344
+ "index": "0",
2345
+ "label": [
2346
+ "Read the article and summarize each steps.",
2347
+ "Give a summary of each steps according to the article.",
2348
+ "After reading the article, summarize each steps.",
2349
+ "Read through the article and provide a concise overview of every step.",
2350
+ "Upon reading the article, summarize the steps one by one.",
2351
+ "Following your reading of the article, summarize the steps outlined within it."
2352
+ ]
2353
+ },
2354
+ {
2355
+ "id": 1305,
2356
+ "text": "The summary should be given in an unnumbered list.",
2357
+ "task": "constant",
2358
+ "index": "0",
2359
+ "label": [
2360
+ "Please summarize in the form of an unnumbered list.",
2361
+ "It is best to present the summary as an unnumbered list.",
2362
+ "Present the summary in a list format without numbering.",
2363
+ "Depict your summary in the form of a bullet list.",
2364
+ "An unnumbered list should be used to present the summary.",
2365
+ "Create a bullet list to provide the summary.",
2366
+ "Present the summary by listing it without numerical order.",
2367
+ "Use a list format without numbering to present the summary."
2368
+ ]
2369
+ },
2370
+ {
2371
+ "id": 1306,
2372
+ "text": "The summary should be given in a numbered list.",
2373
+ "task": "constant",
2374
+ "index": "0",
2375
+ "label": [
2376
+ "Please summarize in the form of a numbered list.",
2377
+ "It is best to present the summary as a numbered list.",
2378
+ "Present the summary in a list format with numbers.",
2379
+ "A numbered list should be used to present the summary.",
2380
+ "Create a numbered list to provide the summary.",
2381
+ "Present the summary by listing it in numerical order.",
2382
+ "Use a list format with numbers to present the summary."
2383
+ ]
2384
+ },
2385
+ {
2386
+ "id": 1307,
2387
+ "text": "Read the article and answer the question concisely.",
2388
+ "task": "constant",
2389
+ "index": "0",
2390
+ "label": [
2391
+ "Read the article and offer a concise answer for my question.",
2392
+ "Read the article, then succinctly respond to the question.",
2393
+ "After reading the article, answer the question concisely"
2394
+ ]
2395
  }
2396
  ]
instruction_rewriting/annotations/pattern-question.txt ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ My question is: {<title>...}?
2
+ Answer my question: {<title>...}?
3
+ The question I wonder is: {<title>...}(?/.)
4
+ What I (wonder/want to (know/ask)) is: {<title>...}(?/.)
5
+ I (wonder/want to (know/ask)): {<title>...}(?/.)
6
+ (Tell/Answer) me: {<title>...}?
7
+ (Give/Offer) (me the answer/the answer to me): {<title>...}.
8
+ Provide me with the answer: {<title>...}.
instruction_rewriting/instruction_rewriting_wrapper.py CHANGED
@@ -22,6 +22,7 @@ class InstructionRewritingWrapper(base_wrapper.BaseWrapper):
22
  , article_command_pattern_file: str
23
  , categ_pattern_file: str
24
  , author_pattern_file: str
 
25
  , doccano_file: str
26
  ):
27
  # method __init__ {{{ #
@@ -35,6 +36,7 @@ class InstructionRewritingWrapper(base_wrapper.BaseWrapper):
35
  , article_command_pattern_file
36
  , categ_pattern_file
37
  , author_pattern_file
 
38
  , doccano_file
39
  , rng
40
  )
 
22
  , article_command_pattern_file: str
23
  , categ_pattern_file: str
24
  , author_pattern_file: str
25
+ , question_pattern_file: str
26
  , doccano_file: str
27
  ):
28
  # method __init__ {{{ #
 
36
  , article_command_pattern_file
37
  , categ_pattern_file
38
  , author_pattern_file
39
+ , question_pattern_file
40
  , doccano_file
41
  , rng
42
  )
instruction_rewriting/target_handlers.py CHANGED
@@ -192,19 +192,22 @@ class TransformerSet:
192
  , article_command_pattern_file: str
193
  , categ_pattern_file: str
194
  , author_pattern_file: str
 
195
  , doccano_file: str
196
  , rng: np.random.Generator
197
  ):
198
  # method __init__ {{{ #
199
  """
200
  Args:
201
- search_pattern_file: str
202
- article_pattern_file: str
203
- article_command_pattern_file: str
204
- categ_pattern_file: str
205
- author_pattern_file: str
206
- doccano_file: str as the path to the json file exported by doccano
207
- rng: np.random.Generator
 
 
208
  """
209
 
210
  self._search_template_lib: List[sentence_pattern.Item]
@@ -212,6 +215,7 @@ class TransformerSet:
212
  self._article_command_template_lib: List[sentence_pattern.Item]
213
  self._categ_template_lib: List[sentence_pattern.Item]
214
  self._author_template_lib: List[sentence_pattern.Item]
 
215
 
216
  self._search_template_regex: List[Pattern[str]]
217
  #self._article_template_regex: List[Pattern[str]]
@@ -225,6 +229,7 @@ class TransformerSet:
225
  self._article_command_template_lib, _ = parse_file(article_command_pattern_file)
226
  self._categ_template_lib, _ = parse_file(categ_pattern_file)
227
  self._author_template_lib, _ = parse_file(author_pattern_file)
 
228
 
229
  self._annotation_dict: Dict[str, List[str]] = {}
230
  with open(doccano_file) as f:
@@ -260,6 +265,8 @@ class TransformerSet:
260
  transformed = self._transform_categ(sentence)
261
  elif sentence.startswith("Check the author"):
262
  transformed = self._transform_author(sentence)
 
 
263
 
264
  elif sentence in self._annotation_dict:
265
  candidates = self._annotation_dict[sentence]
@@ -299,6 +306,8 @@ class TransformerSet:
299
  if template is None:
300
  return sentence
301
 
 
 
302
  target_pattern = template.get_targets()
303
  if not match_target(target_pattern[0], target_str):
304
  return sentence
@@ -338,6 +347,11 @@ class TransformerSet:
338
  and sentence.endswith(".")
339
  target_str = sentence[25:-1]
340
  return self._apply_new_template(target_str, self._author_template_lib)
 
 
 
 
 
341
 
342
  def _apply_new_template( self
343
  , target_str: str
 
192
  , article_command_pattern_file: str
193
  , categ_pattern_file: str
194
  , author_pattern_file: str
195
+ , question_pattern_file: str
196
  , doccano_file: str
197
  , rng: np.random.Generator
198
  ):
199
  # method __init__ {{{ #
200
  """
201
  Args:
202
+ search_pattern_file (str): pattern file for search instructions
203
+ article_pattern_file (str): pattern file for article instructions
204
+ article_command_pattern_file (str): pattern file for article
205
+ instructions in "command" field
206
+ categ_pattern_file (str): pattern file for category instructions
207
+ author_pattern_file (str): pattern file for author instructions
208
+ question_pattern_file (str): pattern file for WikiHowNFQA questions
209
+ doccano_file (str): the path to the json file exported by doccano
210
+ rng (np.random.Generator): used to generate random indices
211
  """
212
 
213
  self._search_template_lib: List[sentence_pattern.Item]
 
215
  self._article_command_template_lib: List[sentence_pattern.Item]
216
  self._categ_template_lib: List[sentence_pattern.Item]
217
  self._author_template_lib: List[sentence_pattern.Item]
218
+ self._question_template_lib: List[sentence_pattern.Item]
219
 
220
  self._search_template_regex: List[Pattern[str]]
221
  #self._article_template_regex: List[Pattern[str]]
 
229
  self._article_command_template_lib, _ = parse_file(article_command_pattern_file)
230
  self._categ_template_lib, _ = parse_file(categ_pattern_file)
231
  self._author_template_lib, _ = parse_file(author_pattern_file)
232
+ self._question_template_lib, _ = parse_file(question_pattern_file)
233
 
234
  self._annotation_dict: Dict[str, List[str]] = {}
235
  with open(doccano_file) as f:
 
265
  transformed = self._transform_categ(sentence)
266
  elif sentence.startswith("Check the author"):
267
  transformed = self._transform_author(sentence)
268
+ elif sentence.startswith("My question is"):
269
+ transformed: str = self._transform_question(sentence)
270
 
271
  elif sentence in self._annotation_dict:
272
  candidates = self._annotation_dict[sentence]
 
306
  if template is None:
307
  return sentence
308
 
309
+ # if the template seems to match the sentence according to the regex,
310
+ # but the target isn't consistent, we should ignore it.
311
  target_pattern = template.get_targets()
312
  if not match_target(target_pattern[0], target_str):
313
  return sentence
 
347
  and sentence.endswith(".")
348
  target_str = sentence[25:-1]
349
  return self._apply_new_template(target_str, self._author_template_lib)
350
+ def _transform_question(self, sentence: str) -> str:
351
+ assert sentence.startswith("My question is: ")\
352
+ and sentence.endswith("?")
353
+ target_str: str = sentence[16:-1]
354
+ return self._apply_new_template(target_str, self._question_template_lib)
355
 
356
  def _apply_new_template( self
357
  , target_str: str
wikihow-canonical.tar.xz CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4e7168327b2aae6e85bf5561fc65b8ad5bdad71791fb8fccdeeaa21c03102a7f
3
- size 30684
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e00499008ec1043ecdf8d3d33c5fac6ad5faee4ca5a9fc0676011fc065dcdee7
3
+ size 123828
wikihow-extended.tar.xz DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:df7650423ada1a94689f0b30d0d11a186da5749dff5167924716bf109c16ef3c
3
- size 946740
 
 
 
 
wikihow-microcanon.tar.xz CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:33279811d397b5741b791130cca8e6d5990ae6d46e70dbe201cdc48f24614a55
3
- size 13312
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a94694a43657122d19a9acb8e426c28b7de6dc52e255a69562b194d3a0fe806
3
+ size 14484