Commit 01d89cdc authored by Baber's avatar Baber
Browse files

rename

parent be7be189
include: niah_1.yaml include: niah_single_1.yaml
task: ruler_cwe task: ruler_cwe
download_dataset: !function cwe_utils.get_cw_dataset download_dataset: !function cwe_utils.get_cw_dataset
generation_kwargs: generation_kwargs:
......
include: niah_1.yaml include: niah_single_1.yaml
task: ruler_fwe task: ruler_fwe
download_dataset: !function fwe_utils.fwe_download download_dataset: !function fwe_utils.fwe_download
generation_kwargs: generation_kwargs:
......
task: niah_4 task: niah_multikey_1
include: niah_1.yaml include: niah_single_1.yaml
download_dataset: !function utils.niah_multikey_1 download_dataset: !function utils.niah_multikey_1
task: niah_5 task: niah_multikey_2
include: niah_1.yaml include: niah_single_1.yaml
download_dataset: !function utils.niah_multikey_2 download_dataset: !function utils.niah_multikey_2
task: niah_6 task: niah_multikey_3
include: niah_1.yaml include: niah_single_1.yaml
download_dataset: !function utils.niah_multikey_3 download_dataset: !function utils.niah_multikey_3
task: niah_8 task: niah_multiquery
include: niah_1.yaml include: niah_single_1.yaml
download_dataset: !function utils.niah_multiquery download_dataset: !function utils.niah_multiquery
task: niah_7 task: niah_multivalue
include: niah_1.yaml include: niah_single_1.yaml
download_dataset: !function utils.niah_multivalue download_dataset: !function utils.niah_multivalue
tag: tag:
- ruler - longcxt
task: niah_1 task: niah_single_1
dataset_path: "" dataset_path: ""
dataset_name: "" dataset_name: ""
output_type: generate_until output_type: generate_until
...@@ -10,7 +10,6 @@ doc_to_text: "{{input}}" ...@@ -10,7 +10,6 @@ doc_to_text: "{{input}}"
doc_to_target: "{{outputs[0]}}" doc_to_target: "{{outputs[0]}}"
gen_prefix: "{{gen_prefix}}" gen_prefix: "{{gen_prefix}}"
process_results: !function utils.process_results process_results: !function utils.process_results
metric_list: metric_list:
- metric: "4096" - metric: "4096"
aggregation: !function utils.aggregate_metrics aggregation: !function utils.aggregate_metrics
......
task: niah_2 task: niah_single_2
include: niah_1.yaml include: niah_single_1.yaml
download_dataset: !function utils.niah_single_2 download_dataset: !function utils.niah_single_2
task: niah_3 task: niah_single_3
include: niah_1.yaml include: niah_single_1.yaml
download_dataset: !function utils.niah_single_3 download_dataset: !function utils.niah_single_3
include: niah_1.yaml include: niah_single_1.yaml
task: ruler_qa_squad task: ruler_qa_squad
download_dataset: !function qa_utils.get_squad download_dataset: !function qa_utils.get_squad
test_split: test test_split: test
......
...@@ -15,7 +15,7 @@ ...@@ -15,7 +15,7 @@
import itertools # noqa: I001 import itertools # noqa: I001
import random import random
from functools import cache, partial from functools import cache
import datasets import datasets
import requests import requests
...@@ -237,5 +237,13 @@ def get_qa_dataset(ds, **kwargs): ...@@ -237,5 +237,13 @@ def get_qa_dataset(ds, **kwargs):
} }
get_squad = partial(get_qa_dataset, "squad") def get_squad(**kwargs):
get_hotpotqa = partial(get_qa_dataset, "hotpotqa") return get_qa_dataset("squad", **kwargs)
def get_hotpotqa(**kwargs):
return get_qa_dataset("hotpotqa", **kwargs)
# get_squad = lambda **kwargs: partial(get_qa_dataset, "squad")(**kwargs)
# get_hotpotqa = lambda **kwargs: partial(get_qa_dataset, "hotpotqa")(**kwargs)
group: ruler
task:
- niah_single_1
- niah_single_2
- niah_single_3
- niah_multikey_1
- niah_multikey_2
- niah_multikey_3
- niah_multiquery
- niah_multivalue
- ruler_vt
- ruler_cwe
- ruler_fwe
- ruler_qa_squad
- ruler_qa_hotpot
aggregate_metric_list:
- metric: acc
weight_by_size: False
metadata:
version: 1
include: niah_1.yaml include: niah_single_1.yaml
task: ruler_vt task: ruler_vt
download_dataset: !function vt_utils.get_vt_dataset download_dataset: !function vt_utils.get_vt_dataset
generation_kwargs: generation_kwargs:
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment