Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
gaoqiong
lm-evaluation-harness
Commits
bbb8386c
Commit
bbb8386c
authored
Apr 16, 2024
by
lintangsutawika
Browse files
removed alt worlds prompts
parent
3e5e9da2
Changes
1000
Hide whitespace changes
Inline
Side-by-side
Showing
20 changed files
with
0 additions
and
240 deletions
+0
-240
lm_eval/tasks/hellaswag/alternative_worlds/style_03/c.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_03/c.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_04/a.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_04/a.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_04/b.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_04/b.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_04/c.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_04/c.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_05/a.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_05/a.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_05/b.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_05/b.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_05/c.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_05/c.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_06/a.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_06/a.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_06/b.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_06/b.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_06/c.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_06/c.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_07/a.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_07/a.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_07/b.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_07/b.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_07/c.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_07/c.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_08/a.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_08/a.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_08/b.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_08/b.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/style_08/c.yaml
lm_eval/tasks/hellaswag/alternative_worlds/style_08/c.yaml
+0
-7
lm_eval/tasks/hellaswag/alternative_worlds/styles.py
lm_eval/tasks/hellaswag/alternative_worlds/styles.py
+0
-90
lm_eval/tasks/mathqa/alternative_worlds/output_variation/README.md
...asks/mathqa/alternative_worlds/output_variation/README.md
+0
-18
lm_eval/tasks/mathqa/alternative_worlds/output_variation/_mathqa_alt_yaml
...thqa/alternative_worlds/output_variation/_mathqa_alt_yaml
+0
-10
lm_eval/tasks/mathqa/alternative_worlds/output_variation/mathqa_alt.yaml
...athqa/alternative_worlds/output_variation/mathqa_alt.yaml
+0
-10
No files found.
Too many changes to show.
To preserve performance only
1000 of 1000+
files are displayed.
Plain diff
Email patch
lm_eval/tasks/hellaswag/alternative_worlds/style_03/c.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_03
group_alias
:
style_03
task
:
hellaswag_03c
task_alias
:
c
doc_to_text
:
!function
../styles.template_03
doc_to_choice
:
!function
../styles.choice_03c
lm_eval/tasks/hellaswag/alternative_worlds/style_04/a.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_04
group_alias
:
style_04
task
:
hellaswag_04a
task_alias
:
a
doc_to_text
:
!function
../styles.template_04
doc_to_choice
:
!function
../styles.choice_04a
lm_eval/tasks/hellaswag/alternative_worlds/style_04/b.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_04
group_alias
:
style_04
task
:
hellaswag_04b
task_alias
:
b
doc_to_text
:
!function
../styles.template_04
doc_to_choice
:
!function
../styles.choice_04b
lm_eval/tasks/hellaswag/alternative_worlds/style_04/c.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_04
group_alias
:
style_04
task
:
hellaswag_04c
task_alias
:
c
doc_to_text
:
!function
../styles.template_04
doc_to_choice
:
!function
../styles.choice_04c
lm_eval/tasks/hellaswag/alternative_worlds/style_05/a.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_05
group_alias
:
style_05
task
:
hellaswag_05a
task_alias
:
a
doc_to_text
:
!function
../styles.template_05
doc_to_choice
:
!function
../styles.choice_05a
