[Dataset] LongBench (#236)

Co-authored-by: wangchonghua <wangchonghua@pjlab.org.cn>

configs/datasets/longbench/longbench2wikimqa/longbench_2wikimqa_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_2wikimqa_gen_6b3efc import LongBench_2wikimqa_datasets  # noqa: F401, F403

configs/datasets/longbench/longbench2wikimqa/longbench_2wikimqa_gen_6b3efc.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchF1Evaluator, LongBench2wikimqaDataset
+
+LongBench_2wikimqa_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_2wikimqa_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='Answer the question based on the given passages. Only give me the answer and do not output any other words.\n\nThe following are given passages.\n{context}\n\nAnswer the question based on the given passages. Only give me the answer and do not output any other words.\n\nQuestion: {input}\nAnswer:'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=32)
+)
+
+LongBench_2wikimqa_eval_cfg = dict(
+    evaluator=dict(type=LongBenchF1Evaluator),
+    pred_role='BOT'
+)
+
+LongBench_2wikimqa_datasets = [
+    dict(
+        type=LongBench2wikimqaDataset,
+        abbr='LongBench_2wikimqa',
+        path='THUDM/LongBench',
+        name='2wikimqa',
+        reader_cfg=LongBench_2wikimqa_reader_cfg,
+        infer_cfg=LongBench_2wikimqa_infer_cfg,
+        eval_cfg=LongBench_2wikimqa_eval_cfg)
+]

configs/datasets/longbench/longbenchdureader/longbench_dureader_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_dureader_gen_c6c7e4 import LongBench_dureader_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchdureader/longbench_dureader_gen_c6c7e4.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchRougeEvaluator, LongBenchdureaderDataset
+
+LongBench_dureader_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_dureader_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='请基于给定的文章回答下述问题。\n\n文章：{context}\n\n请基于上述文章回答下面的问题。\n\n问题：{input}\n回答：'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=128)
+)
+
+LongBench_dureader_eval_cfg = dict(
+    evaluator=dict(type=LongBenchRougeEvaluator, language='zh'),
+    pred_role='BOT'
+)
+
+LongBench_dureader_datasets = [
+    dict(
+        type=LongBenchdureaderDataset,
+        abbr='LongBench_dureader',
+        path='THUDM/LongBench',
+        name='dureader',
+        reader_cfg=LongBench_dureader_reader_cfg,
+        infer_cfg=LongBench_dureader_infer_cfg,
+        eval_cfg=LongBench_dureader_eval_cfg)
+]

configs/datasets/longbench/longbenchgov_report/longbench_gov_report_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_gov_report_gen_54c5b0 import LongBench_gov_report_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchgov_report/longbench_gov_report_gen_54c5b0.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchRougeEvaluator, LongBenchgov_reportDataset
+
+LongBench_gov_report_reader_cfg = dict(
+    input_columns=['context'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_gov_report_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='You are given a report by a government agency. Write a one-page summary of the report.\n\nReport:\n{context}\n\nNow, write a one-page summary of the report.\n\nSummary:'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=512)
+)
+
+LongBench_gov_report_eval_cfg = dict(
+    evaluator=dict(type=LongBenchRougeEvaluator),
+    pred_role='BOT'
+)
+
+LongBench_gov_report_datasets = [
+    dict(
+        type=LongBenchgov_reportDataset,
+        abbr='LongBench_gov_report',
+        path='THUDM/LongBench',
+        name='gov_report',
+        reader_cfg=LongBench_gov_report_reader_cfg,
+        infer_cfg=LongBench_gov_report_infer_cfg,
+        eval_cfg=LongBench_gov_report_eval_cfg)
+]

configs/datasets/longbench/longbenchhotpotqa/longbench_hotpotqa_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_hotpotqa_gen_6b3efc import LongBench_hotpotqa_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchhotpotqa/longbench_hotpotqa_gen_6b3efc.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchF1Evaluator, LongBenchhotpotqaDataset
+
+LongBench_hotpotqa_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_hotpotqa_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='Answer the question based on the given passages. Only give me the answer and do not output any other words.\n\nThe following are given passages.\n{context}\n\nAnswer the question based on the given passages. Only give me the answer and do not output any other words.\n\nQuestion: {input}\nAnswer:'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=32)
+)
+
+LongBench_hotpotqa_eval_cfg = dict(
+    evaluator=dict(type=LongBenchF1Evaluator),
+    pred_role='BOT'
+)
+
+LongBench_hotpotqa_datasets = [
+    dict(
+        type=LongBenchhotpotqaDataset,
+        abbr='LongBench_hotpotqa',
+        path='THUDM/LongBench',
+        name='hotpotqa',
+        reader_cfg=LongBench_hotpotqa_reader_cfg,
+        infer_cfg=LongBench_hotpotqa_infer_cfg,
+        eval_cfg=LongBench_hotpotqa_eval_cfg)
+]

configs/datasets/longbench/longbenchlcc/longbench_lcc_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_lcc_gen_6ba507 import LongBench_lcc_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchlcc/longbench_lcc_gen_6ba507.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchCodeSimEvaluator, LongBenchlccDataset
+
+LongBench_lcc_reader_cfg = dict(
+    input_columns=['context'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_lcc_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='Please complete the code given below. \n{context}Next line of code:\n'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=64)
+)
+
+LongBench_lcc_eval_cfg = dict(
+    evaluator=dict(type=LongBenchCodeSimEvaluator),
+    pred_role='BOT'
+)
+
+LongBench_lcc_datasets = [
+    dict(
+        type=LongBenchlccDataset,
+        abbr='LongBench_lcc',
+        path='THUDM/LongBench',
+        name='lcc',
+        reader_cfg=LongBench_lcc_reader_cfg,
+        infer_cfg=LongBench_lcc_infer_cfg,
+        eval_cfg=LongBench_lcc_eval_cfg)
+]

configs/datasets/longbench/longbenchlsht/longbench_lsht_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_lsht_gen_e8a339 import LongBench_lsht_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchlsht/longbench_lsht_gen_e8a339.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchClassificationEvaluator, LongBenchlshtDataset
+
+LongBench_lsht_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='all_labels',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_lsht_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='请判断给定新闻的类别，下面是一些例子。\n\n{context}\n{input}'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=64)
+)
+
+LongBench_lsht_eval_cfg = dict(
+    evaluator=dict(type=LongBenchClassificationEvaluator),
+    pred_role='BOT'
+)
+
+LongBench_lsht_datasets = [
+    dict(
+        type=LongBenchlshtDataset,
+        abbr='LongBench_lsht',
+        path='THUDM/LongBench',
+        name='lsht',
+        reader_cfg=LongBench_lsht_reader_cfg,
+        infer_cfg=LongBench_lsht_infer_cfg,
+        eval_cfg=LongBench_lsht_eval_cfg)
+]

configs/datasets/longbench/longbenchmultifieldqa_en/longbench_multifieldqa_en_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_multifieldqa_en_gen_d3838e import LongBench_multifieldqa_en_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchmultifieldqa_en/longbench_multifieldqa_en_gen_d3838e.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchF1Evaluator, LongBenchmultifieldqa_enDataset
+
+LongBench_multifieldqa_en_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_multifieldqa_en_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='Read the following text and answer briefly.\n\n{context}\n\nNow, answer the following question based on the above text, only give me the answer and do not output any other words.\n\nQuestion: {input}\nAnswer:'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=64)
+)
+
+LongBench_multifieldqa_en_eval_cfg = dict(
+    evaluator=dict(type=LongBenchF1Evaluator),
+    pred_role='BOT'
+)
+
+LongBench_multifieldqa_en_datasets = [
+    dict(
+        type=LongBenchmultifieldqa_enDataset,
+        abbr='LongBench_multifieldqa_en',
+        path='THUDM/LongBench',
+        name='multifieldqa_en',
+        reader_cfg=LongBench_multifieldqa_en_reader_cfg,
+        infer_cfg=LongBench_multifieldqa_en_infer_cfg,
+        eval_cfg=LongBench_multifieldqa_en_eval_cfg)
+]

configs/datasets/longbench/longbenchmultifieldqa_zh/longbench_multifieldqa_zh_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_multifieldqa_zh_gen_e9a7ef import LongBench_multifieldqa_zh_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchmultifieldqa_zh/longbench_multifieldqa_zh_gen_e9a7ef.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchF1Evaluator, LongBenchmultifieldqa_zhDataset
+
+LongBench_multifieldqa_zh_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_multifieldqa_zh_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='阅读以下文字并用中文简短回答：\n\n{context}\n\n现在请基于上面的文章回答下面的问题，只告诉我答案，不要输出任何其他字词。\n\n问题：{input}\n回答：'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=64)
+)
+
+LongBench_multifieldqa_zh_eval_cfg = dict(
+    evaluator=dict(type=LongBenchF1Evaluator, language='zh'),
+    pred_role='BOT'
+)
+
+LongBench_multifieldqa_zh_datasets = [
+    dict(
+        type=LongBenchmultifieldqa_zhDataset,
+        abbr='LongBench_multifieldqa_zh',
+        path='THUDM/LongBench',
+        name='multifieldqa_zh',
+        reader_cfg=LongBench_multifieldqa_zh_reader_cfg,
+        infer_cfg=LongBench_multifieldqa_zh_infer_cfg,
+        eval_cfg=LongBench_multifieldqa_zh_eval_cfg)
+]

configs/datasets/longbench/longbenchmusique/longbench_musique_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_musique_gen_6b3efc import LongBench_musique_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchmusique/longbench_musique_gen_6b3efc.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchF1Evaluator, LongBenchmusiqueDataset
+
+LongBench_musique_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_musique_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='Answer the question based on the given passages. Only give me the answer and do not output any other words.\n\nThe following are given passages.\n{context}\n\nAnswer the question based on the given passages. Only give me the answer and do not output any other words.\n\nQuestion: {input}\nAnswer:'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=32)
+)
+
+LongBench_musique_eval_cfg = dict(
+    evaluator=dict(type=LongBenchF1Evaluator),
+    pred_role='BOT'
+)
+
+LongBench_musique_datasets = [
+    dict(
+        type=LongBenchmusiqueDataset,
+        abbr='LongBench_musique',
+        path='THUDM/LongBench',
+        name='musique',
+        reader_cfg=LongBench_musique_reader_cfg,
+        infer_cfg=LongBench_musique_infer_cfg,
+        eval_cfg=LongBench_musique_eval_cfg)
+]

configs/datasets/longbench/longbenchnarrativeqa/longbench_narrativeqa_gen.py

+from mmengine.config import read_base
+
+with read_base():
+    from .longbench_narrativeqa_gen_a68305 import LongBench_narrativeqa_datasets  # noqa: F401, F403

configs/datasets/longbench/longbenchnarrativeqa/longbench_narrativeqa_gen_a68305.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import LongBenchF1Evaluator, LongBenchnarrativeqaDataset
+
+LongBench_narrativeqa_reader_cfg = dict(
+    input_columns=['context', 'input'],
+    output_column='answers',
+    train_split='test',
+    test_split='test'
+)
+
+LongBench_narrativeqa_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                dict(role='HUMAN', prompt='You are given a story, which can be either a novel or a movie script, and a question. Answer the question as concisely as you can, using a single phrase if possible. Do not provide any explanation.\n\nStory: {context}\n\nNow, answer the question based on the story as concisely as you can, using a single phrase if possible. Do not provide any explanation.\n\nQuestion: {input}\n\nAnswer:'),
+            ], )),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=128)
+)
+
+LongBench_narrativeqa_eval_cfg = dict(
+    evaluator=dict(type=LongBenchF1Evaluator),
+    pred_role='BOT'
+)
+
+LongBench_narrativeqa_datasets = [
+    dict(
+        type=LongBenchnarrativeqaDataset,
+        abbr='LongBench_narrativeqa',
+        path='THUDM/LongBench',
+        name='narrativeqa',
+        reader_cfg=LongBench_narrativeqa_reader_cfg,
+        infer_cfg=LongBench_narrativeqa_infer_cfg,
+        eval_cfg=LongBench_narrativeqa_eval_cfg)
+]