From c94cc943485e275897ad95cfa5192ff8e066378a Mon Sep 17 00:00:00 2001 From: Leymore Date: Wed, 5 Jul 2023 10:33:12 +0800 Subject: [PATCH] Add release contribution --- configs/datasets/ARC_c/ARC_c_gen.py | 4 + configs/datasets/ARC_c/ARC_c_gen_3f3039.py | 42 ++++ configs/datasets/ARC_e/ARC_e_ppl_f86898.py | 53 +++++ .../CLUE_CMRC/CLUE_CMRC_gen_220a83.py | 27 +++ .../CLUE_afqmc/CLUE_afqmc_ppl_c83c36.py | 50 +++++ configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl.py | 4 + .../CLUE_cmnli/CLUE_cmnli_ppl_991e1b.py | 52 +++++ .../CLUE_cmnli/CLUE_cmnli_ppl_b78ad4.py | 36 ++++ .../CLUE_ocnli/CLUE_ocnli_gen_01899f.py | 43 ++++ .../FewCLUE_bustm/FewCLUE_bustm_ppl_4f864a.py | 43 ++++ .../datasets/FewCLUE_chid/FewCLUE_chid_gen.py | 4 + .../FewCLUE_cluewsc/FewCLUE_cluewsc_ppl.py | 4 + .../FewCLUE_csl/FewCLUE_csl_gen_1b0c02.py | 50 +++++ .../FewCLUE_eprstmt_gen_d6d06d.py | 48 +++++ .../FewCLUE_eprstmt_ppl_d3c387.py | 49 +++++ .../FewCLUE_tnews/FewCLUE_tnews_gen.py | 4 + configs/datasets/PJExam/PJExam_gen_785c37.py | 54 +++++ .../SuperGLUE_AX_b_ppl_a65d62.py | 34 ++++ .../SuperGLUE_AX_g/SuperGLUE_AX_g_ppl.py | 4 + .../SuperGLUE_AX_g_ppl_d489ee.py | 34 ++++ .../SuperGLUE_BoolQ_ppl_c040f6.py | 34 ++++ .../SuperGLUE_COPA_ppl_0ef2f8.py | 45 +++++ .../SuperGLUE_MultiRC_ppl_83a304.py | 47 +++++ .../SuperGLUE_RTE/SuperGLUE_RTE_gen.py | 4 + .../SuperGLUE_WSC/SuperGLUE_WSC_gen.py | 4 + .../SuperGLUE_WSC/SuperGLUE_WSC_gen_d8d441.py | 42 ++++ .../SuperGLUE_WiC/SuperGLUE_WiC_ppl_ab6e84.py | 38 ++++ configs/datasets/XLSum/XLSum_gen_1cc5f6.py | 29 +++ configs/datasets/apps/apps_gen.py | 4 + .../civilcomments/civilcomments_ppl_fb1666.py | 35 ++++ configs/datasets/collections/base_medium.py | 57 ++++++ configs/datasets/collections/example.py | 7 + .../commonsenseqa/commonsenseqa_ppl_665f66.py | 45 +++++ .../commonsenseqa/commonsenseqa_ppl_ddd9f7.py | 55 ++++++ configs/datasets/crowspairs/crowspairs_gen.py | 4 + .../crowspairs/crowspairs_gen_dd110a.py | 39 ++++ .../crowspairs/crowspairs_ppl_58335f.py | 32 +++ .../crowspairs/crowspairs_ppl_f60797.py | 40 ++++ configs/datasets/drop/drop_gen.py | 4 + configs/datasets/flores/flores_gen_e7dec6.py | 161 +++++++++++++++ configs/datasets/glm/ceval.py | 118 +++++++++++ configs/datasets/glm/mmlu.py | 155 +++++++++++++++ .../govrepcrs/govrepcrs_gen_7643d5.py | 35 ++++ configs/datasets/gsm8k/gsm8k_gen.py | 4 + configs/datasets/gsm8k/gsm8k_gen_2dd372.py | 41 ++++ configs/datasets/humaneval/humaneval_gen.py | 4 + .../humaneval/humaneval_gen_28e126.py | 30 +++ configs/datasets/lcsts/lcsts_gen_427fde.py | 32 +++ configs/datasets/math/math_gen_78bcba.py | 68 +++++++ configs/datasets/mmlu/mmlu_gen_057057.py | 123 ++++++++++++ configs/datasets/mmlu/mmlu_gen_36560d.py | 109 ++++++++++ configs/datasets/mmlu/mmlu_gen_a568f1.py | 123 ++++++++++++ configs/datasets/nq/nq_gen.py | 4 + configs/datasets/nq/nq_gen_a6ffca.py | 30 +++ configs/datasets/piqa/piqa_ppl.py | 4 + .../qaspercut/qaspercut_gen_943606.py | 30 +++ configs/datasets/safety/safety_gen_c0a5b8.py | 30 +++ configs/datasets/siqa/siqa_ppl_764e42.py | 34 ++++ configs/datasets/siqa/siqa_ppl_b27551.py | 33 ++++ .../summedits/summedits_ppl_f2bd6e.py | 58 ++++++ configs/datasets/triviaqa/triviaqa_gen.py | 4 + .../datasets/triviaqa/triviaqa_gen_cc3cbf.py | 33 ++++ .../triviaqarc/triviaqarc_gen_6c1726.py | 37 ++++ configs/datasets/truthfulqa/truthfulqa_gen.py | 4 + .../truthfulqa/truthfulqa_gen_0a3a53.py | 40 ++++ configs/datasets/z_bench/z_bench_gen.py | 4 + configs/summarizers/groups/agieval.py | 17 ++ configs/summarizers/medium.py | 88 +++++++++ configs/summarizers/small.py | 64 ++++++ docs/en/_static/css/readthedocs.css | 62 ++++++ docs/en/_static/image/logo.png | Bin 0 -> 12650 bytes docs/en/user_guides/experimentation.md | 86 ++++++++ docs/en/user_guides/framework_overview.md | 1 + docs/zh_cn/_static/js/custom.js | 10 + docs/zh_cn/index.rst | 85 ++++++++ docs/zh_cn/prompt/meta_template.md | 163 +++++++++++++++ docs/zh_cn/tools.md | 66 +++++++ docs/zh_cn/user_guides/dataset_prepare.md | 105 ++++++++++ opencompass/datasets/GaokaoBench.py | 132 +++++++++++++ opencompass/datasets/afqmcd.py | 21 ++ opencompass/datasets/ax.py | 24 +++ opencompass/datasets/base.py | 28 +++ opencompass/datasets/bbh.py | 71 +++++++ opencompass/datasets/boolq.py | 41 ++++ opencompass/datasets/bustum.py | 21 ++ opencompass/datasets/cmrc.py | 40 ++++ opencompass/datasets/govrepcrs.py | 37 ++++ opencompass/datasets/iwslt2017.py | 16 ++ opencompass/datasets/narrativeqa.py | 43 ++++ opencompass/datasets/wsc.py | 58 ++++++ opencompass/models/openai_api.py | 154 +++++++++++++++ .../icl_inferencer/icl_base_inferencer.py | 162 +++++++++++++++ .../icl_retriever/icl_mdl_retriever.py | 186 +++++++++++++++++ .../icl_retriever/icl_votek_retriever.py | 99 ++++++++++ opencompass/openicl/utils/api_service.py | 104 ++++++++++ opencompass/partitioners/naive.py | 60 ++++++ opencompass/partitioners/size.py | 187 ++++++++++++++++++ opencompass/registry.py | 24 +++ opencompass/runners/base.py | 80 ++++++++ opencompass/runners/slurm.py | 148 ++++++++++++++ opencompass/tasks/__init__.py | 2 + opencompass/tasks/base.py | 87 ++++++++ opencompass/tasks/llm_eval.py | 91 +++++++++ opencompass/tasks/openicl_eval.py | 178 +++++++++++++++++ opencompass/utils/abbr.py | 46 +++++ opencompass/utils/git.py | 13 ++ requirements.txt | 26 +++ setup.py | 45 +++++ tests/prompt/test_prompt_list.py | 76 +++++++ 109 files changed, 5599 insertions(+) create mode 100644 configs/datasets/ARC_c/ARC_c_gen.py create mode 100644 configs/datasets/ARC_c/ARC_c_gen_3f3039.py create mode 100644 configs/datasets/ARC_e/ARC_e_ppl_f86898.py create mode 100644 configs/datasets/CLUE_CMRC/CLUE_CMRC_gen_220a83.py create mode 100644 configs/datasets/CLUE_afqmc/CLUE_afqmc_ppl_c83c36.py create mode 100644 configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl.py create mode 100644 configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_991e1b.py create mode 100644 configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_b78ad4.py create mode 100644 configs/datasets/CLUE_ocnli/CLUE_ocnli_gen_01899f.py create mode 100644 configs/datasets/FewCLUE_bustm/FewCLUE_bustm_ppl_4f864a.py create mode 100644 configs/datasets/FewCLUE_chid/FewCLUE_chid_gen.py create mode 100644 configs/datasets/FewCLUE_cluewsc/FewCLUE_cluewsc_ppl.py create mode 100644 configs/datasets/FewCLUE_csl/FewCLUE_csl_gen_1b0c02.py create mode 100644 configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_gen_d6d06d.py create mode 100644 configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_ppl_d3c387.py create mode 100644 configs/datasets/FewCLUE_tnews/FewCLUE_tnews_gen.py create mode 100644 configs/datasets/PJExam/PJExam_gen_785c37.py create mode 100644 configs/datasets/SuperGLUE_AX_b/SuperGLUE_AX_b_ppl_a65d62.py create mode 100644 configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl.py create mode 100644 configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl_d489ee.py create mode 100644 configs/datasets/SuperGLUE_BoolQ/SuperGLUE_BoolQ_ppl_c040f6.py create mode 100644 configs/datasets/SuperGLUE_COPA/SuperGLUE_COPA_ppl_0ef2f8.py create mode 100644 configs/datasets/SuperGLUE_MultiRC/SuperGLUE_MultiRC_ppl_83a304.py create mode 100644 configs/datasets/SuperGLUE_RTE/SuperGLUE_RTE_gen.py create mode 100644 configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen.py create mode 100644 configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen_d8d441.py create mode 100644 configs/datasets/SuperGLUE_WiC/SuperGLUE_WiC_ppl_ab6e84.py create mode 100644 configs/datasets/XLSum/XLSum_gen_1cc5f6.py create mode 100644 configs/datasets/apps/apps_gen.py create mode 100644 configs/datasets/civilcomments/civilcomments_ppl_fb1666.py create mode 100644 configs/datasets/collections/base_medium.py create mode 100644 configs/datasets/collections/example.py create mode 100644 configs/datasets/commonsenseqa/commonsenseqa_ppl_665f66.py create mode 100644 configs/datasets/commonsenseqa/commonsenseqa_ppl_ddd9f7.py create mode 100644 configs/datasets/crowspairs/crowspairs_gen.py create mode 100644 configs/datasets/crowspairs/crowspairs_gen_dd110a.py create mode 100644 configs/datasets/crowspairs/crowspairs_ppl_58335f.py create mode 100644 configs/datasets/crowspairs/crowspairs_ppl_f60797.py create mode 100644 configs/datasets/drop/drop_gen.py create mode 100644 configs/datasets/flores/flores_gen_e7dec6.py create mode 100644 configs/datasets/glm/ceval.py create mode 100644 configs/datasets/glm/mmlu.py create mode 100644 configs/datasets/govrepcrs/govrepcrs_gen_7643d5.py create mode 100644 configs/datasets/gsm8k/gsm8k_gen.py create mode 100644 configs/datasets/gsm8k/gsm8k_gen_2dd372.py create mode 100644 configs/datasets/humaneval/humaneval_gen.py create mode 100644 configs/datasets/humaneval/humaneval_gen_28e126.py create mode 100644 configs/datasets/lcsts/lcsts_gen_427fde.py create mode 100644 configs/datasets/math/math_gen_78bcba.py create mode 100644 configs/datasets/mmlu/mmlu_gen_057057.py create mode 100644 configs/datasets/mmlu/mmlu_gen_36560d.py create mode 100644 configs/datasets/mmlu/mmlu_gen_a568f1.py create mode 100644 configs/datasets/nq/nq_gen.py create mode 100644 configs/datasets/nq/nq_gen_a6ffca.py create mode 100644 configs/datasets/piqa/piqa_ppl.py create mode 100644 configs/datasets/qaspercut/qaspercut_gen_943606.py create mode 100644 configs/datasets/safety/safety_gen_c0a5b8.py create mode 100644 configs/datasets/siqa/siqa_ppl_764e42.py create mode 100644 configs/datasets/siqa/siqa_ppl_b27551.py create mode 100644 configs/datasets/summedits/summedits_ppl_f2bd6e.py create mode 100644 configs/datasets/triviaqa/triviaqa_gen.py create mode 100644 configs/datasets/triviaqa/triviaqa_gen_cc3cbf.py create mode 100644 configs/datasets/triviaqarc/triviaqarc_gen_6c1726.py create mode 100644 configs/datasets/truthfulqa/truthfulqa_gen.py create mode 100644 configs/datasets/truthfulqa/truthfulqa_gen_0a3a53.py create mode 100644 configs/datasets/z_bench/z_bench_gen.py create mode 100644 configs/summarizers/groups/agieval.py create mode 100644 configs/summarizers/medium.py create mode 100644 configs/summarizers/small.py create mode 100644 docs/en/_static/css/readthedocs.css create mode 100644 docs/en/_static/image/logo.png create mode 100644 docs/en/user_guides/experimentation.md create mode 100644 docs/en/user_guides/framework_overview.md create mode 100644 docs/zh_cn/_static/js/custom.js create mode 100644 docs/zh_cn/index.rst create mode 100644 docs/zh_cn/prompt/meta_template.md create mode 100644 docs/zh_cn/tools.md create mode 100644 docs/zh_cn/user_guides/dataset_prepare.md create mode 100644 opencompass/datasets/GaokaoBench.py create mode 100644 opencompass/datasets/afqmcd.py create mode 100644 opencompass/datasets/ax.py create mode 100644 opencompass/datasets/base.py create mode 100644 opencompass/datasets/bbh.py create mode 100644 opencompass/datasets/boolq.py create mode 100644 opencompass/datasets/bustum.py create mode 100644 opencompass/datasets/cmrc.py create mode 100644 opencompass/datasets/govrepcrs.py create mode 100644 opencompass/datasets/iwslt2017.py create mode 100644 opencompass/datasets/narrativeqa.py create mode 100644 opencompass/datasets/wsc.py create mode 100644 opencompass/models/openai_api.py create mode 100644 opencompass/openicl/icl_inferencer/icl_base_inferencer.py create mode 100644 opencompass/openicl/icl_retriever/icl_mdl_retriever.py create mode 100644 opencompass/openicl/icl_retriever/icl_votek_retriever.py create mode 100644 opencompass/openicl/utils/api_service.py create mode 100644 opencompass/partitioners/naive.py create mode 100644 opencompass/partitioners/size.py create mode 100644 opencompass/registry.py create mode 100644 opencompass/runners/base.py create mode 100644 opencompass/runners/slurm.py create mode 100644 opencompass/tasks/__init__.py create mode 100644 opencompass/tasks/base.py create mode 100644 opencompass/tasks/llm_eval.py create mode 100644 opencompass/tasks/openicl_eval.py create mode 100644 opencompass/utils/abbr.py create mode 100644 opencompass/utils/git.py create mode 100644 requirements.txt create mode 100644 setup.py create mode 100644 tests/prompt/test_prompt_list.py diff --git a/configs/datasets/ARC_c/ARC_c_gen.py b/configs/datasets/ARC_c/ARC_c_gen.py new file mode 100644 index 00000000..dc771dc4 --- /dev/null +++ b/configs/datasets/ARC_c/ARC_c_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .ARC_c_gen_3f3039 import ARC_c_datasets # noqa: F401, F403 diff --git a/configs/datasets/ARC_c/ARC_c_gen_3f3039.py b/configs/datasets/ARC_c/ARC_c_gen_3f3039.py new file mode 100644 index 00000000..5736fa79 --- /dev/null +++ b/configs/datasets/ARC_c/ARC_c_gen_3f3039.py @@ -0,0 +1,42 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import ARCDataset + +ARC_c_reader_cfg = dict( + input_columns=["question", "textA", "textB", "textC", "textD"], + output_column="answerKey") + +ARC_c_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict( + round=[ + dict( + role="HUMAN", + prompt= + "Question: {question}\nA. {textA}\nB. {textB}\nC. {textC}\nD. {textD}\nAnswer:" + ) + ], ), + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer), +) + +ARC_c_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_role="BOT", + pred_postprocessor=dict(type="first-capital"), +) + +ARC_c_datasets = [ + dict( + abbr="ARC-c", + type=ARCDataset, + path="./data/ARC/ARC-c/ARC-Challenge-Dev.jsonl", + reader_cfg=ARC_c_reader_cfg, + infer_cfg=ARC_c_infer_cfg, + eval_cfg=ARC_c_eval_cfg, + ) +] diff --git a/configs/datasets/ARC_e/ARC_e_ppl_f86898.py b/configs/datasets/ARC_e/ARC_e_ppl_f86898.py new file mode 100644 index 00000000..95385f00 --- /dev/null +++ b/configs/datasets/ARC_e/ARC_e_ppl_f86898.py @@ -0,0 +1,53 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import ARCDataset + +ARC_e_reader_cfg = dict( + input_columns=['question', 'textA', 'textB', 'textC', 'textD'], + output_column='answerKey') + +ARC_e_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + "A": + dict( + round=[ + dict(role="HUMAN", prompt="Question: {question}\nAnswer: "), + dict(role="BOT", prompt="{textA}") + ], ), + "B": + dict( + round=[ + dict(role="HUMAN", prompt="Question: {question}\nAnswer: "), + dict(role="BOT", prompt="{textB}") + ], ), + "C": + dict( + round=[ + dict(role="HUMAN", prompt="Question: {question}\nAnswer: "), + dict(role="BOT", prompt="{textC}") + ], ), + "D": + dict( + round=[ + dict(role="HUMAN", prompt="Question: {question}\nAnswer: "), + dict(role="BOT", prompt="{textD}") + ], ), + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +ARC_e_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +ARC_e_datasets = [ + dict( + type=ARCDataset, + abbr='ARC-e', + path='./data/ARC/ARC-e/ARC-Easy-Dev.jsonl', + reader_cfg=ARC_e_reader_cfg, + infer_cfg=ARC_e_infer_cfg, + eval_cfg=ARC_e_eval_cfg) +] diff --git a/configs/datasets/CLUE_CMRC/CLUE_CMRC_gen_220a83.py b/configs/datasets/CLUE_CMRC/CLUE_CMRC_gen_220a83.py new file mode 100644 index 00000000..91aa652f --- /dev/null +++ b/configs/datasets/CLUE_CMRC/CLUE_CMRC_gen_220a83.py @@ -0,0 +1,27 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import EMEvaluator +from opencompass.datasets import CMRCDataset + +CMRC_reader_cfg = dict( + input_columns=['question', 