lm_eval/tasks/hellaswag/alternative_worlds/style_05/b.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_05
group_alias
:
style_05
task
:
hellaswag_05b
task_alias
:
b
doc_to_text
:
!function
../styles.template_05
doc_to_choice
:
!function
../styles.choice_05b
lm_eval/tasks/hellaswag/alternative_worlds/style_05/c.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_05
group_alias
:
style_05
task
:
hellaswag_05c
task_alias
:
c
doc_to_text
:
!function
../styles.template_05
doc_to_choice
:
!function
../styles.choice_05c
lm_eval/tasks/hellaswag/alternative_worlds/style_06/a.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_06
group_alias
:
style_06
task
:
hellaswag_06a
task_alias
:
a
doc_to_text
:
!function
../styles.template_06
doc_to_choice
:
!function
../styles.choice_06a
lm_eval/tasks/hellaswag/alternative_worlds/style_06/b.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_06
group_alias
:
style_06
task
:
hellaswag_06b
task_alias
:
b
doc_to_text
:
!function
../styles.template_06
doc_to_choice
:
!function
../styles.choice_06b
lm_eval/tasks/hellaswag/alternative_worlds/style_06/c.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_06
group_alias
:
style_06
task
:
hellaswag_06c
task_alias
:
c
doc_to_text
:
!function
../styles.template_06
doc_to_choice
:
!function
../styles.choice_06c
lm_eval/tasks/hellaswag/alternative_worlds/style_07/a.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_07
group_alias
:
style_07
task
:
hellaswag_07a
task_alias
:
a
doc_to_text
:
!function
../styles.template_07
doc_to_choice
:
!function
../styles.choice_07a
lm_eval/tasks/hellaswag/alternative_worlds/style_07/b.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_07
group_alias
:
style_07
task
:
hellaswag_07b
task_alias
:
b
doc_to_text
:
!function
../styles.template_07
doc_to_choice
:
!function
../styles.choice_07b
lm_eval/tasks/hellaswag/alternative_worlds/style_07/c.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_07
group_alias
:
style_07
task
:
hellaswag_07c
task_alias
:
c
doc_to_text
:
!function
../styles.template_07
doc_to_choice
:
!function
../styles.choice_07c
lm_eval/tasks/hellaswag/alternative_worlds/style_08/a.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_08
group_alias
:
style_08
task
:
hellaswag_08a
task_alias
:
a
doc_to_text
:
!function
../styles.template_08
doc_to_choice
:
!function
../styles.choice_08a
lm_eval/tasks/hellaswag/alternative_worlds/style_08/b.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_08
group_alias
:
style_08
task
:
hellaswag_08b
task_alias
:
b
doc_to_text
:
!function
../styles.template_08
doc_to_choice
:
!function
../styles.choice_08b
lm_eval/tasks/hellaswag/alternative_worlds/style_08/c.yaml
deleted
100644 → 0
View file @
3e5e9da2
include
:
../_hellaswag_alt_yaml
group
:
hellaswag_08
group_alias
:
style_08
task
:
hellaswag_08c
task_alias
:
c
doc_to_text
:
!function
../styles.template_08
doc_to_choice
:
!function
../styles.choice_08c
lm_eval/tasks/hellaswag/alternative_worlds/styles.py
deleted
100644 → 0
View file @
3e5e9da2
import
string
from
functools
import
partial
def
doc_to_text_base
(
alphabet
,
style
,
doc
):
choices
=
doc
[
"choices"
]
num
=
len
(
choices
)
letter_list
=
[
style
.
format
(
letter
)
for
letter
in
alphabet
[
0
:
num
]]
if
"
\t
"
in
style
:
choice_string
=
"{}{}"
else
:
choice_string
=
"{} {}"
doc_to_text
=
"
\n\n
"
.
join
(
[
doc
[
"query"
]
+
"..."
,
" What is the most appropriate continuation?"
,
]
+
[
choice_string
.
format
(
i
,
j
)
for
i
,
j
in
zip
(
letter_list
,
choices
)]
)
return
doc_to_text
# Full continuation
def
choice_A
(
doc
):
return
doc
[
"choices"
]
# Letters only
def
choice_B
(
alphabet
,
style
,
doc
):
choices
=
doc
[
"choices"
]
num
=
len
(
choices
)
letter_list
=
[
style
.
format
(
letter
)
for
letter
in
alphabet
[
0
:
num
]]
if
"
\t
"
in
style
:
letter_list
=
[
letter
.
replace
(
"
\t
"
,
""
)
for
letter
in
letter_list
]
return
letter_list
# Letters + Full continuation
def
choice_C
(
alphabet
,
style
,
doc
):
choices
=
doc
[
"choices"
]
num
=
len
(
choices
)
letter_list
=
[
style
.
format
(
letter
)
for
letter
in
alphabet
[
0
:
num
]]
if
"
\t
"
not
in
style
:
letter_list
=
[
letter
+
" "
for
letter
in
letter_list
]
return
[
letter
+
choice
for
letter
,
choice
in
zip
(
letter_list
,
doc
[
"choices"
])]
template_01
=
partial
(
doc_to_text_base
,
string
.
