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
451a1873
Commit
451a1873
authored
Aug 29, 2023
by
lintangsutawika
Browse files
adjustments
parent
a9757f02
Changes
190
Hide whitespace changes
Inline
Side-by-side
Showing
20 changed files
with
103 additions
and
2 deletions
+103
-2
lm_eval/tasks/model_written_evals/advanced_ai_risk/_generate_configs.py
...model_written_evals/advanced_ai_risk/_generate_configs.py
+29
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/_template_yaml
...tasks/model_written_evals/advanced_ai_risk/_template_yaml
+2
-2
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-coordinate-itself.yaml
...ten_evals/advanced_ai_risk/fewshot-coordinate-itself.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-coordinate-other-ais.yaml
..._evals/advanced_ai_risk/fewshot-coordinate-other-ais.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-coordinate-other-versions.yaml
...s/advanced_ai_risk/fewshot-coordinate-other-versions.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-corrigible-less-HHH.yaml
...n_evals/advanced_ai_risk/fewshot-corrigible-less-HHH.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-corrigible-more-HHH.yaml
...n_evals/advanced_ai_risk/fewshot-corrigible-more-HHH.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-corrigible-neutral-HHH.yaml
...vals/advanced_ai_risk/fewshot-corrigible-neutral-HHH.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-myopic-reward.yaml
...written_evals/advanced_ai_risk/fewshot-myopic-reward.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-one-box-tendency.yaml
...tten_evals/advanced_ai_risk/fewshot-one-box-tendency.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-power-seeking-inclination.yaml
...s/advanced_ai_risk/fewshot-power-seeking-inclination.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-general-ai.yaml
...s/advanced_ai_risk/fewshot-self-awareness-general-ai.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-good-text-model.yaml
...anced_ai_risk/fewshot-self-awareness-good-text-model.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-text-model.yaml
...s/advanced_ai_risk/fewshot-self-awareness-text-model.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-training-architecture.yaml
...ai_risk/fewshot-self-awareness-training-architecture.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-training-web-gpt.yaml
...nced_ai_risk/fewshot-self-awareness-training-web-gpt.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-survival-instinct.yaml
...ten_evals/advanced_ai_risk/fewshot-survival-instinct.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-wealth-seeking-inclination.yaml
.../advanced_ai_risk/fewshot-wealth-seeking-inclination.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/human-coordinate-itself.yaml
...itten_evals/advanced_ai_risk/human-coordinate-itself.yaml
+4
-0
lm_eval/tasks/model_written_evals/advanced_ai_risk/human-coordinate-other-ais.yaml
...en_evals/advanced_ai_risk/human-coordinate-other-ais.yaml
+4
-0
No files found.
lm_eval/tasks/model_written_evals/generate_configs.py
→
lm_eval/tasks/model_written_evals/
advanced_ai_risk/_
generate_configs.py
View file @
451a1873
...
...
@@ -2,24 +2,21 @@ import yaml
import
inspect
import
datasets
from
importlib
import
import_module
from
tqdm
import
tqdm
def
main
()
->
None
:
dataset_path
=
"persona"
dataset_full_path
=
inspect
.
getfile
(
import_module
(
f
"lm_eval.datasets.model_written_evals.
{
dataset_path
}
"
)
)
for
task
in
datasets
.
get_dataset_infos
(
dataset_full_path
).
keys
():
file_name
=
f
"
{
dataset_path
}
/
{
task
}
.yaml"
dataset_path
=
"EleutherAI/advanced_ai_risk"
for
task
in
tqdm
(
datasets
.
get_dataset_infos
(
dataset_path
).
keys
()):
file_name
=
f
"
{
task
}
.yaml"
try
:
with
open
(
f
"
{
file_name
}
"
,
"w"
)
as
f
:
f
.
write
(
"# Generated by generate_configs.py
\n
"
)
f
.
write
(
"# Generated by
_
generate_configs.py
\n
"
)
yaml
.
dump
(
{
"include"
:
"template_yaml"
,
"task"
:
f
"
{
dataset_path
}
_
{
task
}
"
,
"include"
:
"
_
template_yaml"
,
"task"
:
f
"
{
dataset_path
.
split
(
'/'
)[
-
1
]
}
_
{
task
}
"
,
"dataset_name"
:
task
,
},
f
,
...
...
lm_eval/tasks/model_written_evals/advanced_ai_risk/_template_yaml
View file @
451a1873
group: advance_ai_risk
dataset_path: EleutherAI/advance_ai_risk
group: advance
d
_ai_risk
dataset_path: EleutherAI/advance
d
_ai_risk
output_type: multiple_choice
validation_split: train
doc_to_text: "Human: {question}\n\nAssistant:"
...
...
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-coordinate-itself.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-coordinate-itself
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-coordinate-itself
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-coordinate-other-ais.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-coordinate-other-ais
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-coordinate-other-ais
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-coordinate-other-versions.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-coordinate-other-versions
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-coordinate-other-versions
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-corrigible-less-HHH.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-corrigible-less-HHH
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-corrigible-less-HHH
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-corrigible-more-HHH.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-corrigible-more-HHH
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-corrigible-more-HHH
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-corrigible-neutral-HHH.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-corrigible-neutral-HHH
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-corrigible-neutral-HHH
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-myopic-reward.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-myopic-reward
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-myopic-reward
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-one-box-tendency.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-one-box-tendency
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-one-box-tendency
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-power-seeking-inclination.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-power-seeking-inclination
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-power-seeking-inclination
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-general-ai.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-self-awareness-general-ai
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-self-awareness-general-ai
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-good-text-model.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-self-awareness-good-text-model
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-self-awareness-good-text-model
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-text-model.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-self-awareness-text-model
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-self-awareness-text-model
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-training-architecture.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-self-awareness-training-architecture
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-self-awareness-training-architecture
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-self-awareness-training-web-gpt.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-self-awareness-training-web-gpt
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-self-awareness-training-web-gpt
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-survival-instinct.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-survival-instinct
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-survival-instinct
lm_eval/tasks/model_written_evals/advanced_ai_risk/fewshot-wealth-seeking-inclination.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
fewshot-wealth-seeking-inclination
include
:
_template_yaml
task
:
advanced_ai_risk_fewshot-wealth-seeking-inclination
lm_eval/tasks/model_written_evals/advanced_ai_risk/human-coordinate-itself.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
human-coordinate-itself
include
:
_template_yaml
task
:
advanced_ai_risk_human-coordinate-itself
lm_eval/tasks/model_written_evals/advanced_ai_risk/human-coordinate-other-ais.yaml
0 → 100644
View file @
451a1873
# Generated by _generate_configs.py
dataset_name
:
human-coordinate-other-ais
include
:
_template_yaml
task
:
advanced_ai_risk_human-coordinate-other-ais
Prev
1
2
3
4
5
…
10
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