ver Dec18th

updated WikiHow task set with new subsets

README.md

@@ -16,16 +16,12 @@ task set, 107,448 pages are crawled, and the dumped website data occupy about
 88 GiB totally.
 
 Several task definition templates are designed according to the functions of
-WikiHow app and 5,522 task definitions are instantiated through the template
-toolkit in Mobile-Env. This task set is named the *extended set*
-(`wikihow-extended.tar.xz`). There may be several faults that may make the
-system or the task fail in the auto-generated tasks. Therefore, 178 tasks are
-sampled from the extended set and have been verified by human beings to ensure
-correctness and stability, which is named the *canonical set*
-(`wikihow-canonical.tar.xz`). Owing to the limit of the budgets, only 70 tasks
-are tested using the proposed LLM-based agent in the corresponding pager.
-These 70 tasks are given in `wikihow-microcanon.tar.xz`. We call it the
-*canonical subset* or the *micro canonical set*.
+WikiHow app and task definitions are instantiated through the template toolkit
+in Mobile-Env. 577 tasks are sampled from the extended set, which is named the
+*canonical set* (`wikihow-canonical.tar.xz`). Owing to the limit of the
+budgets, only 150 tasks are tested using the proposed LLM-based agent.  These
+150 tasks are given in `wikihow-microcanon.tar.xz`.  We call it the *canonical
+subset* or the *micro canonical set*.
 
 ### Website Data Replay
 

instruction_rewriting/annotations/merged_doccano.json

@@ -2310,5 +2310,87 @@
 			"Open the article \"Buy Stocks (for Beginners)\".",
 			"The target article is \"Buy Stocks (for Beginners)\"."
 		]
+	},
+	{
+		"id": 1302,
+		"text": "Read the article and summarize the needed ingredients.",
+		"task": "constant",
+		"index": "0",
+		"label": [
+			"Read the article and give a summary of the needed ingredients.",
+			"Give a summary of the needed ingredients according to the article.",
+			"After reading the article, list the necessary ingredients.",
+			"Read through the article and encapsulate the needed ingredients.",
+			"Upon reading the article, note down the essential ingredients."
+		]
+	},
+	{
+		"id": 1303,
+		"text": "Read the article and summarize the needed items for the task.",
+		"task": "constant",
+		"index": "0",
+		"label": [
+			"Read the article and give a summary of the needed items.",
+			"Give a summary of the needed items according to the article.",
+			"After reading the article, list all the matierials required for this job.",
+			"Read through the article and encapsulate the needed items.",
+			"Upon reading the article, note down the necessary matierials."
+		]
+	},
+	{
+		"id": 1304,
+		"text": "Read the article and give a summary of each steps.",
+		"task": "constant",
+		"index": "0",
+		"label": [
+			"Read the article and summarize each steps.",
+			"Give a summary of each steps according to the article.",
+			"After reading the article, summarize each steps.",
+			"Read through the article and provide a concise overview of every step.",
+			"Upon reading the article, summarize the steps one by one.",
+			"Following your reading of the article, summarize the steps outlined within it."
+		]
+	},
+	{
+		"id": 1305,
+		"text": "The summary should be given in an unnumbered list.",
+		"task": "constant",
+		"index": "0",
+		"label": [
+			"Please summarize in the form of an unnumbered list.",
+			"It is best to present the summary as an unnumbered list.",
+			"Present the summary in a list format without numbering.",
+			"Depict your summary in the form of a bullet list.",
+			"An unnumbered list should be used to present the summary.",
+			"Create a bullet list to provide the summary.",
+			"Present the summary by listing it without numerical order.",
+			"Use a list format without numbering to present the summary."
+		]
+	},
+	{
+		"id": 1306,
+		"text": "The summary should be given in a numbered list.",
+		"task": "constant",
+		"index": "0",
+		"label": [
+			"Please summarize in the form of a numbered list.",
+			"It is best to present the summary as a numbered list.",
+			"Present the summary in a list format with numbers.",
+			"A numbered list should be used to present the summary.",
+			"Create a numbered list to provide the summary.",
+			"Present the summary by listing it in numerical order.",
+			"Use a list format with numbers to present the summary."
+		]
+	},
+	{
+		"id": 1307,
+		"text": "Read the article and answer the question concisely.",
+		"task": "constant",
+		"index": "0",
+		"label": [
+			"Read the article and offer a concise answer for my question.",
+			"Read the article, then succinctly respond to the question.",
+			"After reading the article, answer the question concisely"
+		]
 	}
 ]

instruction_rewriting/annotations/pattern-question.txt

@@ -0,0 +1,8 @@
+My question is: {<title>...}?
+Answer my question: {<title>...}?
+The question I wonder is: {<title>...}(?/.)
+What I (wonder/want to (know/ask)) is: {<title>...}(?/.)
+I (wonder/want to (know/ask)): {<title>...}(?/.)
+(Tell/Answer) me: {<title>...}?
+(Give/Offer) (me the answer/the answer to me): {<title>...}.
+Provide me with the answer: {<title>...}.

instruction_rewriting/instruction_rewriting_wrapper.py

@@ -22,6 +22,7 @@ class InstructionRewritingWrapper(base_wrapper.BaseWrapper):
                 , article_command_pattern_file: str
                 , categ_pattern_file: str
                 , author_pattern_file: str
+                , question_pattern_file: str
                 , doccano_file: str
                 ):
         #  method __init__ {{{ # 
@@ -35,6 +36,7 @@ class InstructionRewritingWrapper(base_wrapper.BaseWrapper):
                               , article_command_pattern_file
                               , categ_pattern_file
                               , author_pattern_file
+                              , question_pattern_file
                               , doccano_file
                               , rng
                               )

instruction_rewriting/target_handlers.py

@@ -192,19 +192,22 @@ class TransformerSet:
                 , article_command_pattern_file: str
                 , categ_pattern_file: str
                 , author_pattern_file: str
+                , question_pattern_file: str
                 , doccano_file: str
                 , rng: np.random.Generator
                 ):
         #  method __init__ {{{ # 
         """
         Args:
-            search_pattern_file: str
-            article_pattern_file: str
-            article_command_pattern_file: str
-            categ_pattern_file: str
-            author_pattern_file: str
-            doccano_file: str as the path to the json file exported by doccano
-            rng: np.random.Generator
+            search_pattern_file (str): pattern file for search instructions
+            article_pattern_file (str): pattern file for article instructions
+            article_command_pattern_file (str): pattern file for article
+              instructions in "command" field
+            categ_pattern_file (str): pattern file for category instructions
+            author_pattern_file (str): pattern file for author instructions
+            question_pattern_file (str): pattern file for WikiHowNFQA questions
+            doccano_file (str): the path to the json file exported by doccano
+            rng (np.random.Generator): used to generate random indices
         """
 
         self._search_template_lib: List[sentence_pattern.Item]
@@ -212,6 +215,7 @@ class TransformerSet:
         self._article_command_template_lib: List[sentence_pattern.Item]
         self._categ_template_lib: List[sentence_pattern.Item]
         self._author_template_lib: List[sentence_pattern.Item]
+        self._question_template_lib: List[sentence_pattern.Item]
 
         self._search_template_regex: List[Pattern[str]]
         #self._article_template_regex: List[Pattern[str]]
@@ -225,6 +229,7 @@ class TransformerSet:
         self._article_command_template_lib, _ = parse_file(article_command_pattern_file)
         self._categ_template_lib, _ = parse_file(categ_pattern_file)
         self._author_template_lib, _ = parse_file(author_pattern_file)
+        self._question_template_lib, _ = parse_file(question_pattern_file)
 
         self._annotation_dict: Dict[str, List[str]] = {}
         with open(doccano_file) as f:
@@ -260,6 +265,8 @@ class TransformerSet:
             transformed = self._transform_categ(sentence)
         elif sentence.startswith("Check the author"):
             transformed = self._transform_author(sentence)
+        elif sentence.startswith("My question is"):
+            transformed: str = self._transform_question(sentence)
 
         elif sentence in self._annotation_dict:
             candidates = self._annotation_dict[sentence]
@@ -299,6 +306,8 @@ class TransformerSet:
         if template is None:
             return sentence
 
+        # if the template seems to match the sentence according to the regex,
+        # but the target isn't consistent, we should ignore it.
         target_pattern = template.get_targets()
         if not match_target(target_pattern[0], target_str):
             return sentence
@@ -338,6 +347,11 @@ class TransformerSet:
            and sentence.endswith(".")
         target_str = sentence[25:-1]
         return self._apply_new_template(target_str, self._author_template_lib)
+    def _transform_question(self, sentence: str) -> str:
+        assert sentence.startswith("My question is: ")\
+           and sentence.endswith("?")
+        target_str: str = sentence[16:-1]
+        return self._apply_new_template(target_str, self._question_template_lib)
 
     def _apply_new_template( self
                            , target_str: str

wikihow-canonical.tar.xz

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e7168327b2aae6e85bf5561fc65b8ad5bdad71791fb8fccdeeaa21c03102a7f
-size 30684
+oid sha256:e00499008ec1043ecdf8d3d33c5fac6ad5faee4ca5a9fc0676011fc065dcdee7
+size 123828

wikihow-extended.tar.xz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:df7650423ada1a94689f0b30d0d11a186da5749dff5167924716bf109c16ef3c
-size 946740

wikihow-microcanon.tar.xz

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:33279811d397b5741b791130cca8e6d5990ae6d46e70dbe201cdc48f24614a55
-size 13312
+oid sha256:0a94694a43657122d19a9acb8e426c28b7de6dc52e255a69562b194d3a0fe806
+size 14484