ascii_lowercase
,
"({})"
)
choice_01a
=
choice_A
choice_01b
=
partial
(
choice_B
,
string
.
ascii_lowercase
,
"({})"
)
choice_01c
=
partial
(
choice_C
,
string
.
ascii_lowercase
,
"({})"
)
template_02
=
partial
(
doc_to_text_base
,
string
.
ascii_lowercase
,
"{})"
)
choice_02a
=
choice_A
choice_02b
=
partial
(
choice_B
,
string
.
ascii_lowercase
,
"{})"
)
choice_02c
=
partial
(
choice_C
,
string
.
ascii_lowercase
,
"{})"
)
template_03
=
partial
(
doc_to_text_base
,
string
.
ascii_lowercase
,
"{}."
)
choice_03a
=
choice_A
choice_03b
=
partial
(
choice_B
,
string
.
ascii_lowercase
,
"{}."
)
choice_03c
=
partial
(
choice_C
,
string
.
ascii_lowercase
,
"{}."
)
template_04
=
partial
(
doc_to_text_base
,
string
.
ascii_lowercase
,
"{}
\t
"
)
choice_04a
=
choice_A
choice_04b
=
partial
(
choice_B
,
string
.
ascii_lowercase
,
"{}
\t
"
)
choice_04c
=
partial
(
choice_C
,
string
.
ascii_lowercase
,
"{}
\t
"
)
template_05
=
partial
(
doc_to_text_base
,
string
.
ascii_uppercase
,
"({})"
)
choice_05a
=
choice_A
choice_05b
=
partial
(
choice_B
,
string
.
ascii_uppercase
,
"({})"
)
choice_05c
=
partial
(
choice_C
,
string
.
ascii_uppercase
,
"({})"
)
template_06
=
partial
(
doc_to_text_base
,
string
.
ascii_uppercase
,
"{})"
)
choice_06a
=
choice_A
choice_06b
=
partial
(
choice_B
,
string
.
ascii_uppercase
,
"{})"
)
choice_06c
=
partial
(
choice_C
,
string
.
ascii_uppercase
,
"{})"
)
template_07
=
partial
(
doc_to_text_base
,
string
.
ascii_uppercase
,
"{}."
)
choice_07a
=
choice_A
choice_07b
=
partial
(
choice_B
,
string
.
ascii_uppercase
,
"{}."
)
choice_07c
=
partial
(
choice_C
,
string
.
ascii_uppercase
,
"{}."
)
template_08
=
partial
(
doc_to_text_base
,
string
.
ascii_uppercase
,
"{}
\t
"
)
choice_08a
=
choice_A
choice_08b
=
partial
(
choice_B
,
string
.
ascii_uppercase
,
"{}
\t
"
)
choice_08c
=
partial
(
choice_C
,
string
.
ascii_uppercase
,
"{}
\t
"
)
lm_eval/tasks/mathqa/alternative_worlds/output_variation/README.md
deleted
100644 → 0
View file @
3e5e9da2
Investigate affect of letter options
-
(A)
-
A)
-
A.
-
A
\t
-
(a)
-
a)
-
a.
-
a
\t
Answer types:
-
letters only
-
original option
-
just letter
-
letters + continuation
-
original option
-
just letter
-
continuation
lm_eval/tasks/mathqa/alternative_worlds/output_variation/_mathqa_alt_yaml
deleted
100644 → 0
View file @
3e5e9da2
dataset_path: math_qa
output_type: multiple_choice
training_split: train
validation_split: validation
test_split: test
doc_to_target: "{{['a', 'b', 'c', 'd', 'e'].index(correct)}}"
metric_list:
- metric: acc
- metric: acc_norm
- metric: brier_score
lm_eval/tasks/mathqa/alternative_worlds/output_variation/mathqa_alt.yaml
deleted
100644 → 0
View file @
3e5e9da2
group
:
mathqa_alt_ov
task
:
-
mathqa_alt_ov_01
-
mathqa_alt_ov_02
-
mathqa_alt_ov_03
-
mathqa_alt_ov_04
-
mathqa_alt_ov_05
-
mathqa_alt_ov_06
-
mathqa_alt_ov_07
-
mathqa_alt_ov_08
Prev
1
…
8
9
10
11
12
13
14
15
16
…
50
Next
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment