edsl 0.1.33.dev2__py3-none-any.whl → 0.1.33.dev3__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- edsl/Base.py +9 -3
- edsl/__init__.py +1 -0
- edsl/__version__.py +1 -1
- edsl/agents/Agent.py +6 -6
- edsl/agents/Invigilator.py +6 -3
- edsl/agents/InvigilatorBase.py +8 -27
- edsl/agents/{PromptConstructionMixin.py → PromptConstructor.py} +101 -29
- edsl/config.py +26 -34
- edsl/coop/coop.py +11 -2
- edsl/data_transfer_models.py +27 -73
- edsl/enums.py +2 -0
- edsl/inference_services/GoogleService.py +1 -1
- edsl/inference_services/InferenceServiceABC.py +44 -13
- edsl/inference_services/OpenAIService.py +7 -4
- edsl/inference_services/TestService.py +24 -15
- edsl/inference_services/TogetherAIService.py +170 -0
- edsl/inference_services/registry.py +2 -0
- edsl/jobs/Jobs.py +18 -8
- edsl/jobs/buckets/BucketCollection.py +24 -15
- edsl/jobs/buckets/TokenBucket.py +64 -10
- edsl/jobs/interviews/Interview.py +115 -47
- edsl/jobs/interviews/{interview_exception_tracking.py → InterviewExceptionCollection.py} +16 -0
- edsl/jobs/interviews/InterviewExceptionEntry.py +2 -0
- edsl/jobs/runners/JobsRunnerAsyncio.py +86 -161
- edsl/jobs/runners/JobsRunnerStatus.py +331 -0
- edsl/jobs/tasks/TaskHistory.py +17 -0
- edsl/language_models/LanguageModel.py +26 -31
- edsl/language_models/registry.py +13 -9
- edsl/questions/QuestionBase.py +64 -16
- edsl/questions/QuestionBudget.py +93 -41
- edsl/questions/QuestionFreeText.py +6 -0
- edsl/questions/QuestionMultipleChoice.py +11 -26
- edsl/questions/QuestionNumerical.py +5 -4
- edsl/questions/Quick.py +41 -0
- edsl/questions/ResponseValidatorABC.py +6 -5
- edsl/questions/derived/QuestionLinearScale.py +4 -1
- edsl/questions/derived/QuestionTopK.py +4 -1
- edsl/questions/derived/QuestionYesNo.py +8 -2
- edsl/questions/templates/budget/__init__.py +0 -0
- edsl/questions/templates/budget/answering_instructions.jinja +7 -0
- edsl/questions/templates/budget/question_presentation.jinja +7 -0
- edsl/questions/templates/extract/__init__.py +0 -0
- edsl/questions/templates/rank/__init__.py +0 -0
- edsl/results/DatasetExportMixin.py +5 -1
- edsl/results/Result.py +1 -1
- edsl/results/Results.py +4 -1
- edsl/scenarios/FileStore.py +71 -10
- edsl/scenarios/Scenario.py +86 -21
- edsl/scenarios/ScenarioImageMixin.py +2 -2
- edsl/scenarios/ScenarioList.py +13 -0
- edsl/scenarios/ScenarioListPdfMixin.py +150 -4
- edsl/study/Study.py +32 -0
- edsl/surveys/Rule.py +10 -1
- edsl/surveys/RuleCollection.py +19 -3
- edsl/surveys/Survey.py +7 -0
- edsl/templates/error_reporting/interview_details.html +6 -1
- edsl/utilities/utilities.py +9 -1
- {edsl-0.1.33.dev2.dist-info → edsl-0.1.33.dev3.dist-info}/METADATA +2 -1
- {edsl-0.1.33.dev2.dist-info → edsl-0.1.33.dev3.dist-info}/RECORD +61 -55
- edsl/jobs/interviews/retry_management.py +0 -39
- edsl/jobs/runners/JobsRunnerStatusMixin.py +0 -333
- {edsl-0.1.33.dev2.dist-info → edsl-0.1.33.dev3.dist-info}/LICENSE +0 -0
- {edsl-0.1.33.dev2.dist-info → edsl-0.1.33.dev3.dist-info}/WHEEL +0 -0
edsl/questions/QuestionBudget.py
