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
55eff889
Commit
55eff889
authored
Dec 08, 2023
by
lintangsutawika
Browse files
made prompt and output variations
parent
c74e2761
Changes
32
Hide whitespace changes
Inline
Side-by-side
Showing
12 changed files
with
63 additions
and
30 deletions
+63
-30
lm_eval/tasks/arc/alternative_worlds/output_variation/style_06/c.yaml
...s/arc/alternative_worlds/output_variation/style_06/c.yaml
+3
-4
lm_eval/tasks/arc/alternative_worlds/output_variation/style_07/a.yaml
...s/arc/alternative_worlds/output_variation/style_07/a.yaml
+3
-4
lm_eval/tasks/arc/alternative_worlds/output_variation/style_07/b.yaml
...s/arc/alternative_worlds/output_variation/style_07/b.yaml
+3
-4
lm_eval/tasks/arc/alternative_worlds/output_variation/style_07/c.yaml
...s/arc/alternative_worlds/output_variation/style_07/c.yaml
+3
-4
lm_eval/tasks/arc/alternative_worlds/output_variation/style_08/a.yaml
...s/arc/alternative_worlds/output_variation/style_08/a.yaml
+3
-4
lm_eval/tasks/arc/alternative_worlds/output_variation/style_08/b.yaml
...s/arc/alternative_worlds/output_variation/style_08/b.yaml
+3
-4
lm_eval/tasks/arc/alternative_worlds/output_variation/style_08/c.yaml
...s/arc/alternative_worlds/output_variation/style_08/c.yaml
+3
-4
lm_eval/tasks/arc/alternative_worlds/output_variation/styles.py
...l/tasks/arc/alternative_worlds/output_variation/styles.py
+3
-2
lm_eval/tasks/arc/alternative_worlds/prompt_variation/_arc_easy_alt_yaml
...rc/alternative_worlds/prompt_variation/_arc_easy_alt_yaml
+24
-0
lm_eval/tasks/arc/alternative_worlds/prompt_variation/style_01.yaml
...sks/arc/alternative_worlds/prompt_variation/style_01.yaml
+5
-0
lm_eval/tasks/arc/alternative_worlds/prompt_variation/style_02.yaml
...sks/arc/alternative_worlds/prompt_variation/style_02.yaml
+5
-0
lm_eval/tasks/arc/alternative_worlds/prompt_variation/style_03.yaml
...sks/arc/alternative_worlds/prompt_variation/style_03.yaml
+5
-0
No files found.
lm_eval/tasks/arc/alternative_worlds/style_06/c.yaml
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
style_06/c.yaml
View file @
55eff889
include
:
../_arc_easy_alt_yaml
include
:
../_arc_easy_alt_yaml
group
:
arc_easy_06
group
:
arc_easy_alt_ov_06
group_alias
:
style_06
task
:
arc_easy_alt_ov_06c
task
:
arc_easy_06c
task_alias
:
c
doc_to_text
:
!function
../styles.template_06
doc_to_text
:
!function
../styles.template_06
doc_to_choice
:
!function
../styles.choice_06c
doc_to_choice
:
!function
../styles.choice_06c
doc_to_decontamination_query
:
!function
../styles.template_06
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/style_07/a.yaml
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
style_07/a.yaml
View file @
55eff889
include
:
../_arc_easy_alt_yaml
include
:
../_arc_easy_alt_yaml
group
:
arc_easy_07
group
:
arc_easy_alt_ov_07
group_alias
:
style_07
task
:
arc_easy_alt_ov_07a
task
:
arc_easy_07a
task_alias
:
a
doc_to_text
:
!function
../styles.template_07
doc_to_text
:
!function
../styles.template_07
doc_to_choice
:
!function
../styles.choice_07a
doc_to_choice
:
!function
../styles.choice_07a
doc_to_decontamination_query
:
!function
../styles.template_07
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/style_07/b.yaml
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
style_07/b.yaml
View file @
55eff889
include
:
../_arc_easy_alt_yaml
include
:
../_arc_easy_alt_yaml
group
:
arc_easy_07
group
:
arc_easy_alt_ov_07
group_alias
:
style_07
task
:
arc_easy_alt_ov_07b
task
:
arc_easy_07b
task_alias
:
b
doc_to_text
:
!function
../styles.template_07
doc_to_text
:
!function
../styles.template_07
doc_to_choice
:
!function
../styles.choice_07b
doc_to_choice
:
!function
../styles.choice_07b
doc_to_decontamination_query
:
!function
../styles.template_07
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/style_07/c.yaml
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
style_07/c.yaml
View file @
55eff889
include
:
../_arc_easy_alt_yaml
include
:
../_arc_easy_alt_yaml
group
:
arc_easy_07
group
:
arc_easy_alt_ov_07
group_alias
:
style_07
task
:
arc_easy_alt_ov_07c
task
:
arc_easy_07c
task_alias
:
c
doc_to_text
:
!function
../styles.template_07
doc_to_text
:
!function
../styles.template_07
doc_to_choice
:
!function
../styles.choice_07c
doc_to_choice
:
!function