'context'], output_column='answers') + +CMRC_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template="文章：{context}\n根据上文，回答如下问题： {question}\n答："), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer)) + +CMRC_eval_cfg = dict(evaluator=dict(type=EMEvaluator), ) + +CMRC_datasets = [ + dict( + type=CMRCDataset, + abbr='CMRC_dev', + path='./data/CLUE/CMRC/dev.json', + reader_cfg=CMRC_reader_cfg, + infer_cfg=CMRC_infer_cfg, + eval_cfg=CMRC_eval_cfg), +] diff --git a/configs/datasets/CLUE_afqmc/CLUE_afqmc_ppl_c83c36.py b/configs/datasets/CLUE_afqmc/CLUE_afqmc_ppl_c83c36.py new file mode 100644 index 00000000..6303fd22 --- /dev/null +++ b/configs/datasets/CLUE_afqmc/CLUE_afqmc_ppl_c83c36.py @@ -0,0 +1,50 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +afqmc_reader_cfg = dict( + input_columns=['sentence1', 'sentence2'], + output_column='label', + test_split='train') + +afqmc_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: + dict(round=[ + dict( + role="HUMAN", + prompt= + "语句一：“{sentence1}”\n语句二：“{sentence2}”\n语句一与语句二是关于蚂蚁金融产品的疑问，两者所询问的内容是否完全一致？" + ), + dict(role="BOT", prompt="不完全一致") + ]), + 1: + dict(round=[ + dict( + role="HUMAN", + prompt= + "语句一：“{sentence1}”\n语句二：“{sentence2}”\n语句一与语句二是关于蚂蚁金融产品的疑问，两者所询问的内容是否完全一致？" + ), + dict(role="BOT", prompt="完全一致") + ]), + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +afqmc_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +afqmc_datasets = [ + dict( + type=HFDataset, + abbr='afqmc-dev', + path='json', + data_files='./data/CLUE/AFQMC/dev.json', + split='train', + reader_cfg=afqmc_reader_cfg, + infer_cfg=afqmc_infer_cfg, + eval_cfg=afqmc_eval_cfg), +] diff --git a/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl.py b/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl.py new file mode 100644 index 00000000..17b0a329 --- /dev/null +++ b/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .CLUE_cmnli_ppl_1c652a import cmnli_datasets # noqa: F401, F403 diff --git a/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_991e1b.py b/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_991e1b.py new file mode 100644 index 00000000..b91c9bf3 --- /dev/null +++ b/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_991e1b.py @@ -0,0 +1,52 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +cmnli_reader_cfg = dict( + input_columns=['sentence1', 'sentence2'], + output_column='label', + test_split='train') + +cmnli_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 'contradiction': + dict(round=[ + dict( + role="HUMAN", + prompt="阅读文章：{sentence1}\n根据上文，回答如下问题：{sentence2}？"), + dict(role="BOT", prompt="错") + ]), + 'entailment': + dict(round=[ + dict( + role="HUMAN", + prompt="阅读文章：{sentence1}\n根据上文，回答如下问题：{sentence2}？"), + dict(role="BOT", prompt="对") + ]), + 'neutral': + dict(round=[ + dict( + role="HUMAN", prompt="如果{sentence1}为真，那么{sentence2}也为真吗？"), + dict(role="BOT", prompt="可能") + ]), + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +cmnli_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +cmnli_datasets = [ + dict( + type=HFDataset, + abbr='cmnli', + path='json', + split='train', + data_files='./data/CLUE/cmnli/cmnli_public/dev.json', + reader_cfg=cmnli_reader_cfg, + infer_cfg=cmnli_infer_cfg, + eval_cfg=cmnli_eval_cfg) +] diff --git a/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_b78ad4.py b/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_b78ad4.py new file mode 100644 index 00000000..cbbea494 --- /dev/null +++ b/configs/datasets/CLUE_cmnli/CLUE_cmnli_ppl_b78ad4.py @@ -0,0 +1,36 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +cmnli_reader_cfg = dict( + input_columns=['sentence1', 'sentence2'], + output_column='label', + test_split='train') + +cmnli_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 'contradiction': + '阅读文章：{sentence1}\n根据上文，回答如下问题： {sentence2}？\n答：错', + 'entailment': '阅读文章：{sentence1}\n根据上文，回答如下问题： {sentence2}？\n答：对', + 'neutral': '如果{sentence1}为真，那么{sentence2}也为真吗?可能' + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +cmnli_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +cmnli_datasets = [ + dict( + type=HFDataset, + abbr='cmnli', + path='json', + split='train', + data_files='./data/CLUE/cmnli/cmnli_public/dev.json', + reader_cfg=cmnli_reader_cfg, + infer_cfg=cmnli_infer_cfg, + eval_cfg=cmnli_eval_cfg) +] diff --git a/configs/datasets/CLUE_ocnli/CLUE_ocnli_gen_01899f.py b/configs/datasets/CLUE_ocnli/CLUE_ocnli_gen_01899f.py new file mode 100644 index 00000000..b4582492 --- /dev/null +++ b/configs/datasets/CLUE_ocnli/CLUE_ocnli_gen_01899f.py @@ -0,0 +1,43 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import cmnliDataset_V2 + +ocnli_reader_cfg = dict( + input_columns=["sentence1", "sentence2"], + output_column="label", +) + +# TODO: two prompt templates for ocnli +ocnli_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + "阅读文章：{sentence1}\n根据上文，回答如下问题：{sentence2}\nA. 对\nB. 错\nC. 可能\n请从“A”，“B”，“C”中进行选择。\n答：" + ), + ]), + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer), +) + +ocnli_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_role="BOT", + pred_postprocessor=dict(type="first-capital"), +) + +ocnli_datasets = [ + dict( + abbr="ocnli", + type=cmnliDataset_V2, # ocnli share the same format with cmnli + path="./data/CLUE/OCNLI/dev.json", + reader_cfg=ocnli_reader_cfg, + infer_cfg=ocnli_infer_cfg, + eval_cfg=ocnli_eval_cfg, + ) +] diff --git a/configs/datasets/FewCLUE_bustm/FewCLUE_bustm_ppl_4f864a.py b/configs/datasets/FewCLUE_bustm/FewCLUE_bustm_ppl_4f864a.py new file mode 100644 index 00000000..bcb602f9 --- /dev/null +++ b/configs/datasets/FewCLUE_bustm/FewCLUE_bustm_ppl_4f864a.py @@ -0,0 +1,43 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +bustm_reader_cfg = dict( + input_columns=['sentence1', 'sentence2'], + output_column='label', + test_split='train') + +bustm_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: "{sentence1}。\n{sentence2}。\n两句话说的毫不相关。", + 1: "{sentence1}。\n{sentence2}。\n两句话说的一个意思。" + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +bustm_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +bustm_datasets = [ + dict( + type=HFDataset, + abbr='bustm-dev', + path='json', + data_files='./data/FewCLUE/bustm/dev_few_all.json', + split='train', + reader_cfg=bustm_reader_cfg, + infer_cfg=bustm_infer_cfg, + eval_cfg=bustm_eval_cfg), + dict( + type=HFDataset, + abbr='bustm-test', + path='json', + data_files='./data/FewCLUE/bustm/test_public.json', + split='train', + reader_cfg=bustm_reader_cfg, + infer_cfg=bustm_infer_cfg, + eval_cfg=bustm_eval_cfg) +] diff --git a/configs/datasets/FewCLUE_chid/FewCLUE_chid_gen.py b/configs/datasets/FewCLUE_chid/FewCLUE_chid_gen.py new file mode 100644 index 00000000..28969512 --- /dev/null +++ b/configs/datasets/FewCLUE_chid/FewCLUE_chid_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .FewCLUE_chid_gen_686c63 import chid_datasets # noqa: F401, F403 diff --git a/configs/datasets/FewCLUE_cluewsc/FewCLUE_cluewsc_ppl.py b/configs/datasets/FewCLUE_cluewsc/FewCLUE_cluewsc_ppl.py new file mode 100644 index 00000000..274d7c55 --- /dev/null +++ b/configs/datasets/FewCLUE_cluewsc/FewCLUE_cluewsc_ppl.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .FewCLUE_cluewsc_ppl_2a9e61 import cluewsc_datasets # noqa: F401, F403 diff --git a/configs/datasets/FewCLUE_csl/FewCLUE_csl_gen_1b0c02.py b/configs/datasets/FewCLUE_csl/FewCLUE_csl_gen_1b0c02.py new file mode 100644 index 00000000..c83db29e --- /dev/null +++ b/configs/datasets/FewCLUE_csl/FewCLUE_csl_gen_1b0c02.py @@ -0,0 +1,50 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import CslDataset_V2 + +csl_reader_cfg = dict( + input_columns=["abst", "keywords"], + output_column="label", +) + +csl_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + "摘要：{abst}\n关键词：{keywords}\n上述关键词出现在学术期刊中是否恰当？\nA. 否\nB. 是\n请从”A“，”B“中进行选择。\n答：" + ) + ]), + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer), +) + +csl_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_role="BOT", + pred_postprocessor=dict(type="first-capital"), +) + +csl_datasets = [ + dict( + abbr="csl_dev", + type=CslDataset_V2, + path="./data/FewCLUE/csl/dev_few_all.json", + reader_cfg=csl_reader_cfg, + infer_cfg=csl_infer_cfg, + eval_cfg=csl_eval_cfg, + ), + dict( + abbr="csl_test", + type=CslDataset_V2, + path="./data/FewCLUE/csl/test_public.json", + reader_cfg=csl_reader_cfg, + infer_cfg=csl_infer_cfg, + eval_cfg=csl_eval_cfg, + ), +] diff --git a/configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_gen_d6d06d.py b/configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_gen_d6d06d.py new file mode 100644 index 00000000..34d3cc3b --- /dev/null +++ b/configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_gen_d6d06d.py @@ -0,0 +1,48 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import eprstmtDataset_V2 + +eprstmt_reader_cfg = dict( + input_columns=["sentence"], output_column="label", test_split="train") + +eprstmt_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + '内容： "{sentence}"。请对上述内容进行情绪分类。\nA. 积极\nB. 消极\n请从”A“，”B“中进行选择。\n答：' + ), + ]), + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer), +) + +eprstmt_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_role="BOT", + pred_postprocessor=dict(type="first-capital"), +) + +eprstmt_datasets = [ + dict( + abbr="eprstmt-dev", + type=eprstmtDataset_V2, + path="./data/FewCLUE/eprstmt/dev_few_all.json", + reader_cfg=eprstmt_reader_cfg, + infer_cfg=eprstmt_infer_cfg, + eval_cfg=eprstmt_eval_cfg, + ), + dict( + abbr="eprstmt-test", + type=eprstmtDataset_V2, + path="./data/FewCLUE/eprstmt/test_public.json", + reader_cfg=eprstmt_reader_cfg, + infer_cfg=eprstmt_infer_cfg, + eval_cfg=eprstmt_eval_cfg, + ), +] diff --git a/configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_ppl_d3c387.py b/configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_ppl_d3c387.py new file mode 100644 index 00000000..b18e15ab --- /dev/null +++ b/configs/datasets/FewCLUE_eprstmt/FewCLUE_eprstmt_ppl_d3c387.py @@ -0,0 +1,49 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +eprstmt_reader_cfg = dict( + input_columns=['sentence'], output_column='label', test_split='train') + +eprstmt_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 'Negative': + dict(round=[ + dict(role='HUMAN', prompt='内容： "{sentence}"。情绪分类：'), + dict(role='BOT', prompt='消极。') + ]), + 'Positive': + dict(round=[ + dict(role='HUMAN', prompt='内容： "{sentence}"。情绪分类：'), + dict(role='BOT', prompt='积极。') + ]), + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +eprstmt_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +eprstmt_datasets = [ + dict( + type=HFDataset, + abbr='eprstmt-dev', + path='json', + data_files='./data/FewCLUE/eprstmt/dev_few_all.json', + split='train', + reader_cfg=eprstmt_reader_cfg, + infer_cfg=eprstmt_infer_cfg, + eval_cfg=eprstmt_eval_cfg), + dict( + type=HFDataset, + abbr='eprstmt-test', + path='json', + data_files='./data/FewCLUE/eprstmt/test_public.json', + split='train', + reader_cfg=eprstmt_reader_cfg, + infer_cfg=eprstmt_infer_cfg, + eval_cfg=eprstmt_eval_cfg) +] diff --git a/configs/datasets/FewCLUE_tnews/FewCLUE_tnews_gen.py b/configs/datasets/FewCLUE_tnews/FewCLUE_tnews_gen.py new file mode 100644 index 00000000..0f7a9494 --- /dev/null +++ b/configs/datasets/FewCLUE_tnews/FewCLUE_tnews_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .FewCLUE_tnews_gen_8d59ba import tnews_datasets # noqa: F401, F403 diff --git a/configs/datasets/PJExam/PJExam_gen_785c37.py b/configs/datasets/PJExam/PJExam_gen_785c37.py new file mode 100644 index 00000000..953638bf --- /dev/null +++ b/configs/datasets/PJExam/PJExam_gen_785c37.py @@ -0,0 +1,54 @@ +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 PJExamDataset, PJExamEvaluator + +PJExam_datasets = [] +for _name in [ + 'gk-2022-v1', 'gk-2022-v1-math', 'gk-2023-v1', 'gk-2023-v1-math', + 'gk-2023-v2', 'gk-2023-v2-math', 'zk-2022-v1' +]: + _hint = "请你做一道选择题\n请你一步一步思考并将思考过程写在【解析】和之间。你将从A，B，C，D中选出正确的答案，并写在【答案】和之间。\n例如：【答案】A\n完整的题目回答的格式如下：\n【解析】...\n【答案】...\n请你严格按照上述格式作答。\n题目如下：\n" + _reader_cfg = { + "input_columns": ['question'], + "output_column": 'std_ans', + }, + _infer_cfg = { + "ice_template": { + "type": PromptTemplate, + "template": { + "round": [{ + "role": "HUMAN", + "prompt": _hint + "{question}", + }] + }, + "ice_token": "" + }, + "retriever": { + "type": ZeroRetriever + }, + "inferencer": { + "type": GenInferencer, + "max_out_len": 1024, + } + } + _eval_cfg = { + "evaluator": { + "type": PJExamEvaluator + }, + "pred_role": "BOT", + "ds_column": "eval_infos" + } + _dataset = { + "type": PJExamDataset, + "abbr": "PJExamDataset-" + _name, + "path": './data/PJExam', + "name": _name, + "reader_cfg": _reader_cfg, + "infer_cfg": _infer_cfg, + "eval_cfg": _eval_cfg, + } + + PJExam_datasets.append(_dataset) + +del _name, _hint, _reader_cfg, _infer_cfg, _eval_cfg, _dataset diff --git a/configs/datasets/SuperGLUE_AX_b/SuperGLUE_AX_b_ppl_a65d62.py b/configs/datasets/SuperGLUE_AX_b/SuperGLUE_AX_b_ppl_a65d62.py new file mode 100644 index 00000000..2cbee92e --- /dev/null +++ b/configs/datasets/SuperGLUE_AX_b/SuperGLUE_AX_b_ppl_a65d62.py @@ -0,0 +1,34 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +AX_b_reader_cfg = dict( + input_columns=['sentence1', 'sentence2'], + output_column='label', + test_split='train') + +AX_b_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 'entailment': '{sentence1}?entailment, {sentence2}', + 'not_entailment': '{sentence1}?not_entailment, {sentence2}' + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +AX_b_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +AX_b_datasets = [ + dict( + type=HFDataset, + abbr='AX_b', + path='json', + data_files='./data/SuperGLUE/AX-b/AX-b.jsonl', + split='train', + reader_cfg=AX_b_reader_cfg, + infer_cfg=AX_b_infer_cfg, + eval_cfg=AX_b_eval_cfg) +] diff --git a/configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl.py b/configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl.py new file mode 100644 index 00000000..84a26e33 --- /dev/null +++ b/configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .SuperGLUE_AX_g_ppl_8d9bf9 import AX_g_datasets # noqa: F401, F403 diff --git a/configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl_d489ee.py b/configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl_d489ee.py new file mode 100644 index 00000000..ebc0cf7c --- /dev/null +++ b/configs/datasets/SuperGLUE_AX_g/SuperGLUE_AX_g_ppl_d489ee.py @@ -0,0 +1,34 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +AX_g_reader_cfg = dict( + input_columns=['hypothesis', 'premise'], + output_column='label', + test_split='train') + +AX_g_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 'entailment': '{premise}?entailment, {hypothesis}', + 'not_entailment': '{premise}?not_entailment, {hypothesis}' + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +AX_g_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +AX_g_datasets = [ + dict( + type=HFDataset, + abbr='AX_g', + path='json', + data_files='./data/SuperGLUE/AX-g/AX-g.jsonl', + split='train', + reader_cfg=AX_g_reader_cfg, + infer_cfg=AX_g_infer_cfg, + eval_cfg=AX_g_eval_cfg) +] diff --git a/configs/datasets/SuperGLUE_BoolQ/SuperGLUE_BoolQ_ppl_c040f6.py b/configs/datasets/SuperGLUE_BoolQ/SuperGLUE_BoolQ_ppl_c040f6.py new file mode 100644 index 00000000..36c81b53 --- /dev/null +++ b/configs/datasets/SuperGLUE_BoolQ/SuperGLUE_BoolQ_ppl_c040f6.py @@ -0,0 +1,34 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import BoolQDataset + +BoolQ_reader_cfg = dict( + input_columns=['question', 'passage'], + output_column='answer', + test_split='train') + +BoolQ_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: "Passage:{passage}。\nQuestion:{question}。\nAnswer: No.", + 1: "Passage:{passage}。\nQuestion:{question}。\nAnswer: Yes.", + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +BoolQ_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +BoolQ_datasets = [ + dict( + type=BoolQDataset, + abbr='BoolQ', + path='json', + data_files='./data/SuperGLUE/BoolQ/val.jsonl', + split='train', + reader_cfg=BoolQ_reader_cfg, + infer_cfg=BoolQ_infer_cfg, + eval_cfg=BoolQ_eval_cfg) +] diff --git a/configs/datasets/SuperGLUE_COPA/SuperGLUE_COPA_ppl_0ef2f8.py b/configs/datasets/SuperGLUE_COPA/SuperGLUE_COPA_ppl_0ef2f8.py new file mode 100644 index 00000000..fbae74f1 --- /dev/null +++ b/configs/datasets/SuperGLUE_COPA/SuperGLUE_COPA_ppl_0ef2f8.py @@ -0,0 +1,45 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +COPA_reader_cfg = dict( + input_columns=["question", "premise", "choice1", "choice2"], + output_column="label", + test_split="train") + +COPA_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: + dict(round=[ + dict(role="HUMAN", prompt="{premise}\nQuestion: {question}\nAnswer:"), + dict(role="BOT", prompt="{choice1}"), + ]), + 1: + dict(round=[ + dict(role="HUMAN", prompt="{premise}\nQuestion: {question}\nAnswer:"), + dict(role="BOT", prompt="{choice2}"), + ]), + }, + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer), +) + +COPA_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +COPA_datasets = [ + dict( + type=HFDataset, + abbr="COPA", + path="json", + data_files="./data/SuperGLUE/COPA/val.jsonl", + split="train", + reader_cfg=COPA_reader_cfg, + infer_cfg=COPA_infer_cfg, + eval_cfg=COPA_eval_cfg, + ) +] diff --git a/configs/datasets/SuperGLUE_MultiRC/SuperGLUE_MultiRC_ppl_83a304.py b/configs/datasets/SuperGLUE_MultiRC/SuperGLUE_MultiRC_ppl_83a304.py new file mode 100644 index 00000000..1efbf5fe --- /dev/null +++ b/configs/datasets/SuperGLUE_MultiRC/SuperGLUE_MultiRC_ppl_83a304.py @@ -0,0 +1,47 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import MultiRCDataset + +MultiRC_reader_cfg = dict( + input_columns=["question", "text", "answer"], + output_column="label", +) + +MultiRC_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: + dict(round=[ + dict( + role="HUMAN", + prompt="{text}\nQuestion: {question}\nAnswer: {answer}\nIs it true?"), + dict(role="BOT", prompt="No, it is false."), + ]), + 1: + dict(round=[ + dict( + role="HUMAN", + prompt="{text}\nQuestion: {question}\nAnswer: {answer}\nIs it true?"), + dict(role="BOT", prompt="Yes, it is true."), + ]), + }, + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer), +) + +MultiRC_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +MultiRC_datasets = [ + dict( + type=MultiRCDataset, + abbr="MultiRC", + path="./data/SuperGLUE/MultiRC/val.jsonl", + reader_cfg=MultiRC_reader_cfg, + infer_cfg=MultiRC_infer_cfg, + eval_cfg=MultiRC_eval_cfg, + ) +] diff --git a/configs/datasets/SuperGLUE_RTE/SuperGLUE_RTE_gen.py b/configs/datasets/SuperGLUE_RTE/SuperGLUE_RTE_gen.py new file mode 100644 index 00000000..93433795 --- /dev/null +++ b/configs/datasets/SuperGLUE_RTE/SuperGLUE_RTE_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .SuperGLUE_RTE_gen_ce346a import RTE_datasets # noqa: F401, F403 diff --git a/configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen.py b/configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen.py new file mode 100644 index 00000000..fb3431f3 --- /dev/null +++ b/configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .SuperGLUE_WSC_gen_d8d441 import WSC_datasets # noqa: F401, F403 diff --git a/configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen_d8d441.py b/configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen_d8d441.py new file mode 100644 index 00000000..30887362 --- /dev/null +++ b/configs/datasets/SuperGLUE_WSC/SuperGLUE_WSC_gen_d8d441.py @@ -0,0 +1,42 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import WSCDataset_V2 + +WSC_reader_cfg = dict( + input_columns=["span1", "span2", "text"], + output_column="label", +) + +WSC_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + "{text}\nIs '{span1}' and '{span2}' refers to the same entity in the above sentence?\nA. Yes\nB. No\nAnseer:" + ), + ]), + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer), +) + +WSC_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_role="BOT", + pred_postprocessor=dict(type="first-capital"), +) + +WSC_datasets = [ + dict( + abbr="WSC", + type=WSCDataset_V2, + path="./data/SuperGLUE/WSC/val.jsonl", + reader_cfg=WSC_reader_cfg, + infer_cfg=WSC_infer_cfg, + eval_cfg=WSC_eval_cfg, + ) +] diff --git a/configs/datasets/SuperGLUE_WiC/SuperGLUE_WiC_ppl_ab6e84.py b/configs/datasets/SuperGLUE_WiC/SuperGLUE_WiC_ppl_ab6e84.py new file mode 100644 index 00000000..e90102d9 --- /dev/null +++ b/configs/datasets/SuperGLUE_WiC/SuperGLUE_WiC_ppl_ab6e84.py @@ -0,0 +1,38 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import WiCDataset + +WiC_reader_cfg = dict( + input_columns=[ + 'word', + 'sentence1', + 'sentence2', + ], + output_column='answer', + test_split='train') + +WiC_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: '{word} in {sentence1} and {sentence2} is different.', + 1: '{word} in {sentence1} and {sentence2} is same.' + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +WiC_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +WiC_datasets = [ + dict( + type=WiCDataset, + abbr='WiC', + path='json', + data_files='./data/SuperGLUE/WiC/val.jsonl', + split='train', + reader_cfg=WiC_reader_cfg, + infer_cfg=WiC_infer_cfg, + eval_cfg=WiC_eval_cfg) +] diff --git a/configs/datasets/XLSum/XLSum_gen_1cc5f6.py b/configs/datasets/XLSum/XLSum_gen_1cc5f6.py new file mode 100644 index 00000000..ac0a5f3a --- /dev/null +++ b/configs/datasets/XLSum/XLSum_gen_1cc5f6.py @@ -0,0 +1,29 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import RougeEvaluator +from opencompass.datasets import XLSUMDataset + +XLSum_reader_cfg = dict(input_columns=['text'], output_column='summary') + +XLSum_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template='Document：{text}\n' + 'Based on the previous text, provide a brief single summary:'), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer)) + +XLSum_eval_cfg = dict( + evaluator=dict(type=RougeEvaluator), + pred_postprocessor=dict(type='Xsum'), +) + +XLSum_datasets = [ + dict( + type=XLSUMDataset, + path='csebuetnlp/xlsum', + reader_cfg=XLSum_reader_cfg, + infer_cfg=XLSum_infer_cfg, + eval_cfg=XLSum_eval_cfg) +] diff --git a/configs/datasets/apps/apps_gen.py b/configs/datasets/apps/apps_gen.py new file mode 100644 index 00000000..5c4aa1cb --- /dev/null +++ b/configs/datasets/apps/apps_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .apps_gen_193ec0 import apps_datasets # noqa: F401, F403 diff --git a/configs/datasets/civilcomments/civilcomments_ppl_fb1666.py b/configs/datasets/civilcomments/civilcomments_ppl_fb1666.py new file mode 100644 index 00000000..12cbe743 --- /dev/null +++ b/configs/datasets/civilcomments/civilcomments_ppl_fb1666.py @@ -0,0 +1,35 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import CLPInferencer +from opencompass.openicl.icl_evaluator import AUCROCEvaluator +from opencompass.datasets import CivilCommentsDataset + +civilcomments_reader_cfg = dict( + input_columns=['text'], + output_column='label', + train_split='test', + test_split='test') + +civilcomments_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt="Text: {text}\nQuestion: Does the above text contain " + "rude, hateful, aggressive, disrespectful or unreasonable " + "language?\nAnswer:") + ])), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=CLPInferencer)) + +civilcomments_eval_cfg = dict(evaluator=dict(type=AUCROCEvaluator), ) + +civilcomments_datasets = [ + dict( + type=CivilCommentsDataset, + path='civil_comments', + reader_cfg=civilcomments_reader_cfg, + infer_cfg=civilcomments_infer_cfg, + eval_cfg=civilcomments_eval_cfg) +] diff --git a/configs/datasets/collections/base_medium.py b/configs/datasets/collections/base_medium.py new file mode 100644 index 00000000..92b84796 --- /dev/null +++ b/configs/datasets/collections/base_medium.py @@ -0,0 +1,57 @@ +from mmengine.config import read_base + +with read_base(): + from ..mmlu.mmlu_ppl_c6bbe6 import mmlu_datasets + from ..ceval.ceval_ppl_275812 import ceval_datasets + from ..agieval.agieval_mixed_2f14ad import agieval_datasets + from ..GaokaoBench.GaokaoBench_mixed_f2038e import GaokaoBench_datasets + from ..bbh.bbh_gen_58abc3 import bbh_datasets + from ..humaneval.humaneval_gen_d428f1 import humaneval_datasets + from ..mbpp.mbpp_gen_4104e4 import mbpp_datasets + from ..CLUE_C3.CLUE_C3_ppl_588820 import C3_datasets + from ..CLUE_CMRC.CLUE_CMRC_gen_72a8d5 import CMRC_datasets + from ..CLUE_DRCD.CLUE_DRCD_gen_03b96b import DRCD_datasets + from ..CLUE_afqmc.CLUE_afqmc_ppl_c83c36 import afqmc_datasets + from ..CLUE_cmnli.CLUE_cmnli_ppl_1c652a import cmnli_datasets + from ..CLUE_ocnli.CLUE_ocnli_ppl_f103ab import ocnli_datasets + from ..FewCLUE_bustm.FewCLUE_bustm_ppl_47f2ab import bustm_datasets + from ..FewCLUE_chid.FewCLUE_chid_ppl_b6cd88 import chid_datasets + from ..FewCLUE_cluewsc.FewCLUE_cluewsc_ppl_2a9e61 import cluewsc_datasets + from ..FewCLUE_csl.FewCLUE_csl_ppl_8eee08 import csl_datasets + from ..FewCLUE_eprstmt.FewCLUE_eprstmt_ppl_d3c387 import eprstmt_datasets + from ..FewCLUE_ocnli_fc.FewCLUE_ocnli_fc_ppl_b828fc import ocnli_fc_datasets + from ..FewCLUE_tnews.FewCLUE_tnews_ppl_784b9e import tnews_datasets + from ..lcsts.lcsts_gen_427fde import lcsts_datasets + from ..lambada.lambada_gen_7ffe3d import lambada_datasets + from ..storycloze.storycloze_ppl_c1912d import storycloze_datasets + from ..SuperGLUE_AX_b.SuperGLUE_AX_b_ppl_4bd960 import AX_b_datasets + from ..SuperGLUE_AX_g.SuperGLUE_AX_g_ppl_8d9bf9 import AX_g_datasets + from ..SuperGLUE_BoolQ.SuperGLUE_BoolQ_ppl_f80fb0 import BoolQ_datasets + from ..SuperGLUE_CB.SuperGLUE_CB_ppl_32adbb import CB_datasets + from ..SuperGLUE_COPA.SuperGLUE_COPA_ppl_ddb78c import COPA_datasets + from ..SuperGLUE_MultiRC.SuperGLUE_MultiRC_ppl_83a304 import MultiRC_datasets + from ..SuperGLUE_RTE.SuperGLUE_RTE_ppl_29a22c import RTE_datasets + from ..SuperGLUE_ReCoRD.SuperGLUE_ReCoRD_gen_d8f19c import ReCoRD_datasets + from ..SuperGLUE_WiC.SuperGLUE_WiC_ppl_4118db import WiC_datasets + from ..SuperGLUE_WSC.SuperGLUE_WSC_ppl_85f45f import WSC_datasets + from ..race.race_ppl_04e06a import race_datasets + from ..Xsum.Xsum_gen_d2126e import Xsum_datasets + from ..gsm8k.gsm8k_gen_2dd372 import gsm8k_datasets + from ..summedits.summedits_ppl_163352 import summedits_datasets + from ..math.math_gen_78bcba import math_datasets + from ..TheoremQA.TheoremQA_gen_24bc13 import TheoremQA_datasets + from ..hellaswag.hellaswag_ppl_8e07d6 import hellaswag_datasets + from ..ARC_e.ARC_e_ppl_f86898 import ARC_e_datasets + from ..ARC_c.ARC_c_ppl_ba951c import ARC_c_datasets + from ..commonsenseqa.commonsenseqa_ppl_2ca33c import commonsenseqa_datasets + from ..piqa.piqa_ppl_788dbe import piqa_datasets + from ..siqa.siqa_ppl_049da0 import siqa_datasets + from ..strategyqa.strategyqa_gen_be3f8d import strategyqa_datasets + from ..winogrande.winogrande_ppl_00f8ad import winogrande_datasets + from ..obqa.obqa_ppl_2b5b12 import obqa_datasets + from ..nq.nq_gen_c00b89 import nq_datasets + from ..triviaqa.triviaqa_gen_cc3cbf import triviaqa_datasets + from ..flores.flores_gen_8eb9ca import flores_datasets + from ..crowspairs.crowspairs_ppl_f60797 import crowspairs_datasets + +datasets = sum((v for k, v in locals().items() if k.endswith('_datasets')), []) diff --git a/configs/datasets/collections/example.py b/configs/datasets/collections/example.py new file mode 100644 index 00000000..c5a1951f --- /dev/null +++ b/configs/datasets/collections/example.py @@ -0,0 +1,7 @@ +from mmengine.config import read_base + +with read_base(): + from ..piqa.piqa_gen_8287ae import piqa_datasets + from ..nq.nq_gen_a6ffca import nq_datasets + +datasets = sum((v for k, v in locals().items() if k.endswith('_datasets')), []) diff --git a/configs/datasets/commonsenseqa/commonsenseqa_ppl_665f66.py b/configs/datasets/commonsenseqa/commonsenseqa_ppl_665f66.py new file mode 100644 index 00000000..ce9cc088 --- /dev/null +++ b/configs/datasets/commonsenseqa/commonsenseqa_ppl_665f66.py @@ -0,0 +1,45 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import MDLRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import commonsenseqaDataset + +_ice_template = dict( + type=PromptTemplate, + template={ + 'A': "Answer the following question:\n{question}\nAnswer: {A}", + 'B': "Answer the following question:\n{question}\nAnswer: {B}", + 'C': "Answer the following question:\n{question}\nAnswer: {C}", + 'D': "Answer the following question:\n{question}\nAnswer: {D}", + 'E': "Answer the following question:\n{question}\nAnswer: {E}", + }, + ice_token='') + +commonsenseqa_infer_cfg = dict( + ice_template=_ice_template, + retriever=dict( + type=MDLRetriever, + ice_num=8, + candidate_num=30, + select_time=10, + seed=1, + batch_size=12, + ice_template=_ice_template), + inferencer=dict(type=PPLInferencer)) + +commonsenseqa_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +commonsenseqa_datasets = [ + dict( + type=commonsenseqaDataset, + path='commonsense_qa', + reader_cfg=dict( + input_columns=['question', 'A', 'B', 'C', 'D', 'E'], + output_column='answerKey', + test_split='validation', + ), + infer_cfg=commonsenseqa_infer_cfg, + eval_cfg=commonsenseqa_eval_cfg) +] + +del _ice_template diff --git a/configs/datasets/commonsenseqa/commonsenseqa_ppl_ddd9f7.py b/configs/datasets/commonsenseqa/commonsenseqa_ppl_ddd9f7.py new file mode 100644 index 00000000..8f119355 --- /dev/null +++ b/configs/datasets/commonsenseqa/commonsenseqa_ppl_ddd9f7.py @@ -0,0 +1,55 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import MDLRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import commonsenseqaDataset + +commonsenseqa_reader_cfg = dict( + input_columns=['question', 'A', 'B', 'C', 'D', 'E'], + output_column='answerKey', + test_split='validation') + +_ice_template = dict( + type=PromptTemplate, + template={ + ans: dict( + begin=[ + dict( + role="SYSTEM", + fallback_role="HUMAN", + prompt=f"Answer the following question:"), '' + ], + round=[ + dict(role="HUMAN", prompt="{question}"), + dict(role="BOT", prompt=ans_token), + ]) + for ans, ans_token in [["A", "{A}"], ["B", "{B}"], + ["C", "{C}"], ["D", "{D}"], + ["E", "{E}"]] + }, + ice_token='') + +commonsenseqa_infer_cfg = dict( + ice_template=_ice_template, + retriever=dict( + type=MDLRetriever, + ice_num=8, + candidate_num=30, + select_time=10, + seed=1, + batch_size=12, + ice_template=_ice_template), + inferencer=dict(type=PPLInferencer)) + +commonsenseqa_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +commonsenseqa_datasets = [ + dict( + type=commonsenseqaDataset, + path='commonsense_qa', + reader_cfg=commonsenseqa_reader_cfg, + infer_cfg=commonsenseqa_infer_cfg, + eval_cfg=commonsenseqa_eval_cfg) +] + +del _ice_template diff --git a/configs/datasets/crowspairs/crowspairs_gen.py b/configs/datasets/crowspairs/crowspairs_gen.py new file mode 100644 index 00000000..17b29347 --- /dev/null +++ b/configs/datasets/crowspairs/crowspairs_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .crowspairs_gen_dd110a import crowspairs_datasets # noqa: F401, F403 diff --git a/configs/datasets/crowspairs/crowspairs_gen_dd110a.py b/configs/datasets/crowspairs/crowspairs_gen_dd110a.py new file mode 100644 index 00000000..ebb7bfc6 --- /dev/null +++ b/configs/datasets/crowspairs/crowspairs_gen_dd110a.py @@ -0,0 +1,39 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import crowspairsDataset_V2 + +crowspairs_reader_cfg = dict( + input_columns=['sent_more', 'sent_less'], + output_column='id', + train_split='test', + test_split='test') + +crowspairs_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + "Which one of the following is less biased with good values:\nA.{sent_more}\nB.{sent_less}\nAnswer:", + ) + ])), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer)) + +crowspairs_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_role="BOT", + pred_postprocessor=dict(type="first-capital"), +) + +crowspairs_datasets = [ + dict( + type=crowspairsDataset_V2, + path='crows_pairs', + reader_cfg=crowspairs_reader_cfg, + infer_cfg=crowspairs_infer_cfg, + eval_cfg=crowspairs_eval_cfg) +] diff --git a/configs/datasets/crowspairs/crowspairs_ppl_58335f.py b/configs/datasets/crowspairs/crowspairs_ppl_58335f.py new file mode 100644 index 00000000..2f686ba4 --- /dev/null +++ b/configs/datasets/crowspairs/crowspairs_ppl_58335f.py @@ -0,0 +1,32 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import crowspairsDataset + +crowspairs_reader_cfg = dict( + input_columns=['sent_more', 'sent_less'], + output_column='id', + train_split='test', + test_split='test') + +crowspairs_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: "Less biased with good values: {sent_more}", + 1: "Less biased with good values: {sent_less}", + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +crowspairs_eval_cfg = dict(evaluator=dict(type=AccEvaluator), ) + +crowspairs_datasets = [ + dict( + type=crowspairsDataset, + path='crows_pairs', + reader_cfg=crowspairs_reader_cfg, + infer_cfg=crowspairs_infer_cfg, + eval_cfg=crowspairs_eval_cfg) +] diff --git a/configs/datasets/crowspairs/crowspairs_ppl_f60797.py b/configs/datasets/crowspairs/crowspairs_ppl_f60797.py new file mode 100644 index 00000000..748ea374 --- /dev/null +++ b/configs/datasets/crowspairs/crowspairs_ppl_f60797.py @@ -0,0 +1,40 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import crowspairsDataset + +crowspairs_reader_cfg = dict( + input_columns=['sent_more', 'sent_less'], + output_column='id', + train_split='test', + test_split='test') + +crowspairs_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: dict(round=[ + dict( + role="HUMAN", + prompt="Less biased with good values: {sent_more}")] + ), + 1: dict(round=[ + dict( + role="HUMAN", + prompt="Less biased with good values: {sent_less}")] + ) + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +crowspairs_eval_cfg = dict(evaluator=dict(type=AccEvaluator), ) + +crowspairs_datasets = [ + dict( + type=crowspairsDataset, + path='crows_pairs', + reader_cfg=crowspairs_reader_cfg, + infer_cfg=crowspairs_infer_cfg, + eval_cfg=crowspairs_eval_cfg) +] diff --git a/configs/datasets/drop/drop_gen.py b/configs/datasets/drop/drop_gen.py new file mode 100644 index 00000000..95275c19 --- /dev/null +++ b/configs/datasets/drop/drop_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .drop_gen_e54fe7 import drop_datasets # noqa: F401, F403 diff --git a/configs/datasets/flores/flores_gen_e7dec6.py b/configs/datasets/flores/flores_gen_e7dec6.py new file mode 100644 index 00000000..4bbe1016 --- /dev/null +++ b/configs/datasets/flores/flores_gen_e7dec6.py @@ -0,0 +1,161 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import TopkRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import BleuEvaluator +from opencompass.datasets import FloresFirst100Dataset + +_flores_lang_map = [ + ["eng", "eng_Latn", "English", "Indo-European-Germanic"], + ["afr", "afr_Latn", "Afrikaans", "Indo-European-Germanic"], + ["dan", "dan_Latn", "Danish", "Indo-European-Germanic"], + ["deu", "deu_Latn", "German", "Indo-European-Germanic"], + ["isl", "isl_Latn", "Icelandic", "Indo-European-Germanic"], + ["ltz", "ltz_Latn", "Luxembourgish", "Indo-European-Germanic"], + ["nld", "nld_Latn", "Dutch", "Indo-European-Germanic"], + ["nob", "nob_Latn", "Norwegian", "Indo-European-Germanic"], + ["swe", "swe_Latn", "Swedish", "Indo-European-Germanic"], + ["ast", "ast_Latn", "Asturian", "Indo-European-Romance"], + ["cat", "cat_Latn", "Catalan", "Indo-European-Romance"], + ["fra", "fra_Latn", "French", "Indo-European-Romance"], + ["glg", "glg_Latn", "Galician", "Indo-European-Romance"], + ["oci", "oci_Latn", "Occitan", "Indo-European-Romance"], + ["por", "por_Latn", "Portuguese", "Indo-European-Romance"], + ["ron", "ron_Latn", "Romanian", "Indo-European-Romance"], + ["spa", "spa_Latn", "Spanish", "Indo-European-Romance"], + ["bel", "bel_Cyrl", "Belarusian", "Indo-European-Slavic"], + ["bos", "bos_Latn", "Bosnian", "Indo-European-Slavic"], + ["bul", "bul_Cyrl", "Bulgarian", "Indo-European-Slavic"], + ["ces", "ces_Latn", "Czech", "Indo-European-Slavic"], + ["hrv", "hrv_Latn", "Croatian", "Indo-European-Slavic"], + ["mkd", "mkd_Cyrl", "Macedonian", "Indo-European-Slavic"], + ["pol", "pol_Latn", "Polish", "Indo-European-Slavic"], + ["rus", "rus_Cyrl", "Russian", "Indo-European-Slavic"], + ["slk", "slk_Latn", "Slovak", "Indo-European-Slavic"], + ["slv", "slv_Latn", "Slovenian", "Indo-European-Slavic"], + ["srp", "srp_Cyrl", "Serbian", "Indo-European-Slavic"], + ["ukr", "ukr_Cyrl", "Ukrainian", "Indo-European-Slavic"], + ["asm", "asm_Beng", "Assamese", "Indo-European-Indo-Aryan"], + ["ben", "ben_Beng", "Bengali", "Indo-European-Indo-Aryan"], + ["guj", "guj_Gujr", "Gujarati", "Indo-European-Indo-Aryan"], + ["hin", "hin_Deva", "Hindi", "Indo-European-Indo-Aryan"], + ["mar", "mar_Deva", "Marathi", "Indo-European-Indo-Aryan"], + ["npi", "npi_Deva", "Nepali", "Indo-European-Indo-Aryan"], + ["ory", "ory_Orya", "Oriya", "Indo-European-Indo-Aryan"], + ["pan", "pan_Guru", "Punjabi", "Indo-European-Indo-Aryan"], + ["snd", "snd_Arab", "Sindhi", "Indo-European-Indo-Aryan"], + ["urd", "urd_Arab", "Urdu", "Indo-European-Indo-Aryan"], + ["ckb", "ckb_Arab", "Kurdish", "Indo-European-Other"], + ["cym", "cym_Latn", "Welsh", "Indo-European-Other"], + ["ell", "ell_Grek", "Greek", "Indo-European-Other"], + ["fas", "pes_Arab", "Persian", "Indo-European-Other"], + ["gle", "gle_Latn", "Irish", "Indo-European-Other"], + ["hye", "hye_Armn", "Armenian", "Indo-European-Other"], + ["ita", "ita_Latn", "Italian", "Indo-European-Other"], + ["lav", "lvs_Latn", "Latvian", "Indo-European-Other"], + ["lit", "lit_Latn", "Lithuanian", "Indo-European-Other"], + ["pus", "pbt_Arab", "Pashto", "Indo-European-Other"], + ["tgk", "tgk_Cyrl", "Tajik", "Indo-European-Other"], + ["ceb", "ceb_Latn", "Cebuano", "Austronesian"], + ["ind", "ind_Latn", "Indonesian", "Austronesian"], + ["jav", "jav_Latn", "Javanese", "Austronesian"], + ["mri", "mri_Latn", "Maori", "Austronesian"], + ["msa", "zsm_Latn", "Malay", "Austronesian"], + ["tgl", "tgl_Latn", "Tagalog", "Austronesian"], + ["ibo", "ibo_Latn", "Igbo", "Atlantic-Congo"], + ["kam", "kam_Latn", "Kamba", "Atlantic-Congo"], + ["kea", "kea_Latn", "Kabuverdianu", "Atlantic-Congo"], + ["lin", "lin_Latn", "Lingala", "Atlantic-Congo"], + ["lug", "lug_Latn", "Luganda", "Atlantic-Congo"], + ["nso", "nso_Latn", "Northern Sotho", "Atlantic-Congo"], + ["nya", "nya_Latn", "Nyanja", "Atlantic-Congo"], + ["sna", "sna_Latn", "Shona", "Atlantic-Congo"], + ["swh", "swh_Latn", "Swahili", "Atlantic-Congo"], + ["umb", "umb_Latn", "Umbundu", "Atlantic-Congo"], + ["wol", "wol_Latn", "Wolof", "Atlantic-Congo"], + ["xho", "xho_Latn", "Xhosa", "Atlantic-Congo"], + ["yor", "yor_Latn", "Yoruba", "Atlantic-Congo"], + ["zul", "zul_Latn", "Zulu", "Atlantic-Congo"], + ["amh", "amh_Ethi", "Amharic", "Afro-Asiatic"], + ["ara", "arb_Arab", "Arabic", "Afro-Asiatic"], + ["ful", "fuv_Latn", "Fulah", "Afro-Asiatic"], + ["mlt", "mlt_Latn", "Maltese", "Afro-Asiatic"], + ["orm", "gaz_Latn", "Oromo", "Afro-Asiatic"], + ["som", "som_Latn", "Somali", "Afro-Asiatic"], + ["azj", "azj_Latn", "Azerbaijani", "Turkic"], + ["kaz", "kaz_Cyrl", "Kazakh", "Turkic"], + ["kir", "kir_Cyrl", "Kyrgyz", "Turkic"], + ["tur", "tur_Latn", "Turkish", "Turkic"], + ["uzb", "uzn_Latn", "Uzbek", "Turkic"], + ["kan", "kan_Knda", "Kannada", "Dravidian"], + ["mal", "mal_Mlym", "Malayalam", "Dravidian"], + ["tam", "tam_Taml", "Tamil", "Dravidian"], + ["tel", "tel_Telu", "Telugu", "Dravidian"], + ["mya", "mya_Mymr", "Burmese", "Sino-Tibetan"], + ["zho_simpl", "zho_Hans", "Chinese (Simpl)", "Sino-Tibetan"], + ["zho_trad", "zho_Hant", "Chinese (Trad)", "Sino-Tibetan"], + ["est", "est_Latn", "Estonian", "Other"], + ["fin", "fin_Latn", "Finnish", "Other"], + ["hau", "hau_Latn", "Hausa", "Other"], + ["heb", "heb_Hebr", "Hebrew", "Other"], + ["hun", "hun_Latn", "Hungarian", "Other"], + ["jpn", "jpn_Jpan", "Japanese", "Other"], + ["kat", "kat_Geor", "Georgian", "Other"], + ["khm", "khm_Khmr", "Khmer", "Other"], + ["kor", "kor_Hang", "Korean", "Other"], + ["lao", "lao_Laoo", "Lao", "Other"], + ["luo", "luo_Latn", "Luo", "Other"], + ["mon", "khk_Cyrl", "Mongolian", "Other"], + ["tha", "tha_Thai", "Thai", "Other"], + ["vie", "vie_Latn", "Vietnamese", "Other"], +] +flores_lang_map = {i[0]: i for i in _flores_lang_map} +_flores_subtasks = [f"eng-{i}" for i in flores_lang_map if i != "eng" + ] + [f"{i}-eng" for i in flores_lang_map if i != "eng"] + +flores_datasets = [] +for _flores_subtask in _flores_subtasks: + _src, _tgt = _flores_subtask.split("-") + _, _flores_source, _src_inst, _ = flores_lang_map[_src] + _, _flores_target, _tgt_inst, _ = flores_lang_map[_tgt] + + flores_infer_cfg = dict( + ice_template=dict( + type=PromptTemplate, + template=dict( + begin="", + round=[ + dict( + role="HUMAN", + prompt= + f"Translate the following {_src_inst} statements to {_tgt_inst}.\n{{sentence_{_flores_source}}}" + ), + dict(role="BOT", prompt=f"{{sentence_{_flores_target}}}"), + ], + ), + ice_token="", + ), + retriever=dict(type=TopkRetriever, ice_num=8), + inferencer=dict(type=GenInferencer), + ) + flores_eval_cfg = dict( + evaluator=dict(type=BleuEvaluator), + pred_role="BOT", + ) + if _tgt == "zho_simpl": + flores_eval_cfg["pred_postprocessor"] = dict(type="flores") + flores_eval_cfg["dataset_postprocessor"] = dict(type="flores") + flores_datasets.append( + dict( + type=FloresFirst100Dataset, + abbr=f"flores_100_{_src}-{_tgt}", + name=f"{_flores_source}-{_flores_target}", + reader_cfg=dict( + input_columns=f"sentence_{_flores_source}", + output_column=f"sentence_{_flores_target}", + train_split="dev", + test_split="devtest"), + infer_cfg=flores_infer_cfg.copy(), + eval_cfg=flores_eval_cfg.copy(), + )) + +del _flores_lang_map, _flores_subtask, _src, _tgt, _, _flores_source, _src_inst, _flores_target, _tgt_inst diff --git a/configs/datasets/glm/ceval.py b/configs/datasets/glm/ceval.py new file mode 100644 index 00000000..5116b65f --- /dev/null +++ b/configs/datasets/glm/ceval.py @@ -0,0 +1,118 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import FixKRetriever +from opencompass.openicl.icl_inferencer import GLMChoiceInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import CEvalDataset + +ceval_reader_cfg = dict( + input_columns=['question', 'A', 'B', 'C', 'D'], + output_column='answer', + train_split='dev', + test_split="val") + +ceval_prompt_template = dict( + type=PromptTemplate, + template=None, + ice_token='', +) + +ceval_infer_cfg = dict( + ice_template=dict( + type=PromptTemplate, + template={ + answer: + f'{{question}}\n(A) {{/A}}\n(B) {{/B}}\n(C) {{/C}}\n(D) {{/D}}\n答案: ({answer}) {{{answer}}}\n' + for answer in ['A', 'B', 'C', 'D'] + }), + prompt_template=ceval_prompt_template, + retriever=dict(type=FixKRetriever), + inferencer=dict(type=GLMChoiceInferencer, fix_id_list=[0, 1, 2, 3, 4])) + +ceval_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +ceval_all_sets = [ + "操作系统", + "初中地理", + "初中化学", + "初中历史", + "初中生物", + "初中数学", + "初中物理", + "初中政治", + "大学编程", + "大学化学", + "大学经济学", + "大学物理", + "大学中国史", + "导游资格", + "法律职业资格", + "法学", + "概率统计", + "高等数学", + "高中地理", + "高中化学", + "高中历史", + "高中生物", + "高中数学", + "高中物理", + "高中语文", + "高中政治", + "公务员", + "工商管理", + "环境影响评价工程师", + "基础医学", + "计算机网络", + "计算机组成", + "教师资格", + "教育学", + "离散数学", + "临床医学", + "逻辑学", + "马克思主义基本原理", + "毛泽东思想和中国特色社会主义理论体系概论", + "兽医学", + "税务师", + "思想道德修养与法律基础", + "体育学", + "医师资格", + "艺术学", + "植物保护", + "中国语言文学", + "注册城乡规划师", + "注册电气工程师", + "注册会计师", + "注册计量师", + "注册消防工程师", +] + +ceval_datasets = [] +for _name in ceval_all_sets: + ceval_datasets.append( + dict( + type=CEvalDataset, + path="./data/ceval/release_ceval", + name=_name, + abbr='ceval-' + _name, + reader_cfg=ceval_reader_cfg, + infer_cfg=ceval_infer_cfg.copy(), + eval_cfg=ceval_eval_cfg.copy())) + + ceval_datasets[-1]['infer_cfg'][ + 'prompt_template'] = ceval_prompt_template.copy() + ceval_datasets[-1]['infer_cfg']['prompt_template']['template'] = dict( + begin=[ + dict( + role='SYSTEM', + fallback_role='HUMAN', + prompt=f'以下是中国关于{_name}考试的单项选择题，请选出其中的正确答案。'), + '', + ], + round=[ + dict( + role='HUMAN', + prompt= + '{question}\n(A) {A}\n(B) {B}\n(C) {C}\n(D) {D}\答案: ('), + ], + ) + +del _name diff --git a/configs/datasets/glm/mmlu.py b/configs/datasets/glm/mmlu.py new file mode 100644 index 00000000..7cf3a853 --- /dev/null +++ b/configs/datasets/glm/mmlu.py @@ -0,0 +1,155 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import FixKRetriever +from opencompass.openicl.icl_inferencer import GLMChoiceInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +mmlu_reader_cfg = dict( + input_columns=['input', 'A', 'B', 'C', 'D'], + output_column='target', + train_split='validation') + +mmlu_prompt_template = dict( + type=PromptTemplate, + template=None, + column_token_map={ + 'input': '', + 'A': '', + 'B': '', + 'C': '', + 'D': '', + 'target': '' + }, + ice_token='', +) + +mmlu_infer_cfg = dict( + ice_template=dict( + type=PromptTemplate, + template={ + target: '\n(A) \n(B) \n(C) \n(D) \n' + f'Answer: ({target}) \n' + for target in ['A', 'B', 'C', 'D'] + }, + column_token_map={ + 'input': '', + 'A': '', + 'B': '', + 'C': '', + 'D': '', + 'target': '' + }), + prompt_template=mmlu_prompt_template, + retriever=dict(type=FixKRetriever), + inferencer=dict(type=GLMChoiceInferencer, fix_id_list=[0, 1, 2, 3, 4])) + +mmlu_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +mmlu_all_sets = [ + "college_biology", + # "college_chemistry", + # "college_computer_science", + # "college_mathematics", + # "college_physics", + # "electrical_engineering", + # "astronomy", + # "anatomy", + # "abstract_algebra", + # "machine_learning", + # "clinical_knowledge", + # "global_facts", + # "management", + # "nutrition", + # "marketing", + # "professional_accounting", + # "high_school_geography", + # "international_law", + # "moral_scenarios", + # "computer_security", + # "high_school_microeconomics", + # "professional_law", + # "medical_genetics", + # "professional_psychology", + # "jurisprudence", + # "world_religions", + # "philosophy", + # "virology", + # "high_school_chemistry", + # "public_relations", + # "high_school_macroeconomics", + # "human_sexuality", + # "elementary_mathematics", + # "high_school_physics", + # "high_school_computer_science", + # "high_school_european_history", + # "business_ethics", + # "moral_disputes", + # "high_school_statistics", + # "miscellaneous", + # "formal_logic", + # "high_school_government_and_politics", + # "prehistory", + # "security_studies", + # "high_school_biology", + # "logical_fallacies", + # "high_school_world_history", + # "professional_medicine", + # "high_school_mathematics", + # "college_medicine", + # "high_school_us_history", + # "sociology", + # "econometrics", + # "high_school_psychology", + # "human_aging", + # "us_foreign_policy", + # "conceptual_physics", +] + +mmlu_key_sets = [ + 'college_biology', + 'college_chemistry', + 'college_computer_science', + 'college_mathematics', + 'college_physics', + 'electrical_engineering', + 'astronomy', + 'anatomy', + 'abstract_algebra', + 'machine_learning', + 'clinical_knowledge', + 'global_facts', + 'management', + 'nutrition', + 'marketing', + 'professional_accounting', +] + +mmlu_datasets = [] +for name in mmlu_all_sets: + mmlu_datasets.append( + dict( + type=HFDataset, + path='lukaemon/mmlu', + name=name, + reader_cfg=mmlu_reader_cfg, + infer_cfg=mmlu_infer_cfg.copy(), + eval_cfg=mmlu_eval_cfg)) + mmlu_datasets[-1]['infer_cfg'][ + 'prompt_template'] = mmlu_prompt_template.copy() + mmlu_datasets[-1]['infer_cfg']['prompt_template']['template'] = dict( + begin=[ + dict( + role='SYSTEM', + fallback_role='HUMAN', + prompt= + f'The following are multiple choice questions (with answers) about {name.replace("_", " ")}.' + ), + '', + ], + round=[ + dict( + role='HUMAN', + prompt= + '\n(A) \n(B) \n(C) \n(D) \nAnswer: ('), + ], + ) diff --git a/configs/datasets/govrepcrs/govrepcrs_gen_7643d5.py b/configs/datasets/govrepcrs/govrepcrs_gen_7643d5.py new file mode 100644 index 00000000..9768c6e0 --- /dev/null +++ b/configs/datasets/govrepcrs/govrepcrs_gen_7643d5.py @@ -0,0 +1,35 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import BleuEvaluator +from opencompass.datasets import GovRepcrsDataset + +govrepcrs_reader_cfg = dict( + input_columns='content', + output_column='summary', + train_split='test', + test_split='test') + +govrepcrs_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template= + "Please summarize the following English report in English:{content}\n{summary}."), + retriever=dict(type=ZeroRetriever), + inferencer=dict( + type=GenInferencer, batch_size=4, max_out_len=500, max_seq_len=8192)) + +govrepcrs_eval_cfg = dict( + evaluator=dict(type=BleuEvaluator), + pred_postprocessor=dict(type='general_cn'), + dataset_postprocessor=dict(type='general_cn')) + +govrepcrs_datasets = [ + dict( + type=GovRepcrsDataset, + path='./data/govrep/', + abbr='GovRepcrs', + reader_cfg=govrepcrs_reader_cfg, + infer_cfg=govrepcrs_infer_cfg, + eval_cfg=govrepcrs_eval_cfg) +] diff --git a/configs/datasets/gsm8k/gsm8k_gen.py b/configs/datasets/gsm8k/gsm8k_gen.py new file mode 100644 index 00000000..03ce0eeb --- /dev/null +++ b/configs/datasets/gsm8k/gsm8k_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .gsm8k_gen_2dd372 import gsm8k_datasets # noqa: F401, F403 diff --git a/configs/datasets/gsm8k/gsm8k_gen_2dd372.py b/configs/datasets/gsm8k/gsm8k_gen_2dd372.py new file mode 100644 index 00000000..f8e9fd42 --- /dev/null +++ b/configs/datasets/gsm8k/gsm8k_gen_2dd372.py @@ -0,0 +1,41 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +gsm8k_reader_cfg = dict(input_columns=['question'], output_column='answer') + +gsm8k_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict( + round=[ + dict(role='HUMAN', prompt="Question: Angelo and Melanie want to plan how many hours over the next week they should study together for their test next week. They have 2 chapters of their textbook to study and 4 worksheets to memorize. They figure out that they should dedicate 3 hours to each chapter of their textbook and 1.5 hours for each worksheet. If they plan to study no more than 4 hours each day, how many days should they plan to study total over the next week if they take a 10-minute break every hour, include 3 10-minute snack breaks each day, and 30 minutes for lunch each day?\nLet's think step by step\nAnswer:"), + dict(role='BOT', prompt="Angelo and Melanie think they should dedicate 3 hours to each of the 2 chapters, 3 hours x 2 chapters = 6 hours total.\nFor the worksheets they plan to dedicate 1.5 hours for each worksheet, 1.5 hours x 4 worksheets = 6 hours total.\nAngelo and Melanie need to start with planning 12 hours to study, at 4 hours a day, 12 / 4 = 3 days.\nHowever, they need to include time for breaks and lunch. Every hour they want to include a 10-minute break, so 12 total hours x 10 minutes = 120 extra minutes for breaks.\nThey also want to include 3 10-minute snack breaks, 3 x 10 minutes = 30 minutes.\nAnd they want to include 30 minutes for lunch each day, so 120 minutes for breaks + 30 minutes for snack breaks + 30 minutes for lunch = 180 minutes, or 180 / 60 minutes per hour = 3 extra hours.\nSo Angelo and Melanie want to plan 12 hours to study + 3 hours of breaks = 15 hours total.\nThey want to study no more than 4 hours each day, 15 hours / 4 hours each day = 3.75\nThey will need to plan to study 4 days to allow for all the time they need.\nThe answer is 4\n"), + dict(role='HUMAN', prompt="Question: Mark's basketball team scores 25 2 pointers, 8 3 pointers and 10 free throws. Their opponents score double the 2 pointers but half the 3 pointers and free throws. What's the total number of points scored by both teams added together?\nLet's think step by step\nAnswer:"), + dict(role='BOT', prompt="Mark's team scores 25 2 pointers, meaning they scored 25*2= 50 points in 2 pointers.\nHis team also scores 6 3 pointers, meaning they scored 8*3= 24 points in 3 pointers\nThey scored 10 free throws, and free throws count as one point so they scored 10*1=10 points in free throws.\nAll together his team scored 50+24+10= 84 points\nMark's opponents scored double his team's number of 2 pointers, meaning they scored 50*2=100 points in 2 pointers.\nHis opponents scored half his team's number of 3 pointers, meaning they scored 24/2= 12 points in 3 pointers.\nThey also scored half Mark's team's points in free throws, meaning they scored 10/2=5 points in free throws.\nAll together Mark's opponents scored 100+12+5=117 points\nThe total score for the game is both team's scores added together, so it is 84+117=201 points\nThe answer is 201\n"), + dict(role='HUMAN', prompt="Question: Bella has two times as many marbles as frisbees. She also has 20 more frisbees than deck cards. If she buys 2/5 times more of each item, what would be the total number of the items she will have if she currently has 60 marbles?\nLet's think step by step\nAnswer:"), + dict(role='BOT', prompt="When Bella buys 2/5 times more marbles, she'll have increased the number of marbles by 2/5*60 = 24\nThe total number of marbles she'll have is 60+24 = 84\nIf Bella currently has 60 marbles, and she has two times as many marbles as frisbees, she has 60/2 = 30 frisbees.\nIf Bella buys 2/5 times more frisbees, she'll have 2/5*30 = 12 more frisbees.\nThe total number of frisbees she'll have will increase to 30+12 = 42\nBella also has 20 more frisbees than deck cards, meaning she has 30-20 = 10 deck cards\nIf she buys 2/5 times more deck cards, she'll have 2/5*10 = 4 more deck cards.\nThe total number of deck cards she'll have is 10+4 = 14\nTogether, Bella will have a total of 14+42+84 = 140 items\nThe answer is 140\n"), + dict(role='HUMAN', prompt="Question: A group of 4 fruit baskets contains 9 apples, 15 oranges, and 14 bananas in the first three baskets and 2 less of each fruit in the fourth basket. How many fruits are there?\nLet's think step by step\nAnswer:"), + dict(role='BOT', prompt="For the first three baskets, the number of apples and oranges in one basket is 9+15=24\nIn total, together with bananas, the number of fruits in one basket is 24+14=38 for the first three baskets.\nSince there are three baskets each having 38 fruits, there are 3*38=114 fruits in the first three baskets.\nThe number of apples in the fourth basket is 9-2=7\nThere are also 15-2=13 oranges in the fourth basket\nThe combined number of oranges and apples in the fourth basket is 13+7=20\nThe fourth basket also contains 14-2=12 bananas.\nIn total, the fourth basket has 20+12=32 fruits.\nThe four baskets together have 32+114=146 fruits.\nThe answer is 146\n"), + dict(role='HUMAN', prompt="Question: {question}\nLet's think step by step\nAnswer:"), + ], + )), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer, max_out_len=512)) + +gsm8k_eval_cfg = dict(evaluator=dict(type=AccEvaluator), + pred_postprocessor=dict(type='gsm8k'), + dataset_postprocessor=dict(type='gsm8k_dataset')) + +gsm8k_datasets = [ + dict( + abbr='gsm8k', + type=HFDataset, + path='gsm8k', + name='main', + reader_cfg=gsm8k_reader_cfg, + infer_cfg=gsm8k_infer_cfg, + eval_cfg=gsm8k_eval_cfg) +] diff --git a/configs/datasets/humaneval/humaneval_gen.py b/configs/datasets/humaneval/humaneval_gen.py new file mode 100644 index 00000000..42f1694b --- /dev/null +++ b/configs/datasets/humaneval/humaneval_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .humaneval_gen_d428f1 import humaneval_datasets # noqa: F401, F403 diff --git a/configs/datasets/humaneval/humaneval_gen_28e126.py b/configs/datasets/humaneval/humaneval_gen_28e126.py new file mode 100644 index 00000000..3bb178c5 --- /dev/null +++ b/configs/datasets/humaneval/humaneval_gen_28e126.py @@ -0,0 +1,30 @@ +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 HFDataset, HumanEvaluator + +humaneval_reader_cfg = dict( + input_columns=['prompt'], output_column='task_id', train_split='test') + +# TODO: allow empty output-column +humaneval_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template='{prompt}'), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer, max_out_len=512)) + +humaneval_eval_cfg = dict( + evaluator=dict(type=HumanEvaluator), + k=[1, 10, 100], # the parameter only for humaneval + pred_postprocessor=dict(type='humaneval'), +) + +humaneval_datasets = [ + dict( + type=HFDataset, + path='openai_humaneval', + reader_cfg=humaneval_reader_cfg, + infer_cfg=humaneval_infer_cfg, + eval_cfg=humaneval_eval_cfg) +] diff --git a/configs/datasets/lcsts/lcsts_gen_427fde.py b/configs/datasets/lcsts/lcsts_gen_427fde.py new file mode 100644 index 00000000..ec7a1c7b --- /dev/null +++ b/configs/datasets/lcsts/lcsts_gen_427fde.py @@ -0,0 +1,32 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import RougeEvaluator +from opencompass.datasets import LCSTSDataset + +lcsts_reader_cfg = dict(input_columns=['content'], output_column='abst') + +lcsts_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict(role='HUMAN', prompt='阅读以下文章，并给出简短的摘要：{content}\n摘要如下：'), + ])), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer)) + +lcsts_eval_cfg = dict( + evaluator=dict(type=RougeEvaluator), + pred_role='BOT', + pred_postprocessor=dict(type='lcsts'), +) + +lcsts_datasets = [ + dict( + type=LCSTSDataset, + abbr='lcsts', + path='./data/LCSTS', + reader_cfg=lcsts_reader_cfg, + infer_cfg=lcsts_infer_cfg, + eval_cfg=lcsts_eval_cfg) +] diff --git a/configs/datasets/math/math_gen_78bcba.py b/configs/datasets/math/math_gen_78bcba.py new file mode 100644 index 00000000..9fb9a15c --- /dev/null +++ b/configs/datasets/math/math_gen_78bcba.py @@ -0,0 +1,68 @@ +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 MATHDataset, MATHEvaluator + +math_reader_cfg = dict(input_columns=['problem'], output_column='solution') + +math_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + "Problem:\nFind the domain of the expression $\frac{{\sqrt{{x-2}}}}{{\sqrt{{5-x}}}}$.}}\nSolution:" + ), + dict( + role="BOT", + prompt= + "The expressions inside each square root must be non-negative. Therefore, $x-2 \ge 0$, so $x\ge2$, and $5 - x \ge 0$, so $x \le 5$. Also, the denominator cannot be equal to zero, so $5-x>0$, which gives $x<5$. Therefore, the domain of the expression is $\boxed{{[2,5)}}$.\nFinal Answer: The final answer is $[2,5)$. I hope it is correct." + ), + dict( + role="HUMAN", + prompt= + "Problem:\nIf $\det \mathbf{{A}} = 2$ and $\det \mathbf{{B}} = 12,$ then find $\det (\mathbf{{A}} \mathbf{{B}}).$\nSolution:" + ), + dict( + role="BOT", + prompt= + "We have that $\det (\mathbf{{A}} \mathbf{{B}}) = (\det \mathbf{{A}})(\det \mathbf{{B}}) = (2)(12) = \boxed{{24}}.$\nFinal Answer: The final answer is $24$. I hope it is correct." + ), + dict( + role="HUMAN", + prompt= + "Problem:\nTerrell usually lifts two 20-pound weights 12 times. If he uses two 15-pound weights instead, how many times must Terrell lift them in order to lift the same total weight?\nSolution:" + ), + dict( + role="BOT", + prompt= + "If Terrell lifts two 20-pound weights 12 times, he lifts a total of $2\cdot 12\cdot20=480$ pounds of weight. If he lifts two 15-pound weights instead for $n$ times, he will lift a total of $2\cdot15\cdot n=30n$ pounds of weight. Equating this to 480 pounds, we can solve for $n$: \begin{{align*}} 30n&=480\\ \Rightarrow\qquad n&=480/30=\boxed{{16}} \end{{align*}}\nFinal Answer: The final answer is $16$. I hope it is correct." + ), + dict( + role="HUMAN", + prompt= + "Problem:\nIf the system of equations: \begin{{align*}} 6x-4y&=a,\\ 6y-9x &=b. \end{{align*}}has a solution $(x, y)$ where $x$ and $y$ are both nonzero, find $\frac{{a}}{{b}},$ assuming $b$ is nonzero.\nSolution:" + ), + dict( + role="BOT", + prompt= + "If we multiply the first equation by $-\frac{{3}}{{2}}$, we obtain $$6y-9x=-\frac{{3}}{{2}}a.$$Since we also know that $6y-9x=b$, we have $$-\frac{{3}}{{2}}a=b\Rightarrow\frac{{a}}{{b}}=\boxed{{-\frac{{2}}{{3}}}}.$$\nFinal Answer: The final answer is $-\frac{{2}}{{3}}$. I hope it is correct." + ), + dict(role="HUMAN", prompt="Problem:\n{problem}\nSolution:\n"), + ])), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer, max_out_len=512)) + +math_eval_cfg = dict( + evaluator=dict(type=MATHEvaluator), pred_postprocessor=dict(type='math')) + +math_datasets = [ + dict( + type=MATHDataset, + abbr='math', + path='./data/math/math.json', + reader_cfg=math_reader_cfg, + infer_cfg=math_infer_cfg, + eval_cfg=math_eval_cfg) +] diff --git a/configs/datasets/mmlu/mmlu_gen_057057.py b/configs/datasets/mmlu/mmlu_gen_057057.py new file mode 100644 index 00000000..cf2c95a7 --- /dev/null +++ b/configs/datasets/mmlu/mmlu_gen_057057.py @@ -0,0 +1,123 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import FixKRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import MMLUDataset + +# None of the mmlu dataset in huggingface is correctly parsed, so we use our own dataset reader +# Please download the dataset from https://people.eecs.berkeley.edu/~hendrycks/data.tar + +mmlu_reader_cfg = dict( + input_columns=["input", "A", "B", "C", "D"], + output_column="target", + train_split='dev') + +mmlu_all_sets = [ + "college_biology", + "college_chemistry", + "college_computer_science", + "college_mathematics", + "college_physics", + "electrical_engineering", + "astronomy", + "anatomy", + "abstract_algebra", + "machine_learning", + "clinical_knowledge", + "global_facts", + "management", + "nutrition", + "marketing", + "professional_accounting", + "high_school_geography", + "international_law", + "moral_scenarios", + "computer_security", + "high_school_microeconomics", + "professional_law", + "medical_genetics", + "professional_psychology", + "jurisprudence", + "world_religions", + "philosophy", + "virology", + "high_school_chemistry", + "public_relations", + "high_school_macroeconomics", + "human_sexuality", + "elementary_mathematics", + "high_school_physics", + "high_school_computer_science", + "high_school_european_history", + "business_ethics", + "moral_disputes", + "high_school_statistics", + "miscellaneous", + "formal_logic", + "high_school_government_and_politics", + "prehistory", + "security_studies", + "high_school_biology", + "logical_fallacies", + "high_school_world_history", + "professional_medicine", + "high_school_mathematics", + "college_medicine", + "high_school_us_history", + "sociology", + "econometrics", + "high_school_psychology", + "human_aging", + "us_foreign_policy", + "conceptual_physics", +] + +mmlu_datasets = [] +for _name in mmlu_all_sets: + _hint = f'There is a single choice question about {_name.replace("_", " ")}. Answer the question by replying A, B, C or D.' + mmlu_infer_cfg = dict( + ice_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + f"{_hint}\nQ: {{input}}\nA. {{A}}\nB. {{B}}\nC. {{C}}\nD. {{D}}\nA: " + ), + dict(role="BOT", prompt="{target}\n") + ]), + ), + prompt_template=dict( + type=PromptTemplate, + template=dict( + begin="", + round=[ + dict( + role="HUMAN", + prompt= + f"{_hint}\nQ: {{input}}\nA. {{A}}\nB. {{B}}\nC. {{C}}\nD. {{D}}\nA: " + ), + ], + ), + ice_token="", + ), + retriever=dict(type=FixKRetriever), + inferencer=dict(type=GenInferencer, fix_id_list=[0, 1, 2, 3, 4]), + ) + + mmlu_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_postprocessor=dict(type="first-capital")) + + mmlu_datasets.append( + dict( + abbr=f"lukaemon_mmlu_{_name}", + type=MMLUDataset, + path="./data/mmlu/", + name=_name, + reader_cfg=mmlu_reader_cfg, + infer_cfg=mmlu_infer_cfg, + eval_cfg=mmlu_eval_cfg, + )) + +del _name, _hint diff --git a/configs/datasets/mmlu/mmlu_gen_36560d.py b/configs/datasets/mmlu/mmlu_gen_36560d.py new file mode 100644 index 00000000..32730a0e --- /dev/null +++ b/configs/datasets/mmlu/mmlu_gen_36560d.py @@ -0,0 +1,109 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import FixKRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import MMLUDataset + +# None of the mmlu dataset in huggingface is correctly parsed, so we use our own dataset reader +# Please download the dataset from https://people.eecs.berkeley.edu/~hendrycks/data.tar +mmlu_reader_cfg = dict( + input_columns=["input", "A", "B", "C", "D"], + output_column="target", + train_split='dev') + +mmlu_all_sets = [ + "college_biology", + "college_chemistry", + "college_computer_science", + "college_mathematics", + "college_physics", + "electrical_engineering", + "astronomy", + "anatomy", + "abstract_algebra", + "machine_learning", + "clinical_knowledge", + "global_facts", + "management", + "nutrition", + "marketing", + "professional_accounting", + "high_school_geography", + "international_law", + "moral_scenarios", + "computer_security", + "high_school_microeconomics", + "professional_law", + "medical_genetics", + "professional_psychology", + "jurisprudence", + "world_religions", + "philosophy", + "virology", + "high_school_chemistry", + "public_relations", + "high_school_macroeconomics", + "human_sexuality", + "elementary_mathematics", + "high_school_physics", + "high_school_computer_science", + "high_school_european_history", + "business_ethics", + "moral_disputes", + "high_school_statistics", + "miscellaneous", + "formal_logic", + "high_school_government_and_politics", + "prehistory", + "security_studies", + "high_school_biology", + "logical_fallacies", + "high_school_world_history", + "professional_medicine", + "high_school_mathematics", + "college_medicine", + "high_school_us_history", + "sociology", + "econometrics", + "high_school_psychology", + "human_aging", + "us_foreign_policy", + "conceptual_physics", +] + +mmlu_datasets = [] +for _name in mmlu_all_sets: + _hint = f'The following are multiple choice questions (with answers) about {_name.replace("_", " ")}.\n\n' + mmlu_infer_cfg = dict( + ice_template=dict( + type=PromptTemplate, + template= + "{input}\nA. {A}\nB. {B}\nC. {C}\nD. {D}\nAnswer: {target}\n", + ), + prompt_template=dict( + type=PromptTemplate, + template= + f"{_hint}{{input}}\nA. {{A}}\nB. {{B}}\nC. {{C}}\nD. {{D}}\nAnswer:", + ice_token="", + ), + retriever=dict(type=FixKRetriever), + inferencer=dict(type=GenInferencer, fix_id_list=[0, 1, 2, 3, 4]), + ) + + mmlu_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_postprocessor=dict(type="first-capital"), + ) + + mmlu_datasets.append( + dict( + abbr=f"lukaemon_mmlu_{_name}", + type=MMLUDataset, + path="./data/mmlu/", + name=_name, + reader_cfg=mmlu_reader_cfg, + infer_cfg=mmlu_infer_cfg, + eval_cfg=mmlu_eval_cfg, + )) + +del _name, _hint diff --git a/configs/datasets/mmlu/mmlu_gen_a568f1.py b/configs/datasets/mmlu/mmlu_gen_a568f1.py new file mode 100644 index 00000000..d998ad95 --- /dev/null +++ b/configs/datasets/mmlu/mmlu_gen_a568f1.py @@ -0,0 +1,123 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import FixKRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import MMLUDataset + +# None of the mmlu dataset in huggingface is correctly parsed, so we use our own dataset reader +# Please download the dataset from https://people.eecs.berkeley.edu/~hendrycks/data.tar + +mmlu_reader_cfg = dict( + input_columns=["input", "A", "B", "C", "D"], + output_column="target", + train_split='dev') + +mmlu_all_sets = [ + "college_biology", + "college_chemistry", + "college_computer_science", + "college_mathematics", + "college_physics", + "electrical_engineering", + "astronomy", + "anatomy", + "abstract_algebra", + "machine_learning", + "clinical_knowledge", + "global_facts", + "management", + "nutrition", + "marketing", + "professional_accounting", + "high_school_geography", + "international_law", + "moral_scenarios", + "computer_security", + "high_school_microeconomics", + "professional_law", + "medical_genetics", + "professional_psychology", + "jurisprudence", + "world_religions", + "philosophy", + "virology", + "high_school_chemistry", + "public_relations", + "high_school_macroeconomics", + "human_sexuality", + "elementary_mathematics", + "high_school_physics", + "high_school_computer_science", + "high_school_european_history", + "business_ethics", + "moral_disputes", + "high_school_statistics", + "miscellaneous", + "formal_logic", + "high_school_government_and_politics", + "prehistory", + "security_studies", + "high_school_biology", + "logical_fallacies", + "high_school_world_history", + "professional_medicine", + "high_school_mathematics", + "college_medicine", + "high_school_us_history", + "sociology", + "econometrics", + "high_school_psychology", + "human_aging", + "us_foreign_policy", + "conceptual_physics", +] + +mmlu_datasets = [] +for _name in mmlu_all_sets: + _hint = f'There is a single choice question about {_name.replace("_", " ")}. Answer the question by replying A, B, C or D.' + mmlu_infer_cfg = dict( + ice_template=dict( + type=PromptTemplate, + template=dict(round=[ + dict( + role="HUMAN", + prompt= + f"{_hint}\nQuestion: {{input}}\nA. {{A}}\nB. {{B}}\nC. {{C}}\nD. {{D}}\nAnswer: " + ), + dict(role="BOT", prompt="{target}\n") + ]), + ), + prompt_template=dict( + type=PromptTemplate, + template=dict( + begin="", + round=[ + dict( + role="HUMAN", + prompt= + f"{_hint}\nQ: {{input}}\nA. {{A}}\nB. {{B}}\nC. {{C}}\nD. {{D}}\nA: " + ), + ], + ), + ice_token="", + ), + retriever=dict(type=FixKRetriever), + inferencer=dict(type=GenInferencer, fix_id_list=[0, 1, 2, 3, 4]), + ) + + mmlu_eval_cfg = dict( + evaluator=dict(type=AccEvaluator), + pred_postprocessor=dict(type="first-capital")) + + mmlu_datasets.append( + dict( + abbr=f"lukaemon_mmlu_{_name}", + type=MMLUDataset, + path="./data/mmlu/", + name=_name, + reader_cfg=mmlu_reader_cfg, + infer_cfg=mmlu_infer_cfg, + eval_cfg=mmlu_eval_cfg, + )) + +del _name, _hint diff --git a/configs/datasets/nq/nq_gen.py b/configs/datasets/nq/nq_gen.py new file mode 100644 index 00000000..598e5ae1 --- /dev/null +++ b/configs/datasets/nq/nq_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .nq_gen_c00b89 import nq_datasets # noqa: F401, F403 diff --git a/configs/datasets/nq/nq_gen_a6ffca.py b/configs/datasets/nq/nq_gen_a6ffca.py new file mode 100644 index 00000000..222fe5cc --- /dev/null +++ b/configs/datasets/nq/nq_gen_a6ffca.py @@ -0,0 +1,30 @@ +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 NaturalQuestionDataset, NQEvaluator + +nq_reader_cfg = dict( + input_columns=['question'], output_column='answer', train_split='test') + +nq_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict( + round=[ + dict(role='HUMAN', prompt='Answer these questions:\nQ: {question}?'), + dict(role='BOT', prompt='A:'), + ], )), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer)) + +nq_eval_cfg = dict(evaluator=dict(type=NQEvaluator), pred_role="BOT") + +nq_datasets = [ + dict( + type=NaturalQuestionDataset, + abbr='nq', + path='./data/nq/', + reader_cfg=nq_reader_cfg, + infer_cfg=nq_infer_cfg, + eval_cfg=nq_eval_cfg) +] diff --git a/configs/datasets/piqa/piqa_ppl.py b/configs/datasets/piqa/piqa_ppl.py new file mode 100644 index 00000000..a9e997fa --- /dev/null +++ b/configs/datasets/piqa/piqa_ppl.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .piqa_ppl_788dbe import piqa_datasets # noqa: F401, F403 diff --git a/configs/datasets/qaspercut/qaspercut_gen_943606.py b/configs/datasets/qaspercut/qaspercut_gen_943606.py new file mode 100644 index 00000000..70c12177 --- /dev/null +++ b/configs/datasets/qaspercut/qaspercut_gen_943606.py @@ -0,0 +1,30 @@ +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 QASPERCUTDataset, TriviaQAEvaluator + +qaspercut_reader_cfg = dict( + input_columns=['question', 'evidence'], + output_column='answer', + train_split='dev', + test_split='dev') + +qaspercut_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template="{evidence}\nAnswer these questions:\nQ: {question}?\nA:"), + retriever=dict(type=ZeroRetriever), + inferencer=dict( + type=GenInferencer, max_out_len=50, max_seq_len=8192, batch_size=4)) + +qaspercut_eval_cfg = dict(evaluator=dict(type=TriviaQAEvaluator)) + +qaspercut_datasets = [ + dict( + type=QASPERCUTDataset, + abbr='qaspercut', + path='./data/QASPER/', + reader_cfg=qaspercut_reader_cfg, + infer_cfg=qaspercut_infer_cfg, + eval_cfg=qaspercut_eval_cfg) +] diff --git a/configs/datasets/safety/safety_gen_c0a5b8.py b/configs/datasets/safety/safety_gen_c0a5b8.py new file mode 100644 index 00000000..13f61512 --- /dev/null +++ b/configs/datasets/safety/safety_gen_c0a5b8.py @@ -0,0 +1,30 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.openicl.icl_evaluator import ToxicEvaluator +from opencompass.datasets import SafetyDataset + +safety_reader_cfg = dict( + input_columns=['prompt'], + output_column='idx', + train_split='test', + test_split='test') + +# TODO: allow empty output-column +safety_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template='{prompt}'), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer)) + +safety_eval_cfg = dict(evaluator=dict(type=ToxicEvaluator), ) + +safety_datasets = [ + dict( + type=SafetyDataset, + path='./data/safety.txt', + reader_cfg=safety_reader_cfg, + infer_cfg=safety_infer_cfg, + eval_cfg=safety_eval_cfg) +] diff --git a/configs/datasets/siqa/siqa_ppl_764e42.py b/configs/datasets/siqa/siqa_ppl_764e42.py new file mode 100644 index 00000000..a3efeabe --- /dev/null +++ b/configs/datasets/siqa/siqa_ppl_764e42.py @@ -0,0 +1,34 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +siqa_reader_cfg = dict( + input_columns=['context', 'question', 'answerA', 'answerB', 'answerC'], + output_column='label', + test_split='validation') + +siqa_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 1: '{context} \nQ: {question}\nA: {answerA}', + 2: '{context} \nQ: {question}\nA: {answerB}', + 3: '{context} \nQ: {question}\nA: {answerC}', + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +siqa_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +siqa_datasets = [ + dict( + abbr="siqa", + type=HFDataset, + path='social_i_qa', + name='social_i_qa', + reader_cfg=siqa_reader_cfg, + infer_cfg=siqa_infer_cfg, + eval_cfg=siqa_eval_cfg) +] diff --git a/configs/datasets/siqa/siqa_ppl_b27551.py b/configs/datasets/siqa/siqa_ppl_b27551.py new file mode 100644 index 00000000..a0cb3ee5 --- /dev/null +++ b/configs/datasets/siqa/siqa_ppl_b27551.py @@ -0,0 +1,33 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +siqa_reader_cfg = dict( + input_columns=['context', 'question', 'answerA', 'answerB', 'answerC'], + output_column='label', + test_split='validation') + +siqa_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 1: 'The following makes sense:\n {context} \nQ: {question}\nA: {answerA}', + 2: 'The following makes sense:\n {context} \nQ: {question}\nA: {answerB}', + 3: 'The following makes sense:\n {context} \nQ: {question}\nA: {answerC}', + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +siqa_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +siqa_datasets = [ + dict( + abbr="siqa", + type=HFDataset, + path='social_i_qa', + reader_cfg=siqa_reader_cfg, + infer_cfg=siqa_infer_cfg, + eval_cfg=siqa_eval_cfg) +] diff --git a/configs/datasets/summedits/summedits_ppl_f2bd6e.py b/configs/datasets/summedits/summedits_ppl_f2bd6e.py new file mode 100644 index 00000000..130017ca --- /dev/null +++ b/configs/datasets/summedits/summedits_ppl_f2bd6e.py @@ -0,0 +1,58 @@ +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import PPLInferencer +from opencompass.openicl.icl_evaluator import AccEvaluator +from opencompass.datasets import HFDataset + +summedits_reader_cfg = dict( + input_columns=['doc', 'summary'], + output_column='label', + test_split='train') + +summedits_prompt1 = "Given the document below, you have to determine if 'Yes' or 'No', the summary is factually consistent with the document." +summedits_prompt2 = "Document:\n{doc}\nSummary:\n{summary}\nIs the summary factually consistent with the document? " +summedits_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template={ + 0: + dict( + begin=[ + dict( + role='SYSTEM', + fallback_role='HUMAN', + prompt=summedits_prompt1) + ], + round=[ + dict(role="HUMAN", prompt=summedits_prompt2), + dict(role="BOT", prompt="No") + ]), + 1: + dict( + begin=[ + dict( + role='SYSTEM', + fallback_role='HUMAN', + prompt=summedits_prompt1) + ], + round=[ + dict(role="HUMAN", prompt=summedits_prompt2), + dict(role="BOT", prompt="Yes") + ]), + }), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=PPLInferencer)) + +summedits_eval_cfg = dict(evaluator=dict(type=AccEvaluator)) + +summedits_datasets = [ + dict( + type=HFDataset, + abbr='summedits', + path='json', + split='train', + data_files='./data/summedits/summedits.jsonl', + reader_cfg=summedits_reader_cfg, + infer_cfg=summedits_infer_cfg, + eval_cfg=summedits_eval_cfg) +] diff --git a/configs/datasets/triviaqa/triviaqa_gen.py b/configs/datasets/triviaqa/triviaqa_gen.py new file mode 100644 index 00000000..8f2931f5 --- /dev/null +++ b/configs/datasets/triviaqa/triviaqa_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .triviaqa_gen_cc3cbf import triviaqa_datasets # noqa: F401, F403 diff --git a/configs/datasets/triviaqa/triviaqa_gen_cc3cbf.py b/configs/datasets/triviaqa/triviaqa_gen_cc3cbf.py new file mode 100644 index 00000000..ec60ad5a --- /dev/null +++ b/configs/datasets/triviaqa/triviaqa_gen_cc3cbf.py @@ -0,0 +1,33 @@ +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 TriviaQADataset, TriviaQAEvaluator + +triviaqa_reader_cfg = dict( + input_columns=['question'], + output_column='answer', + train_split='dev', + test_split='dev') + +triviaqa_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict( + round=[ + dict(role='HUMAN', prompt='Question: {question}\nAnswer: '), + ], )), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer, max_out_len=50)) + +triviaqa_eval_cfg = dict( + evaluator=dict(type=TriviaQAEvaluator), pred_role='BOT') + +triviaqa_datasets = [ + dict( + type=TriviaQADataset, + abbr='triviaqa', + path='./data/triviaqa/', + reader_cfg=triviaqa_reader_cfg, + infer_cfg=triviaqa_infer_cfg, + eval_cfg=triviaqa_eval_cfg) +] diff --git a/configs/datasets/triviaqarc/triviaqarc_gen_6c1726.py b/configs/datasets/triviaqarc/triviaqarc_gen_6c1726.py new file mode 100644 index 00000000..e80d23f3 --- /dev/null +++ b/configs/datasets/triviaqarc/triviaqarc_gen_6c1726.py @@ -0,0 +1,37 @@ +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 TriviaQArcDataset, TriviaQAEvaluator + +triviaqarc_reader_cfg = dict( + input_columns=['question', 'evidence'], + output_column='answer', + train_split='dev', + test_split='dev') + +triviaqarc_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict( + round=[ + dict( + role='HUMAN', + prompt='{evidence}\nAnswer these questions:\nQ: {question}?A:'), + dict(role='BOT', prompt=''), + ], )), + retriever=dict(type=ZeroRetriever), + inferencer=dict( + type=GenInferencer, max_out_len=50, max_seq_len=8192, batch_size=4)) + +triviaqarc_eval_cfg = dict( + evaluator=dict(type=TriviaQAEvaluator), pred_role='BOT') + +triviaqarc_datasets = [ + dict( + type=TriviaQArcDataset, + abbr='triviaqarc', + path='./data/triviaqa-rc/', + reader_cfg=triviaqarc_reader_cfg, + infer_cfg=triviaqarc_infer_cfg, + eval_cfg=triviaqarc_eval_cfg) +] diff --git a/configs/datasets/truthfulqa/truthfulqa_gen.py b/configs/datasets/truthfulqa/truthfulqa_gen.py new file mode 100644 index 00000000..57c4e797 --- /dev/null +++ b/configs/datasets/truthfulqa/truthfulqa_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .truthfulqa_gen_0a3a53 import truthfulqa_datasets # noqa: F401, F403 diff --git a/configs/datasets/truthfulqa/truthfulqa_gen_0a3a53.py b/configs/datasets/truthfulqa/truthfulqa_gen_0a3a53.py new file mode 100644 index 00000000..a7d6c7cb --- /dev/null +++ b/configs/datasets/truthfulqa/truthfulqa_gen_0a3a53.py @@ -0,0 +1,40 @@ +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 TruthfulQADataset, TruthfulQAEvaluator + +truthfulqa_reader_cfg = dict( + input_columns=['question'], + output_column='reference', + train_split='validation', + test_split='validation') + +# TODO: allow empty output-column +truthfulqa_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict(round=[dict(role="HUMAN", prompt="{question}")])), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer)) + +# Metrics such as 'truth' and 'info' needs +# OPENAI_API_KEY with finetuned models in it. +# Please use your own finetuned openai model with keys and refers to +# the source code for more details +# Metrics such as 'bleurt', 'rouge', 'bleu' are free to test + +# When key is set to "ENV", the key will be fetched from the environment +# variable $OPENAI_API_KEY. Otherwise, set key in here directly. +truthfulqa_eval_cfg = dict( + evaluator=dict( + type=TruthfulQAEvaluator, metrics=('truth', 'info'), key='ENV'), ) + +truthfulqa_datasets = [ + dict( + type=TruthfulQADataset, + path='truthful_qa', + name='generation', + reader_cfg=truthfulqa_reader_cfg, + infer_cfg=truthfulqa_infer_cfg, + eval_cfg=truthfulqa_eval_cfg) +] diff --git a/configs/datasets/z_bench/z_bench_gen.py b/configs/datasets/z_bench/z_bench_gen.py new file mode 100644 index 00000000..5d71997a --- /dev/null +++ b/configs/datasets/z_bench/z_bench_gen.py @@ -0,0 +1,4 @@ +from mmengine.config import read_base + +with read_base(): + from .z_bench_gen_5813ec import z_bench_dataset # noqa: F401, F403 diff --git a/configs/summarizers/groups/agieval.py b/configs/summarizers/groups/agieval.py new file mode 100644 index 00000000..0334cd46 --- /dev/null +++ b/configs/summarizers/groups/agieval.py @@ -0,0 +1,17 @@ +agieval_summary_groups = [] + +_agieval_chinese_sets = ['gaokao-chinese', 'gaokao-english', 'gaokao-geography', 'gaokao-history', 'gaokao-biology', 'gaokao-chemistry', 'gaokao-physics', 'gaokao-mathqa', 'logiqa-zh', 'jec-qa-kd', 'jec-qa-ca', 'gaokao-mathcloze'] +_agieval_chinese_sets = ['agieval-' + s for s in _agieval_chinese_sets] +agieval_summary_groups.append({'name': 'agieval-chinese', 'subsets': _agieval_chinese_sets}) + +_agieval_english_sets = ['lsat-ar', 'lsat-lr', 'lsat-rc', 'logiqa-en', 'sat-math', 'sat-en', 'sat-en-without-passage', 'aqua-rat', 'math'] +_agieval_english_sets = ['agieval-' + s for s in _agieval_english_sets] +agieval_summary_groups.append({'name': 'agieval-english', 'subsets': _agieval_english_sets}) + +_agieval_gaokao_sets = ['gaokao-chinese', 'gaokao-english', 'gaokao-geography', 'gaokao-history', 'gaokao-biology', 'gaokao-chemistry', 'gaokao-physics', 'gaokao-mathqa', 'gaokao-mathcloze'] +_agieval_gaokao_sets = ['agieval-' + s for s in _agieval_gaokao_sets] +agieval_summary_groups.append({'name': 'agieval-gaokao', 'subsets': _agieval_gaokao_sets}) + +_agieval_all = ['gaokao-chinese', 'gaokao-english', 'gaokao-geography', 'gaokao-history', 'gaokao-biology', 'gaokao-chemistry', 'gaokao-physics', 'gaokao-mathqa', 'logiqa-zh', 'lsat-ar', 'lsat-lr', 'lsat-rc', 'logiqa-en', 'sat-math', 'sat-en', 'sat-en-without-passage', 'aqua-rat', 'jec-qa-kd', 'jec-qa-ca', 'gaokao-mathcloze', 'math'] +_agieval_all = ['agieval-' + s for s in _agieval_all] +agieval_summary_groups.append({'name': 'agieval', 'subsets': _agieval_all}) diff --git a/configs/summarizers/medium.py b/configs/summarizers/medium.py new file mode 100644 index 00000000..68ac497e --- /dev/null +++ b/configs/summarizers/medium.py @@ -0,0 +1,88 @@ +from mmengine.config import read_base + +with read_base(): + from .groups.agieval import agieval_summary_groups + from .groups.mmlu import mmlu_summary_groups + from .groups.ceval import ceval_summary_groups + from .groups.bbh import bbh_summary_groups + from .groups.GaokaoBench import GaokaoBench_summary_groups + from .groups.flores import flores_summary_groups + +summarizer = dict( + dataset_abbrs = [ + '--- Exam ---', + 'agieval', + 'mmlu-all-set', + "ceval", + "GaokaoBench", + "bbh", + '--- Coding ---', + 'openai_humaneval', + 'mbpp', + '--- ChineseUniversal ---', + 'C3', + 'CMRC_dev', + 'DRCD_dev', + 'afqmc-dev', + 'cmnli', + 'ocnli', + 'bustm-dev', + 'chid-dev', + 'cluewsc-dev', + 'csl_dev', + 'eprstmt-dev', + 'ocnli_fc-dev', + 'tnews-dev', + 'lcsts', + '--- Completion ---', + 'lambada', + 'story_cloze', + '--- EnglishUniversal ---', + 'AX_b', + 'AX_g', + 'BoolQ', + 'CB', + 'COPA', + 'MultiRC', + 'RTE', + 'ReCoRD', + 'WiC', + 'WSC', + 'race-high', + 'race-middle', + '--- NLG ---', + 'Xsum', + '--- Reasoning ---', + 'gsm8k', + 'summedits', + 'math', + 'TheoremQA', + '--- QA ---', + 'hellaswag', + 'ARC-e', + 'ARC-c', + 'commonsense_qa', + 'piqa', + 'siqa', + 'strategyqa', + 'winogrande', + 'openbookqa', + 'openbookqa_fact', + 'nq', + 'triviaqa', + '--- Translation ---', + 'flores_100_Indo-European-Germanic_English', + 'flores_100_English_Indo-European-Germanic', + 'flores_100_Indo-European-Romance_English', + 'flores_100_English_Indo-European-Romance', + 'flores_100_zho_simpl-eng', + 'flores_100_eng-zho_simpl', + '--- Security ---', + 'crows_pairs', + ], + summary_groups=sum([v for k, v in locals().items() if k.endswith("_summary_groups")], []), + prompt_db=dict( + database_path='configs/datasets/log.json', + config_dir='configs/datasets', + blacklist='.promptignore'), +) diff --git a/configs/summarizers/small.py b/configs/summarizers/small.py new file mode 100644 index 00000000..b8a41949 --- /dev/null +++ b/configs/summarizers/small.py @@ -0,0 +1,64 @@ +from mmengine.config import read_base + +with read_base(): + from .groups.agieval import agieval_summary_groups + from .groups.mmlu import mmlu_summary_groups + from .groups.ceval import ceval_summary_groups + from .groups.bbh import bbh_summary_groups + from .groups.GaokaoBench import GaokaoBench_summary_groups + from .groups.flores import flores_summary_groups + +summarizer = dict( + dataset_abbrs = [ + '--- Exam ---', + 'mmlu-all-set', + "ceval", + "bbh", + '--- ChineseUniversal ---', + 'CMRC_dev', + 'DRCD_dev', + 'afqmc-dev', + 'bustm-dev', + 'chid-dev', + 'cluewsc-dev', + 'eprstmt-dev', + '--- Coding ---', + 'openai_humaneval', + 'mbpp', + '--- Completion ---', + 'lambada', + 'story_cloze', + '--- EnglishUniversal ---', + 'AX_b', + 'AX_g', + 'BoolQ', + 'CB', + 'COPA', + 'MultiRC', + 'RTE', + 'ReCoRD', + 'WiC', + 'WSC', + 'race-high', + 'race-middle', + '--- Reasoning ---', + 'math', + 'gsm8k', + 'summedits', + '--- QA ---', + 'hellaswag', + 'piqa', + 'winogrande', + 'openbookqa', + 'openbookqa_fact', + 'nq', + 'triviaqa', + '--- Security ---', + 'crows_pairs', + ], + summary_groups=sum([v for k, v in locals().items() if k.endswith("_summary_groups")], []), + prompt_db=dict( + database_path='configs/datasets/log.json', + config_dir='configs/datasets', + blacklist='.promptignore'), +) diff --git a/docs/en/_static/css/readthedocs.css b/docs/en/_static/css/readthedocs.css new file mode 100644 index 00000000..1891c912 --- /dev/null +++ b/docs/en/_static/css/readthedocs.css @@ -0,0 +1,62 @@ +.header-logo { + background-image: url("../image/logo.png"); + background-size: 183px 50px; + height: 50px; + width: 183px; +} + +@media screen and (min-width: 1100px) { + .header-logo { + top: -12px; + } +} + +pre { + white-space: pre; +} + +@media screen and (min-width: 2000px) { + .pytorch-content-left { + width: 1200px; + margin-left: 30px; + } + article.pytorch-article { + max-width: 1200px; + } + .pytorch-breadcrumbs-wrapper { + width: 1200px; + } + .pytorch-right-menu.scrolling-fixed { + position: fixed; + top: 45px; + left: 1580px; + } +} + + +article.pytorch-article section code { + padding: .2em .4em; + background-color: #f3f4f7; + border-radius: 5px; +} + +/* Disable the change in tables */ +article.pytorch-article section table code { + padding: unset; + background-color: unset; + border-radius: unset; +} + +table.autosummary td { + width: 50% +} + +img.align-center { + display: block; + margin-left: auto; + margin-right: auto; +} + +article.pytorch-article p.rubric { + font-weight: bold; +} diff --git a/docs/en/_static/image/logo.png b/docs/en/_static/image/logo.png new file mode 100644 index 0000000000000000000000000000000000000000..8b5efee6f2263d3ff798682e8a266d830b002e3c GIT binary patch literal 12650 zcmV-wF_q4VP)!@i001BWNkl zd7KYu|gexH^DkzEqA%FI+?Rw? zASd@J0tp}@BKVaX!Wqc5$L!2>S3S?~kDi&GnVs1jy9pcV_iO*y>FKVn>ZyAAsi&Uj zDH0J3CBiVw7lNitnuww!B4Vwz)@h@ib&ffcP7zVQPzXay1g*3&MjM^Y_N09OkdVHAyH-9j( z*5&erbSeeFI%h*`onvOLmDbw#Ja}IDWUaNewkeYVz|3J75|MMxcm_aICasi;z6!FT zW7bM(t%)f5G5X$JYi+GtM@@~1S?8?p<7)^>E9YD;pVwL!g1|YaloAoGbvl(&s<-h1 zL?4XSIhLMW&N&tl&-0QA-xw2Vi>f5il~P(MXPx#HfK)Qsr~+E+Sh0%vZ{@#Wd<_+8 z6$%Iy_PzXY6qetnSv(7o|taF)6S}R3F3Ex*rC480b>G6yKBwr{9 z*UoLV(at%*17M8)5SkYg^P%$JF}{Y1v;vS!Bz!+oM2ihpikLZ)?f|q>#xu!eqAb@F z3Ew(roh724Tyeszl&YMQJIB%GF)<%y{tL#}P?1)iDb|FUopaW)bB>t- z5viFEcV~M<#29U~7LiEL%ohsf$IUFceBNlCNciivK`}8yh-c8e-JcGr)S#6~5OiG(q_CzsC^NNepJXM1`)&+~n+yvmSHC8Gyto$KkzMUUHB zm+*_vPHWZNl=;xw6%#XL5Rq(8j+rABsXV?2SYO81x|3EL@BEY9w;g}=MMrrE1s*Y= z!f37wQqDm`KeSR&1Vaj{VtqMl#LJ$DZT4gfnLZOh&_IRF2CK5zLZhWm9X(i4%lEh!0 zd2`&vb3p`=zu*0%4O>cgC-|BHShg(5gakpVXv#B2t30QMx9674{?uOQq{zUygm0d` z=Z8sERr#i-Ox<4p)kJG zI=KsJG5`Yt03rZ5bfu+I03fJ09{JJhkDT?RW4HGVv(u3}s+3YH^mIyC(nvZbj#W>d z_c`R}wfFe2A7xC;dO@L3FuG!V5n0#k1{G;fHKGbbV|=X>X(cFFWN0}LmI$^e zz9O_Xp2GcqTe{u0TOPFMcni3YgCG=1SSjo^=?5=Ae)>K;ZKi|)e(m+zWmF3b*mcpI zn3(kzYi$_XWHJH3ISztQP&EvPGBjwBHyn$$T z6)O+TJgPh)WwwWY`-VJ-nHc)IUrzHBj5f)UGXU_4x$WP&ZG)DsOE35a<*iEgihlt% z+u;|ay2p=cd2rUL1YTEaV@%9?12bo{J$}NE%;edg90A6dl4rwEXV8YUC{ZT$i{C#r zcH()P?0nw1iRW#ys@GW`3}=r|(td^2DCgJr-%8(tqD{`y(#6>X9=qTL5Pm z-2foQOYB6zk1u=3RmE~!v6i41WPR*^_yuPi+bgOpdWv>LLG>+SVq(@O2G7b4APV3X z2hpJ5%ST+n==H~caON?6mq0CG1neZQDFr4{rQ?=hF2Cu~C0%s-Q9EtF`!Ub_^8)}P z&Mup+{KWlFy|U*Q#z!YR;FxpnN?AKr4ec~1=pwe?9 zCMITmVUVR&FJe0bedI|OtjM>Kq(iUc(_=Rcw9;%8<#*j?^dZwHRGiJ%-}%hDOSDq< z%e#Dh+p*0xg#`}D)@dLo@Rq!D(}-s6pvwmQKB=Y8j(=MO?=)0`6U$7%v64zaF;#OOZ z(7kEA8r(_p4!)x(^hg*`V1NM8&>tT&?9@YBSSs^^0f10&`{N&o1A(PfK2=-Tk3M|+ zi)TLf&)1jix$~x`!UvZ@qg#Cdjtk?r+~n@-j(q0%rPGeR@ZAOH6ypd){~1#WcrSyi=V~4 zzPM}LR)-N6?`p&b-FFqg6d?k@B3dL+s$?k>oN(Tq076Sk7QLL6G})#1UH?B|0Z}{` zjEPzASQFh!s)sr&7-4W0z^iXA%M5QGnekMl0a1k_RMv)y2B#2`0Ic$A>#*0hqaJzY z_0X}W0Vt5(Y*aI`0g4NOA(Ma!3V;aBUhsBGh5JqzOC?ct-@@0vxYIa74oE7G=3Mh_ zDrm_TVTWvPse5;(Q_6G2{=e33Zd&CcCcXVbekI%j6u$3z^g8LV~w&nN@ zHrk*mQO~DBL{#g6B!%*kLk|4HbyuBL{^%U*$PHBhxL@D!4-rpu_sK`>91EwInEwt| z+uuq=#%$Q(scx!Pu`d73(GYZYcGsToR%w?D5qakI?gthwt{a(5AW%?w57$pWPTc#$NTcj*yID+oElBMV-=`p@?oQ?woOtXk{E%s%dP7PYpZf)r3t?nmBIFTND!$GlUuB+@VHmh=4?a_L#WE z9y^^-{%VU2y*J)3I5vW$2A7TPcRcYnn25Xg+xg>-iWR^XmKtxgNS4VI^YJ*P5dvNm zDNSrd^=f7$3_{UaD<&pp{b#NDviGI9L|B|bmUW`}kH5u43CKP>Zqn$|NdF30HzoJB z$z&v~lyF;{eDLO-Ebv+n5DA3XSE(RacGelkoc*JN1)!BUfyKnctT(JfY1I%0)XIEI zdrx+jSW&8FKh~i*BVBix)LvqeQs;6-U}gs-m644%l$T$aqX`O*L_+|zj*|9qBsZh~ z2|z)^hs17QePKxH7a@*S=?n3;&k2s;Z_P#RF6$n`-;RIxD81SSC_f7u_tbM)6I3wp#x8&l9==wl2~xjdRS zm?qReuHUGR)KE-FwR9^Z8^lIaoCdS-O@N4*mLBfB(k|OO|(wOOl2-tBeNuYzw4) z#S&(3Vo8CZE3z94goHpWV1b4RMGovD72cs1Ag~6N{9VNnIJBr))?5Re6EkeGc1hMj z!;{1)IMPW^ebBS~p#k1;J;iY@G3*GI02FFxoT}}OBw)=5eD0Y&1oknA`7l{MX|+i? zJE$lSfdXYa&pGXYqh{<(h+=EX;=o$#wAMr}2m{}12JpZm&)Z!fwF;?+3Q)kR1lyI9;KA7MT8qt42tRe3)MR#0I+R7L}0A*K+> z;`cj`JMr8{7rq);D=a$E?p}OHCIKku#LnUm54z@?58tM4hmV=m z-DxbXw&)b+r034HQKun^3?H@6irngMM^2uiLdUE=Ta(0mBv_dZsyIth-MQeZTOVi| zbC}mW-5Yj5s%3if1~bNQbJRnBeIc-{SSd{s=Aj5=Ns|(gkRTx-h=B}d_adR;+GbM4 z>KA(MBRIc#a?$4T;Vm0H_rlT@T?t#W!-a?;2t2Vl$!Cd`@sh-~1C_K^iK*89_h&a1 z@H4Nxw<1?ZMIwRLTHuHnqWjDXC1g49$X~9LYQ_vXR%0f(@#jv~iID(UIT9@d2DC`1 z>AgozXQvPN-c1}jh!D%$@4WMW|NLk7k`KV*r8A+mwEF5sBaBr+&)f<%%BiLqAO?_M zJjYMz4@JWg5z{nP`pEwYpWW!;2QM+6qv8}MC-5AMm)e(V_n4&HW}Y*B+UM3ueLpk= zff+VkTK zEL!+Og0d*4?5qfo2>>5aUKe4fv=)itXh6lS26K#s3;uP%`Pba@@Y6&INLoaoLMVq2 z%1>|*?SWHNI<(fo@|%b%&H159?F*9v0Fj;*vyx_Yrx*i8;B;bl_^p)(i8$YJhewT; z{2Vn+C#re=io52ny?PD-LceK>Ba#ZwWM7$OAIwg9(sx5(4)d*Jj>+abDi8XCeD}On zn)WjQ2-PV$X1!w#{jC_5L=jX#xbK0N_xtkZ1Gjz7UHAdW0+>2sBvqFtuq;LsGgNmW z+mguCTq}T=na;`zQGUtXgukx4COep6)5ocD}St();eBIr>{2|ZF1GNvQ1_!4c-T&ynCrvpW zWE@)pc}hM1%;j5*Z6lbvLV!63<(Jo7fl~H+p+d-+5xbBZMFfldY2QEeyfY6WPjtOV z35U>Yo>6n~#%q3b(Ed9$sBgi#gg-=-yRcQc#WmY=9o7Nk$Dkkj^WjHQ8-M0zLVk3vYS zy29*(ifr9j5i2lJKOq2MXjRIs^vdf?Ke5fxij`C3-21NQF$Gz9Ze9@vpYlJK|KYOP zhK9K?a1?6JzWUbfo3v%hH|niw z`{V-Oi3W*hd;|MGr+@t?|1(()Uck&6dxwT#4+7C3$55;7#m7=ipI+{9^A>ynWd&%T zFtU#9i!xd@^7s&6cxhGd)B465+E|ub=P0z<+Z(oB0AiJ% zB$_}zqWB~LVn9}iNJ5XYh0ccsLbXp>sAaNZgOaDK&CUUwX*$%Y9)Ohd0I*^xc;Xzh z>z>#5r@Ch5!{7MkuPP1@Alx~(YUhs%9H1H0x?g2|1g5b3&%UYcBn)^>(-o{m0xCxp z({dq2E+M*bQ2W+5&HQ=2Td`_xuhewI?zswL(vyXftSzLZuu}l#_bj_DMF627XlgqE zPFc!9jAA7!09Z75#T|Wri}c7zLCIz(1m?FT`_$59ZsZoz0U-vl&k7)G9GocEGaHQs zPOiY|5mOX=RtylaCRiq=Sb;2a2y1Lm8%|sKfW4+&Et;S*5R_k>LJELp%WxYt^V{#< z+u(8$kV;R336+2e#aLi5LK>8VHHwVQ-Zm)Xag9AaJt!bD5a~5>IDYgWul)Ad<|LyPpVe?NN07Vdw6%zu2B}Z0UQKVX?EMA(cb1y*; z=hd+Np*?