CHANGED
@@ -1,8 +1,60 @@
|
|
1
1
|
from __future__ import annotations
|
2
|
-
import
|
3
|
-
|
2
|
+
from typing import Any, Optional, Union, List
|
3
|
+
|
4
|
+
from pydantic import Field, BaseModel, validator
|
5
|
+
|
4
6
|
from edsl.questions.QuestionBase import QuestionBase
|
5
7
|
from edsl.questions.descriptors import IntegerDescriptor, QuestionOptionsDescriptor
|
8
|
+
from edsl.questions.ResponseValidatorABC import ResponseValidatorABC
|
9
|
+
|
10
|
+
|
11
|
+
class BudgewResponseValidator(ResponseValidatorABC):
|
12
|
+
valid_examples = []
|
13
|
+
|
14
|
+
invalid_examples = []
|
15
|
+
|
16
|
+
def fix(self, response, verbose=False):
|
17
|
+
if verbose:
|
18
|
+
print(f"Fixing list response: {response}")
|
19
|
+
answer = str(response.get("answer") or response.get("generated_tokens", ""))
|
20
|
+
if len(answer.split(",")) > 0:
|
21
|
+
return (
|
22
|
+
{"answer": answer.split(",")} | {"comment": response.get("comment")}
|
23
|
+
if "comment" in response
|
24
|
+
else {}
|
25
|
+
)
|
26
|
+
|
27
|
+
|
28
|
+
def create_budget_model(
|
29
|
+
budget_sum: float, permissive: bool, question_options: List[str]
|
30
|
+
):
|
31
|
+
class BudgetResponse(BaseModel):
|
32
|
+
answer: List[float] = Field(
|
33
|
+
...,
|
34
|
+
description="List of non-negative numbers representing budget allocation",
|
35
|
+
min_items=len(question_options),
|
36
|
+
max_items=len(question_options),
|
37
|
+
)
|
38
|
+
comment: Optional[str] = None
|
39
|
+
generated_tokens: Optional[str] = None
|
40
|
+
|
41
|
+
@validator("answer")
|
42
|
+
def validate_answer(cls, v):
|
43
|
+
if len(v) != len(question_options):
|
44
|
+
raise ValueError(f"Must provide {len(question_options)} values")
|
45
|
+
if any(x < 0 for x in v):
|
46
|
+
raise ValueError("All values must be non-negative")
|
47
|
+
total = sum(v)
|
48
|
+
if not permissive and total != budget_sum:
|
49
|
+
raise ValueError(f"Sum of numbers must equal {budget_sum}")
|
50
|
+
elif permissive and total > budget_sum:
|
51
|
+
raise ValueError(f"Sum of numbers cannot exceed {budget_sum}")
|
52
|
+
return v
|
53
|
+
|
54
|
+
class Config:
|
55
|
+
extra = "forbid"
|
56
|
+
|
57
|
+
return BudgetResponse
|
6
58
|
|
7
59
|
|
8
60
|
class QuestionBudget(QuestionBase):
|
@@ -12,7 +64,7 @@ class QuestionBudget(QuestionBase):
|
|
12
64
|
budget_sum: int = IntegerDescriptor(none_allowed=False)
|
13
65
|
question_options: list[str] = QuestionOptionsDescriptor(q_budget=True)
|
14
66
|
_response_model = None
|
15
|
-
response_validator_class =
|
67
|
+
response_validator_class = BudgewResponseValidator
|
16
68
|
|
17
69
|
def __init__(
|
18
70
|
self,
|
@@ -20,8 +72,10 @@ class QuestionBudget(QuestionBase):
|
|
20
72
|
question_text: str,
|
21
73
|
question_options: list[str],
|
22
74
|
budget_sum: int,
|
75
|
+
include_comment: bool = True,
|
23
76
|
question_presentation: Optional[str] = None,
|
24
77
|
answering_instructions: Optional[str] = None,
|
78
|
+
permissive: bool = False,
|
25
79
|
):
|
26
80
|
"""Instantiate a new QuestionBudget.
|
27
81
|
|
@@ -36,20 +90,17 @@ class QuestionBudget(QuestionBase):
|
|
36
90
|
self.budget_sum = budget_sum
|
37
91
|
self.question_presentation = question_presentation
|
38
92
|
self.answering_instructions = answering_instructions
|
93
|
+
self.permissive = permissive
|
94
|
+
self.include_comment = include_comment
|
39
95
|
|
40
|
-
|
41
|
-
|
42
|
-
|
43
|
-
|
44
|
-
"""Validate the answer."""
|
45
|
-
self._validate_answer_template_basic(answer)
|
46
|
-
self._validate_answer_key_value(answer, "answer", dict)
|
47
|
-
self._validate_answer_budget(answer)
|
48
|
-
return answer
|
96
|
+
def create_response_model(self):
|
97
|
+
return create_budget_model(
|
98
|
+
self.budget_sum, self.permissive, self.question_options
|
99
|
+
)
|
49
100
|
|
50
101
|
def _translate_answer_code_to_answer(
|
51
|
-
self,
|
52
|
-
):
|
102
|
+
self, answer_code, combined_dict
|
103
|
+
) -> list[dict]:
|
53
104
|
"""
|
54
105
|
Translate the answer codes to the actual answers.
|
55
106
|
|
@@ -58,35 +109,35 @@ class QuestionBudget(QuestionBase):
|
|
58
109
|
This code will translate that to "a".
|
59
110
|
"""
|
60
111
|
translated_codes = []
|
61
|
-
for answer_code,
|
62
|
-
translated_codes.append({
|
112
|
+
for answer_code, question_option in zip(answer_code, self.question_options):
|
113
|
+
translated_codes.append({question_option: answer_code})
|
63
114
|
|
64
115
|
return translated_codes
|
65
116
|
|
66
|
-
def _simulate_answer(self, human_readable=True):
|
67
|
-
|
68
|
-
|
69
|
-
|
70
|
-
|
71
|
-
|
72
|
-
|
73
|
-
|
74
|
-
|
75
|
-
|
76
|
-
|
77
|
-
|
78
|
-
|
79
|
-
|
80
|
-
|
81
|
-
|
82
|
-
|
83
|
-
|
84
|
-
|
85
|
-
|
86
|
-
|
87
|
-
|
88
|
-
|
89
|
-
|
117
|
+
# def _simulate_answer(self, human_readable=True):
|
118
|
+
# """Simulate a valid answer for debugging purposes (what the validator expects)."""
|
119
|
+
# from edsl.utilities.utilities import random_string
|
120
|
+
|
121
|
+
# if human_readable:
|
122
|
+
# keys = self.question_options
|
123
|
+
# else:
|
124
|
+
# keys = range(len(self.question_options))
|
125
|
+
# remaining_budget = self.budget_sum
|
126
|
+
# values = []
|
127
|
+
# for _ in range(len(self.question_options)):
|
128
|
+
# if _ == len(self.question_options) - 1:
|
129
|
+
# # Assign remaining budget to the last value
|
130
|
+
# values.append(remaining_budget)
|
131
|
+
# else:
|
132
|
+
# # Generate a random value between 0 and remaining budget
|
133
|
+
# value = random.randint(0, remaining_budget)
|
134
|
+
# values.append(value)
|
135
|
+
# remaining_budget -= value
|
136
|
+
# answer = dict(zip(keys, values))
|
137
|
+
# return {
|
138
|
+
# "answer": answer,
|
139
|
+
# "comment": random_string(),
|
140
|
+
# }
|
90
141
|
|
91
142
|
@property
|
92
143
|
def question_html_content(self) -> str:
|
@@ -133,13 +184,14 @@ class QuestionBudget(QuestionBase):
|
|
133
184
|
# Helpful methods
|
134
185
|
################
|
135
186
|
@classmethod
|
136
|
-
def example(cls) -> QuestionBudget:
|
187
|
+
def example(cls, include_comment: bool = True) -> QuestionBudget:
|
137
188
|
"""Return an example of a budget question."""
|
138
189
|
return cls(
|
139
190
|
question_name="food_budget",
|
140
191
|
question_text="How would you allocate $100?",
|
141
192
|
question_options=["Pizza", "Ice Cream", "Burgers", "Salad"],
|
142
193
|
budget_sum=100,
|
194
|
+
include_comment=include_comment,
|
143
195
|
)
|
144
196
|
|
145
197
|
|
@@ -36,6 +36,12 @@ class FreeTextResponseValidator(ResponseValidatorABC):
|
|
36
36
|
),
|
37
37
|
]
|
38
38
|
|
39
|
+
def fix(self, response, verbose=False):
|
40
|
+
return {
|
41
|
+
"answer": str(response.get("generated_tokens")),
|
42
|
+
"generated_tokens": str(response.get("generated_tokens")),
|
43
|
+
}
|
44
|
+
|
39
45
|
|
40
46
|
class QuestionFreeText(QuestionBase):
|
41
47
|
"""This question prompts the agent to respond with free text."""
|
@@ -1,22 +1,14 @@
|
|
1
1
|
from __future__ import annotations
|
2
|
-
from typing import Union, Literal, Optional
|
3
|
-
from jinja2 import Template
|
2
|
+
from typing import Union, Literal, Optional, List, Any
|
4
3
|
|
4
|
+
from jinja2 import Template
|
5
5
|
from pydantic import BaseModel, Field
|
6
|
-
from typing import Optional, Literal
|
7
6
|
|
7
|
+
from edsl.scenarios.Scenario import Scenario
|
8
8
|
from edsl.questions.QuestionBase import QuestionBase
|
9
9
|
from edsl.questions.descriptors import QuestionOptionsDescriptor
|
10
10
|
from edsl.questions.decorators import inject_exception
|
11
|
-
|
12
11
|
from edsl.questions.ResponseValidatorABC import ResponseValidatorABC
|
13
|
-
from edsl.questions.ResponseValidatorABC import BaseResponse
|
14
|
-
|
15
|
-
from edsl.exceptions import QuestionAnswerValidationError
|
16
|
-
|
17
|
-
from pydantic import BaseModel, Field, create_model
|
18
|
-
|
19
|
-
from typing import List, Any, Literal
|
20
12
|
|
21
13
|
|
22
14
|
def create_response_model(choices: List[str], permissive: bool = False):
|
@@ -27,7 +19,6 @@ def create_response_model(choices: List[str], permissive: bool = False):
|
|
27
19
|
:param permissive: If True, any value will be accepted as an answer.
|
28
20
|
:return: A new Pydantic model class.
|
29
21
|
"""
|
30
|
-
# Convert the choices list to a tuple for use with Literal
|
31
22
|
choice_tuple = tuple(choices)
|
32
23
|
|
33
24
|
if not permissive:
|
@@ -66,16 +57,6 @@ def create_response_model(choices: List[str], permissive: bool = False):
|
|
66
57
|
return ChoiceResponse
|
67
58
|
|
68
59
|
|
69
|
-
def fix_multiple_choice(response, question_options, use_code, verbose=False):
|
70
|
-
"""Fix the response to a multiple choice question.
|
71
|
-
Respnse is a dictionary with keys:
|
72
|
-
- answer: the answer code
|
73
|
-
- generated_tokens: the generated tokens
|
74
|
-
- comment: the comment
|
75
|
-
"""
|
76
|
-
pass
|
77
|
-
|
78
|
-
|
79
60
|
class MultipleChoiceResponseValidator(ResponseValidatorABC):
|
80
61
|
required_params = ["question_options", "use_code"]
|
81
62
|
|
@@ -139,9 +120,9 @@ class QuestionMultipleChoice(QuestionBase):
|
|
139
120
|
|
140
121
|
question_type = "multiple_choice"
|
141
122
|
purpose = "When options are known and limited"
|
142
|
-
question_options: Union[
|
143
|
-
|
144
|
-
|
123
|
+
question_options: Union[list[str], list[list], list[float], list[int]] = (
|
124
|
+
QuestionOptionsDescriptor()
|
125
|
+
)
|
145
126
|
_response_model = None
|
146
127
|
response_validator_class = MultipleChoiceResponseValidator
|
147
128
|
|
@@ -161,6 +142,11 @@ class QuestionMultipleChoice(QuestionBase):
|
|
161
142
|
:param question_name: The name of the question.
|
162
143
|
:param question_text: The text of the question.
|
163
144
|
:param question_options: The options the agent should select from.
|
145
|
+
:param include_comment: Whether to include a comment field.
|
146
|
+
:param use_code: Whether to use code for the options.
|
147
|
+
:param answering_instructions: Instructions for the question.
|
148
|
+
:param question_presentation: The presentation of the question.
|
149
|
+
:param permissive: Whether to force the answer to be one of the options.
|
164
150
|
|
165
151
|
"""
|
166
152
|
self.question_name = question_name
|
@@ -202,7 +188,6 @@ class QuestionMultipleChoice(QuestionBase):
|
|
202
188
|
'Happy'
|
203
189
|
|
204
190
|
"""
|
205
|
-
from edsl.scenarios.Scenario import Scenario
|
206
191
|
|
207
192
|
scenario = scenario or Scenario()
|
208
193
|
|
@@ -1,5 +1,6 @@
|
|
1
1
|
from __future__ import annotations
|
2
|
-
|
2
|
+
|
3
|
+
# from decimal import Decimal
|
3
4
|
from random import uniform
|
4
5
|
from typing import Any, Optional, Union, Literal
|
5
6
|
|
@@ -14,8 +15,8 @@ from edsl.exceptions.questions import QuestionAnswerValidationError
|
|
14
15
|
|
15
16
|
|
16
17
|
def create_numeric_response(
|
17
|
-
min_value: Optional[
|
18
|
-
max_value: Optional[
|
18
|
+
min_value: Optional[float] = None,
|
19
|
+
max_value: Optional[float] = None,
|
19
20
|
permissive=False,
|
20
21
|
):
|
21
22
|
field_kwargs = {}
|
@@ -27,7 +28,7 @@ def create_numeric_response(
|
|
27
28
|
field_kwargs["le"] = max_value
|
28
29
|
|
29
30
|
class ConstrainedNumericResponse(BaseModel):
|
30
|
-
answer: Union[
|
31
|
+
answer: Union[int, float] = Field(**field_kwargs)
|
31
32
|
comment: Optional[str] = Field(None)
|
32
33
|
generated_tokens: Optional[Any] = Field(None)
|
33
34
|
|
edsl/questions/Quick.py
ADDED
@@ -0,0 +1,41 @@
|
|
1
|
+
from edsl import (
|
2
|
+
QuestionFreeText,
|
3
|
+
QuestionMultipleChoice,
|
4
|
+
Survey,
|
5
|
+
QuestionList,
|
6
|
+
Question,
|
7
|
+
)
|
8
|
+
|
9
|
+
|
10
|
+
def Quick(question_text):
|
11
|
+
q_type = QuestionMultipleChoice(
|
12
|
+
question_text=f"A researcher is asking a language model this: {question_text}. What is the most appropriate type of question to ask?",
|
13
|
+
question_name="potential_question_type",
|
14
|
+