../styles.choice_07c
doc_to_decontamination_query
:
!function
../styles.template_07
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/style_08/a.yaml
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
style_08/a.yaml
View file @
55eff889
include
:
../_arc_easy_alt_yaml
include
:
../_arc_easy_alt_yaml
group
:
arc_easy_08
group
:
arc_easy_alt_ov_08
group_alias
:
style_08
task
:
arc_easy_alt_ov_08a
task
:
arc_easy_08a
task_alias
:
a
doc_to_text
:
!function
../styles.template_08
doc_to_text
:
!function
../styles.template_08
doc_to_choice
:
!function
../styles.choice_08a
doc_to_choice
:
!function
../styles.choice_08a
doc_to_decontamination_query
:
!function
../styles.template_08
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/style_08/b.yaml
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
style_08/b.yaml
View file @
55eff889
include
:
../_arc_easy_alt_yaml
include
:
../_arc_easy_alt_yaml
group
:
arc_easy_08
group
:
arc_easy_alt_ov_08
group_alias
:
style_08
task
:
arc_easy_alt_ov_08b
task
:
arc_easy_08b
task_alias
:
b
doc_to_text
:
!function
../styles.template_08
doc_to_text
:
!function
../styles.template_08
doc_to_choice
:
!function
../styles.choice_08b
doc_to_choice
:
!function
../styles.choice_08b
doc_to_decontamination_query
:
!function
../styles.template_08
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/style_08/c.yaml
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
style_08/c.yaml
View file @
55eff889
include
:
../_arc_easy_alt_yaml
include
:
../_arc_easy_alt_yaml
group
:
arc_easy_08
group
:
arc_easy_alt_ov_08
group_alias
:
style_08
task
:
arc_easy_alt_ov_08c
task
:
arc_easy_08c
task_alias
:
c
doc_to_text
:
!function
../styles.template_08
doc_to_text
:
!function
../styles.template_08
doc_to_choice
:
!function
../styles.choice_08c
doc_to_choice
:
!function
../styles.choice_08c
doc_to_decontamination_query
:
!function
../styles.template_08
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/styles.py
→
lm_eval/tasks/arc/alternative_worlds/
output_variation/
styles.py
View file @
55eff889
...
@@ -14,11 +14,12 @@ def doc_to_text_base(alphabet, style, doc):
...
@@ -14,11 +14,12 @@ def doc_to_text_base(alphabet, style, doc):
else
:
else
:
choice_string
=
"{} {}"
choice_string
=
"{} {}"
doc_to_text
=
"
\n
\n
"
.
join
(
doc_to_text
=
"
\n
"
.
join
(
[
[
"Question: "
+
doc
[
"question"
]
+
"
\n
Answer:"
,
"Question: "
+
doc
[
"question"
],
]
]
+
[
choice_string
.
format
(
i
,
j
)
for
i
,
j
in
zip
(
letter_list
,
choices
)]
+
[
choice_string
.
format
(
i
,
j
)
for
i
,
j
in
zip
(
letter_list
,
choices
)]
+
[
"Answer:"
]
)
)
return
doc_to_text
return
doc_to_text
...
...
lm_eval/tasks/arc/alternative_worlds/prompt_variation/_arc_easy_alt_yaml
0 → 100644
View file @
55eff889
group:
- ai2_arc
task: arc_easy
dataset_path: ai2_arc
dataset_name: ARC-Easy
output_type: multiple_choice
training_split: train
validation_split: validation
test_split: test
doc_to_text: "Question: {{question}}\nAnswer:"
doc_to_target: "{{choices.label.index(answerKey)}}"
doc_to_choice: "{{choices.text}}"
should_decontaminate: true
doc_to_decontamination_query: "Question: {{question}}\nAnswer:"
metric_list:
- metric: acc
aggregation: mean
higher_is_better: true
- metric: acc_norm
aggregation: mean
higher_is_better: true
- metric: brier_score
aggregation: brier_score
higher_is_better: false
lm_eval/tasks/arc/alternative_worlds/prompt_variation/style_01.yaml
0 → 100644
View file @
55eff889
include
:
_arc_easy_alt_yaml
group
:
arc_easy_alt_pv
task
:
arc_easy_alt_pv_01
doc_to_text
:
"
{{question}}"
doc_to_decontamination_query
:
"
{{question}}"
lm_eval/tasks/arc/alternative_worlds/prompt_variation/style_02.yaml
0 → 100644
View file @
55eff889
include
:
_arc_easy_alt_yaml
group
:
arc_easy_alt_pv
task
:
arc_easy_alt_pv_02
doc_to_text
:
"
Q:
{{question}}
\n
A:"
doc_to_decontamination_query
:
"
Q:
{{question}}
\n
A:"
\ No newline at end of file
lm_eval/tasks/arc/alternative_worlds/prompt_variation/style_03.yaml
0 → 100644
View file @
55eff889
include
:
_arc_easy_alt_yaml
group
:
arc_easy_alt_pv
task
:
arc_easy_alt_pv_03
doc_to_text
:
"
Question:
{{question}}
\n
Answer:"
doc_to_decontamination_query
:
"
Question:
{{question}}
\n
Answer:"
\ No newline at end of file
Prev
1
2
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