Spaces:
Runtime error
Runtime error
File size: 7,286 Bytes
1a7b8b2 |
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 |
# coding=utf-8
# Copyright 2023 The Google Research Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Registry of all instructions."""
from lm_eval.tasks.ifeval import instructions
_KEYWORD = "keywords:"
_LANGUAGE = "language:"
_LENGTH = "length_constraints:"
_CONTENT = "detectable_content:"
_FORMAT = "detectable_format:"
_MULTITURN = "multi-turn:"
_COMBINATION = "combination:"
_STARTEND = "startend:"
_CHANGE_CASES = "change_case:"
_PUNCTUATION = "punctuation:"
INSTRUCTION_DICT = {
_KEYWORD + "existence": instructions.KeywordChecker,
_KEYWORD + "frequency": instructions.KeywordFrequencyChecker,
# TODO(jeffreyzhou): make a proper set of sentences to choose from
# _KEYWORD + "key_sentences": instructions.KeySentenceChecker,
_KEYWORD + "forbidden_words": instructions.ForbiddenWords,
_KEYWORD + "letter_frequency": instructions.LetterFrequencyChecker,
_LANGUAGE + "response_language": instructions.ResponseLanguageChecker,
_LENGTH + "number_sentences": instructions.NumberOfSentences,
_LENGTH + "number_paragraphs": instructions.ParagraphChecker,
_LENGTH + "number_words": instructions.NumberOfWords,
_LENGTH + "nth_paragraph_first_word": instructions.ParagraphFirstWordCheck,
_CONTENT + "number_placeholders": instructions.PlaceholderChecker,
_CONTENT + "postscript": instructions.PostscriptChecker,
_FORMAT + "number_bullet_lists": instructions.BulletListChecker,
# TODO(jeffreyzhou): Pre-create paragraph or use prompt to replace
# _CONTENT + "rephrase_paragraph": instructions.RephraseParagraph,
_FORMAT + "constrained_response": instructions.ConstrainedResponseChecker,
_FORMAT + "number_highlighted_sections": (instructions.HighlightSectionChecker),
_FORMAT + "multiple_sections": instructions.SectionChecker,
# TODO(tianjianlu): Re-enable rephrasing with preprocessing the message.
# _FORMAT + "rephrase": instructions.RephraseChecker,
_FORMAT + "json_format": instructions.JsonFormat,
_FORMAT + "title": instructions.TitleChecker,
# TODO(tianjianlu): Re-enable with specific prompts.
# _MULTITURN + "constrained_start": instructions.ConstrainedStartChecker,
_COMBINATION + "two_responses": instructions.TwoResponsesChecker,
_COMBINATION + "repeat_prompt": instructions.RepeatPromptThenAnswer,
_STARTEND + "end_checker": instructions.EndChecker,
_CHANGE_CASES + "capital_word_frequency": instructions.CapitalWordFrequencyChecker,
_CHANGE_CASES + "english_capital": instructions.CapitalLettersEnglishChecker,
_CHANGE_CASES + "english_lowercase": instructions.LowercaseLettersEnglishChecker,
_PUNCTUATION + "no_comma": instructions.CommaChecker,
_STARTEND + "quotation": instructions.QuotationChecker,
}
INSTRUCTION_CONFLICTS = {
_KEYWORD + "existence": {_KEYWORD + "existence"},
_KEYWORD + "frequency": {_KEYWORD + "frequency"},
# TODO(jeffreyzhou): make a proper set of sentences to choose from
# _KEYWORD + "key_sentences": instructions.KeySentenceChecker,
_KEYWORD + "forbidden_words": {_KEYWORD + "forbidden_words"},
_KEYWORD + "letter_frequency": {_KEYWORD + "letter_frequency"},
_LANGUAGE
+ "response_language": {
_LANGUAGE + "response_language",
_FORMAT + "multiple_sections",
_KEYWORD + "existence",
_KEYWORD + "frequency",
_KEYWORD + "forbidden_words",
_STARTEND + "end_checker",
_CHANGE_CASES + "english_capital",
_CHANGE_CASES + "english_lowercase",
},
_LENGTH + "number_sentences": {_LENGTH + "number_sentences"},
_LENGTH
+ "number_paragraphs": {
_LENGTH + "number_paragraphs",
_LENGTH + "nth_paragraph_first_word",
_LENGTH + "number_sentences",
_LENGTH + "nth_paragraph_first_word",
},
_LENGTH + "number_words": {_LENGTH + "number_words"},
_LENGTH
+ "nth_paragraph_first_word": {
_LENGTH + "nth_paragraph_first_word",
_LENGTH + "number_paragraphs",
},
_CONTENT + "number_placeholders": {_CONTENT + "number_placeholders"},
_CONTENT + "postscript": {_CONTENT + "postscript"},
_FORMAT + "number_bullet_lists": {_FORMAT + "number_bullet_lists"},
# TODO(jeffreyzhou): Pre-create paragraph or use prompt to replace
# _CONTENT + "rephrase_paragraph": instructions.RephraseParagraph,
_FORMAT + "constrained_response": set(INSTRUCTION_DICT.keys()),
_FORMAT + "number_highlighted_sections": {_FORMAT + "number_highlighted_sections"},
_FORMAT
+ "multiple_sections": {
_FORMAT + "multiple_sections",
_LANGUAGE + "response_language",
_FORMAT + "number_highlighted_sections",
},
# TODO(tianjianlu): Re-enable rephrasing with preprocessing the message.
# _FORMAT + "rephrase": instructions.RephraseChecker,
_FORMAT
+ "json_format": set(INSTRUCTION_DICT.keys()).difference(
{_KEYWORD + "forbidden_words", _KEYWORD + "existence"}
),
_FORMAT + "title": {_FORMAT + "title"},
# TODO(tianjianlu): Re-enable with specific prompts.
# _MULTITURN + "constrained_start": instructions.ConstrainedStartChecker,
_COMBINATION
+ "two_responses": set(INSTRUCTION_DICT.keys()).difference(
{
_KEYWORD + "forbidden_words",
_KEYWORD + "existence",
_LANGUAGE + "response_language",
_FORMAT + "title",
_PUNCTUATION + "no_comma",
}
),
_COMBINATION
+ "repeat_prompt": set(INSTRUCTION_DICT.keys()).difference(
{_KEYWORD + "existence", _FORMAT + "title", _PUNCTUATION + "no_comma"}
),
_STARTEND + "end_checker": {_STARTEND + "end_checker"},
_CHANGE_CASES
+ "capital_word_frequency": {
_CHANGE_CASES + "capital_word_frequency",
_CHANGE_CASES + "english_lowercase",
_CHANGE_CASES + "english_capital",
},
_CHANGE_CASES + "english_capital": {_CHANGE_CASES + "english_capital"},
_CHANGE_CASES
+ "english_lowercase": {
_CHANGE_CASES + "english_lowercase",
_CHANGE_CASES + "english_capital",
},
_PUNCTUATION + "no_comma": {_PUNCTUATION + "no_comma"},
_STARTEND + "quotation": {_STARTEND + "quotation", _FORMAT + "title"},
}
def conflict_make(conflicts):
"""Makes sure if A conflicts with B, B will conflict with A.
Args:
conflicts: Dictionary of potential conflicts where key is instruction id
and value is set of instruction ids that it conflicts with.
Returns:
Revised version of the dictionary. All instructions conflict with
themselves. If A conflicts with B, B will conflict with A.
"""
for key in conflicts:
for k in conflicts[key]:
conflicts[k].add(key)
conflicts[key].add(key)
return conflicts
|