question_options=["multiple_choice", "list", "free_text"],
|
15
|
+
)
|
16
|
+
|
17
|
+
q_name = QuestionFreeText(
|
18
|
+
question_text=f"A researcher is asking a language model this: {question_text}. What is a good name for this question that's a valid python identifier? Just return the proposed identifer",
|
19
|
+
question_name="potential_question_name",
|
20
|
+
)
|
21
|
+
|
22
|
+
q_options = QuestionList(
|
23
|
+
question_text=f"A research is asking this question: { question_text }. What are the possible options for this question?",
|
24
|
+
question_name="potential_question_options",
|
25
|
+
)
|
26
|
+
|
27
|
+
survey = Survey([q_type, q_name, q_options]).add_skip_rule(
|
28
|
+
q_options, "{{ potential_question_type }} != 'multiple_choice'"
|
29
|
+
)
|
30
|
+
return survey
|
31
|
+
# results = survey.run()
|
32
|
+
# question_type = results.select("potential_question_type").first()
|
33
|
+
# question_options = results.select("potential_question_options").first()
|
34
|
+
# question_name = results.select("potential_question_name").first()
|
35
|
+
# print("Question Type: ", question_type)
|
36
|
+
# print("Question Name: ", question_name)
|
37
|
+
# print("Question Options: ", question_options)
|
38
|
+
# if question_options == None:
|
39
|
+
# return Question(question_type, question_name = question_name)
|
40
|
+
# else:
|
41
|
+
# return Question(question_type, question_name = question_name, question_options = question_options)
|
@@ -1,6 +1,7 @@
|
|
1
1
|
from abc import ABC, abstractmethod
|
2
2
|
from pydantic import BaseModel, Field, field_validator
|
3
|
-
|
3
|
+
|
4
|
+
# from decimal import Decimal
|
4
5
|
from typing import Optional, Any, List, TypedDict
|
5
6
|
|
6
7
|
from edsl.exceptions import QuestionAnswerValidationError
|
@@ -64,7 +65,7 @@ class ResponseValidatorABC(ABC):
|
|
64
65
|
>>> rv = ResponseValidatorABC.example()
|
65
66
|
>>> rv.override_answer = {"answer": 42}
|
66
67
|
>>> rv.validate({"answer": 23})
|
67
|
-
{'answer':
|
68
|
+
{'answer': 42, 'comment': None, 'generated_tokens': None}
|
68
69
|
"""
|
69
70
|
if self.exception_to_throw:
|
70
71
|
raise self.exception_to_throw
|
@@ -75,7 +76,7 @@ class ResponseValidatorABC(ABC):
|
|
75
76
|
|
76
77
|
>>> rv = ResponseValidatorABC.example("numerical")
|
77
78
|
>>> rv._base_validate({"answer": 42})
|
78
|
-
ConstrainedNumericResponse(answer=
|
79
|
+
ConstrainedNumericResponse(answer=42, comment=None, generated_tokens=None)
|
79
80
|
"""
|
80
81
|
try:
|
81
82
|
return self.response_model(**data)
|
@@ -97,7 +98,7 @@ class ResponseValidatorABC(ABC):
|
|
97
98
|
|
98
99
|
>>> rv = ResponseValidatorABC.example("numerical")
|
99
100
|
>>> rv.validate({"answer": 42})
|
100
|
-
{'answer':
|
101
|
+
{'answer': 42, 'comment': None, 'generated_tokens': None}
|
101
102
|
>>> rv.max_value
|
102
103
|
86.7
|
103
104
|
>>> rv.validate({"answer": "120"})
|
@@ -109,7 +110,7 @@ class ResponseValidatorABC(ABC):
|
|
109
110
|
>>> q.permissive = True
|
110
111
|
>>> rv = q.response_validator
|
111
112
|
>>> rv.validate({"answer": "120"})
|
112
|
-
{'answer':
|
113
|
+
{'answer': 120, 'comment': None, 'generated_tokens': None}
|
113
114
|
>>> rv.validate({"answer": "poo"})
|
114
115
|
Traceback (most recent call last):
|
115
116
|
...
|
@@ -22,6 +22,7 @@ class QuestionLinearScale(QuestionMultipleChoice):
|
|
22
22
|
option_labels: Optional[dict[int, str]] = None,
|
23
23
|
answering_instructions: Optional[str] = None,
|
24
24
|
question_presentation: Optional[str] = None,
|
25
|
+
include_comment: Optional[bool] = True,
|
25
26
|
):
|
26
27
|
"""Instantiate a new QuestionLinearScale.
|
27
28
|
|
@@ -36,6 +37,7 @@ class QuestionLinearScale(QuestionMultipleChoice):
|
|
36
37
|
question_text=question_text,
|
37
38
|
question_options=question_options,
|
38
39
|
use_code=False, # question linear scale will have it's own code
|
40
|
+
include_comment=include_comment,
|
39
41
|
)
|
40
42
|
self.question_options = question_options
|
41
43
|
self.option_labels = (
|
@@ -49,13 +51,14 @@ class QuestionLinearScale(QuestionMultipleChoice):
|
|
49
51
|
################
|
50
52
|
@classmethod
|
51
53
|
@inject_exception
|
52
|
-
def example(cls) -> QuestionLinearScale:
|
54
|
+
def example(cls, include_comment: bool = True) -> QuestionLinearScale:
|
53
55
|
"""Return an example of a linear scale question."""
|
54
56
|
return cls(
|
55
57
|
question_text="How much do you like ice cream?",
|
56
58
|
question_options=[1, 2, 3, 4, 5],
|
57
59
|
question_name="ice_cream",
|
58
60
|
option_labels={1: "I hate it", 5: "I love it"},
|
61
|
+
include_comment=include_comment,
|
59
62
|
)
|
60
63
|
|
61
64
|
|
@@ -20,6 +20,7 @@ class QuestionTopK(QuestionCheckBox):
|
|
20
20
|
max_selections: int,
|
21
21
|
question_presentation: Optional[str] = None,
|
22
22
|
answering_instructions: Optional[str] = None,
|
23
|
+
include_comment: Optional[bool] = True,
|
23
24
|
):
|
24
25
|
"""Initialize the question.
|
25
26
|
|
@@ -37,6 +38,7 @@ class QuestionTopK(QuestionCheckBox):
|
|
37
38
|
max_selections=max_selections,
|
38
39
|
question_presentation=question_presentation,
|
39
40
|
answering_instructions=answering_instructions,
|
41
|
+
include_comment=include_comment,
|
40
42
|
)
|
41
43
|
if min_selections != max_selections:
|
42
44
|
raise QuestionCreationValidationError(
|
@@ -52,7 +54,7 @@ class QuestionTopK(QuestionCheckBox):
|
|
52
54
|
################
|
53
55
|
@classmethod
|
54
56
|
@inject_exception
|
55
|
-
def example(cls) -> QuestionTopK:
|
57
|
+
def example(cls, include_comment: bool = True) -> QuestionTopK:
|
56
58
|
"""Return an example question."""
|
57
59
|
return cls(
|
58
60
|
question_name="two_fruits",
|
@@ -60,6 +62,7 @@ class QuestionTopK(QuestionCheckBox):
|
|
60
62
|
question_options=["apple", "banana", "carrot", "durian"],
|
61
63
|
min_selections=2,
|
62
64
|
max_selections=2,
|
65
|
+
include_comment=include_comment,
|
63
66
|
)
|
64
67
|
|
65
68
|
|
@@ -19,6 +19,7 @@ class QuestionYesNo(QuestionMultipleChoice):
|
|
19
19
|
question_options: list[str] = ["No", "Yes"],
|
20
20
|
answering_instructions: Optional[str] = None,
|
21
21
|
question_presentation: Optional[str] = None,
|
22
|
+
include_comment: Optional[bool] = True,
|
22
23
|
):
|
23
24
|
"""Instantiate a new QuestionYesNo.
|
24
25
|
|
@@ -33,6 +34,7 @@ class QuestionYesNo(QuestionMultipleChoice):
|
|
33
34
|
use_code=False,
|
34
35
|
answering_instructions=answering_instructions,
|
35
36
|
question_presentation=question_presentation,
|
37
|
+
include_comment=include_comment,
|
36
38
|
)
|
37
39
|
self.question_options = question_options
|
38
40
|
|
@@ -41,9 +43,13 @@ class QuestionYesNo(QuestionMultipleChoice):
|
|
41
43
|
################
|
42
44
|
@classmethod
|
43
45
|
@inject_exception
|
44
|
-
def example(cls) -> QuestionYesNo:
|
46
|
+
def example(cls, include_comment: bool = True) -> QuestionYesNo:
|
45
47
|
"""Return an example of a yes/no question."""
|
46
|
-
return cls(
|
48
|
+
return cls(
|
49
|
+
question_name="is_it_equal",
|
50
|
+
question_text="Is 5 + 5 equal to 11?",
|
51
|
+
include_comment=include_comment,
|
52
|
+
)
|
47
53
|
|
48
54
|
|
49
55
|
def main():
|
File without changes
|
@@ -0,0 +1,7 @@
|
|
1
|
+
Return only a comma-separated list the values in the same order as the options, with 0s included, on one line, in square braces.
|
2
|
+
|
3
|
+
Example: if there are 4 options, the response should be "[25,25,25,25]" to allocate 25 to each option.
|
4
|
+
|
5
|
+
{% if include_comment %}
|
6
|
+
After the answer, you can put a comment explaining your choice on the next line.
|
7
|
+
{% endif %}
|
File without changes
|
File without changes
|
@@ -472,7 +472,11 @@ class DatasetExportMixin:
|
|
472
472
|
from edsl import ScenarioList, Scenario
|
473
473
|
|
474
474
|
list_of_dicts = self.to_dicts(remove_prefix=remove_prefix)
|
475
|
-
|
475
|
+
scenarios = []
|
476
|
+
for d in list_of_dicts:
|
477
|
+
scenarios.append(Scenario(d))
|
478
|
+
return ScenarioList(scenarios)
|
479
|
+
# return ScenarioList([Scenario(d) for d in list_of_dicts])
|
476
480
|
|
477
481
|
def to_agent_list(self, remove_prefix: bool = True):
|
478
482
|
"""Convert the results to a list of dictionaries, one per agent.
|
edsl/results/Result.py
CHANGED
@@ -367,7 +367,7 @@ class Result(Base, UserDict):
|
|
367
367
|
"raw_model_response", {"raw_model_response": "No raw model response"}
|
368
368
|
),
|
369
369
|
question_to_attributes=json_dict.get("question_to_attributes", None),
|
370
|
-
generated_tokens=json_dict.get("generated_tokens",
|
370
|
+
generated_tokens=json_dict.get("generated_tokens", {}),
|
371
371
|
)
|
372
372
|
return result
|
373
373
|
|
edsl/results/Results.py
CHANGED
@@ -245,7 +245,9 @@ class Results(UserList, Mixins, Base):
|
|
245
245
|
)
|
246
246
|
|
247
247
|
def __repr__(self) -> str:
|
248
|
-
|
248
|
+
import reprlib
|
249
|
+
|
250
|
+
return f"Results(data = {reprlib.repr(self.data)}, survey = {repr(self.survey)}, created_columns = {self.created_columns})"
|
249
251
|
|
250
252
|
def _repr_html_(self) -> str:
|
251
253
|
from IPython.display import HTML
|
@@ -1089,6 +1091,7 @@ class Results(UserList, Mixins, Base):
|
|
1089
1091
|
stop_on_exception=True,
|
1090
1092
|
skip_retry=True,
|
1091
1093
|
raise_validation_errors=True,
|
1094
|
+
disable_remote_inference=True,
|
1092
1095
|
)
|
1093
1096
|
return results
|
1094
1097
|
|