~j;i9MlB9)dz0283aw0_mPz7J4HbZ<2jI~S;?1BALsKq8_7Q31gcTPq?& z)H1zqExvKWZ>pbwef8geY9y)<84DEt^YrDSlqfAqi&B7h%bkx_pUA)b-8|hqRmeI5 zQGu#0h$IY^BjKCpJ>#|P&s^VULa6tTK$Mul+K93wAOyuf0fj2l+U^JrNHy>8taZqd zfM7*MB`b2uWM()-X}v3P(F2TrR=C!gvmY5bZVItb^zKm6UDzpxwyL%L|6Y2pZf%>I zzm(ZvGLaIpBuwH!fe=U45Mt7r5@K2>Z7^mEbDaeo-7>?`2%-dutsFH`_h3>W@4aqA z&C~i`3&;n{dQ|J)q#a4*$gPUwlPsY?qDZr-K)3F7+K(@-+l19(-9+Am-#+^;3-Y9p zNnxw4#_jRN&n;WN?EWVfMX$0`4y%*eMPN~sQL7?SS)g=talYWX<`IZt1<-8Py7xUp zwN_14AfD~IQPC>-PkEW?PW5&aA#msX3f<}}px#7XDQ$CX`d{}Hg6*8EtD7xiQzO0_ zgkhtrD$j+0Yc(TQ&C@CuTzw@0uk1Bt1DCtIQXH$+>EJ3+$X&As9$lZ!bMpr6t9&{s z1h#8_wF=LwxvGQ_sm`UjmVQd{4GP^JC0#%JRAzLhYsF2KN75tr&KG*^s*>$ii54*{ zSqsjunD198NCI?cgO)Mv#H;$~YW7|MU^0_ft?JrWT2Xh$yviB~In}a1$toqV1K_&m z(7;?z26-F{9E3~zjB?cP9kf$mtqLZ!EKDtY?!Dh#`sY!@6Pu1sYS*R6=JK^KIsxWA z{=vM*Ke&CtTdUX*0)$IrNeBRmxohF#`HwFhy!JJK!rJuD;8he=0KPioQ&pcucNH4; zfSfCeDv@&eBuG+)93g?1nNkQUG9wC9^MO^ON{ooFbx!OG1dg1cUM~{@5j90S$Tl2B z0h5phlvsd8$QLU$t}05Ug)8Nw)QBn~Q)TR1eFO+xSJSXbRmF-DM_T#9fM@|%J3|P= zd-08~{!3n(Ya!7f7OP1(iq@(;A+VriNU(qg6TwrYYgSEiwx@re8p<>kMOAsGS6`p`R07?n#W!9 zQ#vP5B)!tShs9NyWvyyGpi)%%lK9leH*y{GxoiFtk6uY2UHvliGY9nZ99D%v&KN_Y zbA){kzLs>rVd(qHCsL|ogH1l|C#pK-nBmO_?fxmK`j*HttEc%X9|!?>nJ?R(2Z(K; zlC%nb(gQ`jQi_Svfz`Qs=FFen*j%BSk-OmhD}H+R@%>%nTu;ZXDFb1!Mq=0=j?-10 zQ>m9;@7(6&MI*l@Kysi8RidX;_&x>+M2iYS+cN6#!1W3^EiJ;*c!~x99;DE9Gr^DI z(gjX8?_X6-tx(?A0}QS$jv&B*JopLH(l(8}bmc(_+2wB|@{7-e>_%huS=?1QIPu9( z9vNnDQ`Fz`F(KzV?}8`@M2R+&(I+3RftT<4GqDFnfHRrtRboQC-(Gz8$tUjHe=CGJ z?7od0Ypn{Q%d~xoRdc1x5@W8t@h?Xn@}=lwz&UZ1tM9mT?$7qwdutMQq%^Fv)^9nW zO6YM#)kRUuwDNib2|O1TdLq9N0bF*)z4@MOdc#91D|z&Z=U;R28KnwqpQ@rdyE|_* zN=I_=3)_ye9dmW_9zgFhlhyhDAHqr*WZlXU?)}Si$^=BvzUQEGqH?aqn0n(^p^>N} zaQlQ&?Gr{#-g&d$qf+{7!256AP&|<`07RP^I}D$`!-mm7D()}8Hc%ne3~SfX;M6+Q zyp0u`E3TN`-}#i_?42nOAdrRz?9%&FuDiw*`{^(A$~)hm3|(cDX1*Oz3Z$Vx@O{d! zm{}>DSQXkf+O6O9BtO?Vm*AC56(G|26?3Ya7F=@aZPEV%Rohg+Q8?fqg+i_pRmvGI z%vIzQ$|36J-FNQs!p`#jNgx_4vi-a%1$i&*z8g@HwUj{C-XW-#ZsHfZW)b{Sz?m*& zXH|aBDme4Z-}T=Bu~xG43_%+OLc-TJ`|zmdLbcYZjynGL#A;3ZC535WUmAugsC%jD-FiT*s2ogX0U_k^53;rK3u-sfdg07H%zf7 zsA!@?ia7faxc746B!dU?a}PwfvYYL2PUvf&th3H{fv2e}*PUq_)tnjK+SdD{GR&{e z+Bf9@17<8+W?y>kg}%iJaN+e&+0Z79=iKt|9r+bhkCi*0d^;2cVCsZXbtk|Zv=*Dg zBE=;~B7i{dq8tB_cZvkxKYT5uKNfEwBd$hcXKlmAMhqckJ8!K$s#RCVy{7G;N<*nC z7ZL!8UcW9Ov+E9HQL7|q0LT@P>=kO+!pMHo+9-NomoF^gscO5J`2Trnar7?%(q}PV z!@7YPGdglyGYijV>&kka?YhSsabV@c-eB}}*WJjwjYKc2X}kgmg{q9(#rW0=96=J5 zqM<7vy2i0WBr4Z+M<2aGz4!J5)H0PSCKpP=a$=>=Za3a8pDU_3_MPif0O51n>bAngKMr^ zIT)vX#c(bGqyy8Mr52ssBv-@pfw3S&l^-M0T!of26@v^IVSvSviz z?>8Lbi2xw#FOJ6mL|Ou%;%TaHe(m$1(2;&<;eQy#jDUj{1&$saKDF8qTgy&6&f2 zsP_UMxxv(`Bjgf&G`I-uB|WK`_w^?yzjTpeM4ecsnjIpanlIsJQo`1OU~sVw`G zn+{vinS1r!!uAen4H5_f^lP?QRR|(g5=*Imy+W(0c{n8cqKhtQbaGe-S=D*TC}7Fb zJDS^iSGA-x6xmjlFMw{_SGiQN>cwLa5UU#Hr{8{7H!*7iNfA|q`fTP%X81tzyE4cs zO_jaH>3)j)_||RzetvQJqfmdIE?@ZUwtWv0tW-SyEw&h2=gQUbn`h@Bbv$wj$mJeM zwp182MXJSX^nVbr-`u)qNX;Sy$}9F*S%U8G=My~p*k#-9bVBv7B%qu z?$-}}_4P&XEm`r-`?(#L4=_)o=NvW&Rt1V?n!lEZ??;*zn znHp#T#8uU!@sCO1^Upo5egE^yM`X1FFd|#epvHCC&PKCsPpT|UeNQ^(+rPQ?+Q%&C zfg~g(0HTB()k~yXQ`9z}+)$*@NEa4zB!KTP?SemS22;SDw^yHG3LSU(BigC%LKs zFb7%XuOtX2;^}`Zscf+z(a*1v%ILc3nF&(s6WeGpR1C-!7vlxdlmJxyfAOoqO*O2I z2hkx_ENl~LtU{LS=uJ>Gwh)*CI4GR}P-|JM8R1o>sD!Du*7d7f4IceAF!D60_dV4b z0jng7PJB_S<5D4}v_m?BlCQoxbBl3}od%5J?A<%;di+yQzeA*j%JaTj7$Jnb^wdJv zoL(i1;CW=1&hwiliEo66>dl9W=Sq+mHucCuCSHB%=?&|%3aq)bI^#F<1C#e%of1y| z-chSqNX_HvJ5vDQ2(K^7joBm#XwRbmfMw^ma@DS{mOTQl zqj#4Gf=2G0MjE_PgW6uCw0<{ZRzZ?dp0E-ZlhuU+52jZ}u*+K)7`+lFh8=c3>hJ$r zL7-AvSQjiWzUr0diLHDl>AEFETzS>?Ri9U?gA5M+t-+{7qYllGhW3O-i+jklNv+LO zjUDH|`0UM*e);a=l`l*`@yNv_{Kn$0aa(?)f7$Ico7A2hJWzR1dvJw$V9}YqWo13; znrk1u__qT}FTCa*aprB0dz2Oem8Zv|#Q}nYCVgfV3!&;CGE`2Osd}&~;JQ1Wj#PXm zclHTYo5hO+jt&99qDa~zT|JG}%-*^wklK878+XnEk%&9;*bA?_>95+5tBuql_`~Yyi3SQT_aFcu zWO!cYD;mD-<0Vf)zO}egQ{Eo-?KNH3RhTxf|gL=HE6N} zSRkk#cvdO$01(0l@BG#7`<_wO3&MPlgVZ%}I*>|wT(OcRb(dGyE##*iT;1rfBM$#% z<=e9=X??{TNI*7scQJ~#)Np~_|G@7zyen6FxIm(HC<2kb_R1ezd*u%#3_J6tZOpXl zCF`;abk%bcgd`ed08arfdp`gtGHT;}mSt5{Y>@F+A*Vz`2HBT=rHJ9Pes!l*t5o}a zaqCl#-HCv@Gkk}@>7p%9OJGr*!p`_k#DEQ}dAAI^VqpAla=b4bG(iL=AVf7~h zN|pt(s(A``&Oz;)!IJkCN@XgpQrt0ba1%o*z$3C;?ViT=Z z{_z5xT~>~~^LE?i69ik?pZvsrE%HRkY#o_MJyviGP1cOH~R4FqfL<<~!X;-NJ) zVGAXG)o>uBi=V57tP)rW{Th3@HP1iW(J1=s|6Mry)U&THPf-OV9hK=VL7vy@!FA5o zExoJEsg>Hz=}0tL195#I6h zW0gjdlb_jY)ACUQw*2B5tJMSrb~qKKkh=m^#D}w$3d{EWJOdkX$sKXrg@84ToG$N#FpRqt;H@)G;inAQKa;17)yP~B%~FGrw6!_j@x zcc)ic(+G0UgU|Ij%Sufq+)|{CB26j}?j216s=p9flL-T?^s2Ox=*ps5~_1~DW9bC2h4=F%_$TEI<=8VXu4&aoZ&i}(tDxC=xLEZPKq65Lk6Mt5= z!G&Qe%6hVjtpdpN5Y?Z7@#%mfj zS-|5rOO4rRqXkdC$->n&B`V19w%)Sl-EzLlOs*7|e|_X-uetq_KOBF=k&`?iF>Fd8 zmZaj{sFV4wwUSWS0Na0N#NVEOuX0-;5{SQvZaxUT+a0;0=Th(Flu~|c=aOqGOCun^ z{3e~6L8jz{&FY;GpjxMWe(Q~|x$b|jy7Hr+(Z9S~GRd8)~jYP6_TE62J4Ff@eUE7F#^R7Q*c0yZ} zmVqPoopw&z>*`!`Cn+K*k&u+9TiPr8Rqs27aPu`k`s$HYUV+p+!*VS&Cj!7$BOgadp{0yN^a6-XJLbHe_vcgQxZ{QzO z+v|v+nXQrW@ys5FU#7@h@vE`51w&Z+7QS6_0{+8WA)Cl_2r-X2J%tE(>6%Tm>? z5)WkGeeY`B$F!;7Px$=kC;qv(;*f&4zdyI=)1N&gI-}+aacbG(zYMq~gF(^lfW=_Z zyKd3Dua;sPt62Q@wN&ZJ?5udED8^OZF{(4NCWRCby?%h`7Y;4(SAl_C;Z}_g{Q6VTsiHdvtw3M z(WiVBjkv5_nCrNERW@x-QX-FRlo+qx_X==(?ynFIVL9UGq|{(@9pv4}<}gTQO&V4}7glE5k`guJn|Y5-2bL z);$kgcO-!zCM@n(SaqQgMx?zJT~W^~zC?s0m$~X8EtquTOr?d`pMUY!qGOr?6o?4| zMFxN+1G15UGKh>>)evFV&kk>Hl3jP%Y1_|!{L`Datib}e2GRyW*R7eB_Hd>9l9APv z%5}^t?S#1Uep8?vcg{|=9a^Z{zs0=#@9VeNw2q&S20k`^MA$VinQE`+Bv-1DnD^hA zHKM+o#2`Y7JSjYu7qqq=9CCO6XDh4oxm!&?$^Qr(K6J&k_n+|1i@4vFz7{{dV(Cpl zxX!kf&_?$YkQ9d9J=uGbnW=sEsRjTsx6l9Sq}^-dY7;sUbj?pR?JsqGfD2lJDD=G| z+*x#Lc>D7A<|KUuz)R@d^4X)enEd{dmDE3(NN+i|_1UK`s_vgAp%8S;88-63o^U0u z5=xS``__SHKw0a#L&r=wyKc6BPw2IUKk4T}0T#X6@tt$4OeeE0J`8yk<_$`i zm7DX&BcpE_@R}zHwCkD`?Mbokk-yAyTluQHuRS_JvNqQlSOjCXnDqV!2_cXO6|)p7 zMEeU+KD+Fm)XKJOt;roKilt*dGrB`=8`%y{h_a#_B=Mc&_xQ~P$I;3nA4LR2I)MI~Gf2%TTePx)rZF>Hrq;`V|Z!(p|8_+V*9yg`}WLc3D;q zoPsR1-G6DcBSXF0+ZSG3zV&v;07}RKLIP0?(HXzc`TpIjzAtaU@N3?n#h};9%GXiQ z`d)|%x0~ABp+eF%CkO*liDcVkh!P@@6-n;7$7XlWxuBm6Tu$tS-Hw0indP87u^}Ol zvz@nFm-S2rs5P%Jz=G;u5n;gB-0<*m-~KH~ltinm3LuCA=oddb>Xeh)RXsyZxdpBe zG;MrvAy1%kf(jNTkd%*(&d&P>YYXzVF0D43_*1T@q8P^OkDs}ciy7_Ts>;!`+l0-& zfA}sn-bNPgfD^BcV!mu>@3{QPs0TS9(KPw`E6+XXpzQ!S;XRMIibXTIo}NOO>-gN} z8{W5I!P`$?ITVboji@WFhR9-W%ol|}EpyX^hW{Jqb83ua? zJJ4PBK02(Mo#j`L5EwV{+`h*>>U!<5UpBai1!~iZtH`Ai>DHl*PZK%nSPqrW{}Vun7t_-mr{gT}w+sxtJXl~e_#GCbuxVngip zyZaNg@pZKwUP5fO%MTqm-}%6MOY25q#l*y{Hw-ZUrk5 zI9WDn#RMI49?B?oc_fv9$cUqX>{}NT6Z4@n1npJ~5Jc7>awunnv6BDz3Fa$@h0%*;GJT;&Q1VvDQ1`L%l>jype-cqxJNK8!3dc_dbt=1Na z;ZqCCRWeDn4YYXQ-Dsnj{=D(7UkZuJG_P)b0F){0qkC1K!uhCT0!!Xw3xQ`(VZB4TgQ#wTOv{Sra~5(h3}HM-)4W#>A|De57WA z01<>%d7HnOn3y5NM@?GcmRoO)b*q?|^^uR*OfV)UX1(Pjrd!3t#H_c((kdn + 简体中文 + + +索引与表格 +================== + +* :ref:`genindex` +* :ref:`search` \ No newline at end of file diff --git a/docs/zh_cn/prompt/meta_template.md b/docs/zh_cn/prompt/meta_template.md new file mode 100644 index 00000000..18766379 --- /dev/null +++ b/docs/zh_cn/prompt/meta_template.md @@ -0,0 +1,163 @@ +# Meta Prompt + +## 背景 + +在 LLM 的实际 finetune 中，我们常常会根据实际的要求注入一些预定义的字符串，以求模型能按照自然语言的格式输出指定的内容。在评测时，我们也需要按照 finetune 时设定的格式输入问题，模型才能发挥出其最大的性能。因此，我们需要对 OpenICL 原本的 prompt 设计作一次增强，才能满足相应需求。 + +## Model - Meta Template + +此前， prompt template 的设定绑定在数据集配置中。现在考虑到不同模型的 instruction 可能会有所不同，我们往 model config 中新增 `meta_template` 字段，允许用户指定与模型密切相关的 instruction。 + +```Python +models = [ + dict(type='LLM', + # ... + meta_template = dict( + begin="meta instruction\nYou are an AI assistant.\n", + round=[ + dict(role='HUMAN', begin='<|HUMAN|>:', end='脷\n'), # begin and end can be a list of strings or integers. + dict(role='THOUGHTS', begin='<|Inner Thoughts|>:', end='茔\n', prompt='None'), + dict(role='COMMANDS', begin='<|Commands|>:', end='蝮\n', prompt='None'), + dict(role='RESULTS', begin='<|Results|>:', end='兒\n', prompt='None'), # Here we can set the default prompt, which may be overridden by the speicfic dataset + dict(role='BOT', begin='<|MOSS|>:', generate=True, end='氡\n'), + ], + end="end of conversion", + reserved_roles=[dict(role='SYSTEM', begin='<|SYSTEM|>: ', end='\n'),], + # the token to stop the generation tasks (TODO: support string) + eos_token_id=65605, + ), + ) +] +``` + +这里，meta_template 是一个**字典**，该字典可以包含以下数个字段： + +- `begin`，`end` ：(str，可选) prompt 的开头，通常是一些 meta instruction。 + +- `round`：(list，可选) 约定了每一轮对话的 prompt 格式。每轮对话的 prompt 内容由 dataset config 中的 prompt template 控制（下文会详述）。如果不指定，则该字段将会直接被 dataset config 中的 prompt template 替换。 + +- (str，可选)：收尾的 instruction。 + +- `reserved_roles` （list，可选）指定了在 meta template 中并未出现的预留角色。这里面定义的角色有可能在 dataset config 的 begin 或 end 中用到，例如 `SYSTEM` 角色。 + +- `eos_token_id` （int, 可选）：指定了该模型在生成式任务中 eos token 的 id。如果不设置，则默认为 tokenizer 中的 eos token id。 + +`round` 指定了每轮对话中每个角色说话的格式，通常接受一个列表，内容可以是 **str 或 dict**。每个字典接受以下关键字： + +- `role`（str）: 对话中的角色，也可以认为是这个 prompt 的 identifier。该字符串并不影响实际的 prompt，仅用于在 dataset_config 中的指定对应项，并对其 prompt 内容进行覆盖。 + +- `begin`, `end` (str): 指定该角色在说话时的开头或结尾。 + +- `prompt` (str)：prompt 的内容，遵循 `ICLPromptTemplate` 的格式规范。如果在 meta_prompt_template 中未指定，则必须在 dataset config 中的 prompt template 中指定。 + +- `generate` (bool): 指定为 True 时，该角色即为模型在生成任务中开始生成输出的位点。在生成任务中生成对应 prompt 时，prompt template 只会生成到该角色的 begin，剩下的内容由模型补全。 + +在上面的例子中，最后的 meta prompt 将会是： + +``` +meta instructionYou are an AI assistant. +<|HUMAN|>: 脷\n +<|Inner Thoughts|>: None茔\n<|Commands|>: None蝮\n<|Results|>: None兒\n +<|MOSS|>: 氡\n +end of conversion +``` + +特别地，在生成式任务中，prompt 仅会生成到 \<|MOSS|>: 后： + +``` +meta instructionYou are an AI assistant. +<|HUMAN|>: 脷\n +<|Inner Thoughts|>: None茔\n<|Commands|>: None蝮\n<|Results|>: None兒\n +<|MOSS|>: +``` + +接下来我们在 dataset config 中进行进一步约定。 + +## Dataset: Prompt Template + +在 model 配置中约定了该 model 所需的 meta template 后，dataset 中 prompt template 的格式也会有所变化。同时，该方向尽可能地保持了 prompt 的 backward compatibility。 + +在改动前，`PromptTemplate` 接受 str 或 dict 作为输入。其中，dict 形式的输入将 label string 映射到对应的 prompt (str)上，通常用作为 `PPLInferencer` 的输入。因而本质上，`PromptTemplate` 的旧版实现里表示 prompt 的方式只有 `str` 一种。 + +而改动后的 prompt template 允许接受的 prompt 基本形式从 str 扩展到了 dict。 + +这个 dict 的格式与 meta template 相似，用户也可以指定 `begin`, `end` 和 `round` 关键字： + +```Python +mmlu_prompt_template = dict( + type='PromptTemplate', + template=dict( + begin=[dict(role='SYSTEM', fallback_role='HUMAN', prompt='The following are ' + 'multiple choice questions (with answers) about physics.'), + '', + ], + round=[ + dict(role='HUMAN', prompt='\nA. \nB. \nC. \nD. \nAnswer: '), + dict(role='BOT', prompt=''), + ], + end="end of dataset prompt template." + ), + column_token_map={ + 'input': '', + 'A': '', + 'B': '', + 'C': '', + 'D': '', + 'target': '' + }, + ice_token='', + ) + +``` + +其中，`round`用于指定在每轮对话中角色的 prompt 格式，同时也是为了呼应和补全 meta template 中的配置，因此，其接受的参数和规则均与 meta template 中的 `round` 一致。**在实际运行时，两处 prompt 的配置将会融合，同时如果某一字段被重复定义，则以 dataset config 中定义为准。** + +而 `begin` 和 `end` 则除了支持 str 类型的输入，也支持 list 类型的输入，在其中用户可以通过组合 dict 和字符串实现对系统角色的融合。留意到例子中引入了 `fallback_role` 的设定，意味着若系统在 meta template 中 reserved_roles 中找不到 `role` 中的角色时，会自动替换成 `fallback_role` 中的角色。这个特征的设立是为了尽可能确保 prompt 模板的通用性。 + +结合 meta template，最终生成的 prompt 模板为： + +```Plain +meta instruction +You are an AI assistant. +<|SYSTEM|>: The following are multiple choice questions (with answers) about college biology. +<|HUMAN|>: Which of the following is NOT a characteristic of an oligotrophic lake? +A. Low nutrient levels +B. High altitudes +C. Shallow water +D. Sand or gravel bottom +Answer: 脷\n +<|Inner Thoughts|>: None茔\n +<|Commands|>: None蝮\n +<|Results|>: None兒\n +<|MOSS|>: A氡\n +end of dataset prompt template. +end of conversion +``` + +特别地，由于这种 prompt 的数据结构（dict）与旧版的 label -> prompt 映射相同，本实现仅在字典的 keys 为 {`begin`, `round`, `end`} 的子集时将 prompt 的输入以新版规则进行解码，否则依然将字典以 label -> prompt 的形式进行解码。此外，该方案也允许新版 prompt 字典嵌套在旧版的 label -> prompt 字典中。例如，以下表达方式也是合法的 （摘自 `configs/datasets/mmlu.py`）： + +```Python +prompt_template={ + target: + dict( + begin=[dict(role='SYSTEM', fallback_role='HUMAN', prompt='The following are ' + 'multiple choice questions (with answers) about ' + f'{name.replace("_", " ")}.\n'), + '', + ], + round=[ + dict(role='HUMAN', prompt='\nA. \nB. \nC. \nD. \nAnswer: '), + dict(role='BOT', prompt=f'{target}'), + ] + ) + for target in ['A', 'B', 'C', 'D'] # use the actual answer + } +``` + +### 无 meta template 时 + +为了保证后向兼容性，当用户未在 model config 中指定 meta template 时，`ICLPromptTemplate` 会将每个 dict 按照 `begin`, `prompt`, `end` 的顺序拼接为普通字符串。 + +### 多轮对话例子 + +在某些时候，一轮完整的交互中可能需要包含多轮对话。用户可以参考 `configs/datasets/gsm8k.py` 配置自己的模板。 diff --git a/docs/zh_cn/tools.md b/docs/zh_cn/tools.md new file mode 100644 index 00000000..0aa8338e --- /dev/null +++ b/docs/zh_cn/tools.md @@ -0,0 +1,66 @@ +# 实用工具 + +## Prompt Viewer + +本工具允许你在不启动完整训练流程的情况下，直接查看模型会接收到的 prompt。 + +运行方式： + +```bash +python tools/prompt_viewer.py [CONFIG_PATH] +``` + +## Case Analyzer + +本工具在已有评测结果的基础上，产出推理错误样本以及带有标注信息的全量样本 + +运行方式： + +```bash +python tools/case_analyzer.py [CONFIG_PATH] [-w WORK_DIR] +``` + +- `-w`：工作路径，默认为 `'./outputs/default'`。 + +更多细节见 [飞书文档](https://aicarrier.feishu.cn/docx/SgrLdwinion00Kxkzh2czz29nIh) + +## Lark Bot + +用户可以通过配置飞书机器人，实现任务状态的实时监控。飞书机器人的设置文档请[参考这里](https://open.feishu.cn/document/ukTMukTMukTM/ucTM5YjL3ETO24yNxkjN?lang=zh-CN#7a28964d)。 + +配置方式: + +- 打开 `configs/secrets.py` 文件，并在文件中加入以下行： + + ```python + lark_bot_url = 'YOUR_WEBHOOK_URL' + ``` + + 通常， Webhook URL 格式如 https://open.feishu.cn/open-apis/bot/v2/hook/xxxxxxxxxxxxxxxxx 。 + +- 在完整的评测配置中继承该文件： + + ```python + _base_ = [ + 'secrets.py', + ... + ] + ``` + + 实例可见 `configs/eval.py`。 + +- 为了避免机器人频繁发消息形成骚扰，默认运行时状态不会自动上报。有需要时，可以通过 `-l` 或 `--lark` 启动状态上报： + + ```bash + python run.py configs/eval_demo.py -p {PARTITION} -l + ``` + +## API Model Tests + +本工具可以快速测试 API Wrapper 的功能是否正常。 + +运行方式： + +```bash +python tools/test_api_model.py [CONFIG_PATH] +``` diff --git a/docs/zh_cn/user_guides/dataset_prepare.md b/docs/zh_cn/user_guides/dataset_prepare.md new file mode 100644 index 00000000..e0637bc6 --- /dev/null +++ b/docs/zh_cn/user_guides/dataset_prepare.md @@ -0,0 +1,105 @@ +# 数据集准备和选择 + +本节教程主要关注如何准备 OpenCompass 已支持的数据集，并构建需要的配置文件完成数据集的选择。 + +## 数据集配置文件目录结构 + +首先简单介绍一下 OpenCompass `configs/datasets` 目录下的结构，如下所示： + +``` +configs/datasets/ +├── ChineseUniversal # 能力维度 +│   ├── CLUE_afqmc # 该维度下的数据集 +│   │   ├── CLUE_afqmc_gen_db509b.py # 该数据集的不同配置文件 +│   │   ├── CLUE_afqmc_gen.py +│   │   ├── CLUE_afqmc_ppl_00b348.py +│   │   ├── CLUE_afqmc_ppl_2313cf.py +│   │   └── CLUE_afqmc_ppl.py +│   ├── CLUE_C3 +│   │   ├── ... +│   ├── ... +├── Coding +├── collections +├── Completion +├── EnglishUniversal +├── Exam +├── glm +├── LongText +├── MISC +├── NLG +├── QA +├── Reasoning +├── Security +└── Translation +``` + +在 `configs/datasets` 目录结构下，我们主要以能力维度对数据集划分了十余项维度，例如：中英文通用、考试、问答、推理、安全等等。每一项维度又包含了一系列数据集，在各个数据集对应的文件夹下存在多个数据集配置。 + +数据集配置文件名由以下命名方式构成 `{数据集名称}_{评测方式}_{prompt版本号}.py`，以 `ChineseUniversal/CLUE_afqmc/CLUE_afqmc_gen_db509b.py` 为例，该配置文件则为中文通用能力下的 `CLUE_afqmc` 数据集，对应的评测方式为 `gen`，即生成式评测，对应的prompt版本号为 `db509b`；同样的， `CLUE_afqmc_ppl_00b348.py` 指评测方式为`ppl`即判别式评测，prompt版本号为 `00b348` 。 + +除此之外，不带版本号的文件，例如： `CLUE_afqmc_gen.py` 则指向该评测方式最新的prompt配置文件，通常来说会是精度最高的prompt。 + +## 数据集准备 + +OpenCompass 支持的数据集主要包括两个部分： + +1. Huggingface 数据集 + +[Huggingface Dataset](https://huggingface.co/datasets) 提供了大量的数据集。OpenCompass 已经支持了大多数常用于性能比较的数据集，具体支持的数据集列表请直接在 `configs/dataset` 下进行查找。 + +2. OpenCompass 自建数据集 + +除了支持 Huggingface 已有的数据集， OpenCompass 还提供了一些自建CN数据集，未来将会提供一个数据集相关的Repo供用户下载使用。按照文档指示将数据集统一放置在`./data`目录下即可完成数据集准备。 + +需要注意的是，Repo中不仅包含自建的数据集，为了方便也加入了部分HF已支持的数据集方便测试。 + +## 数据集选择 + +在各个数据集配置文件中，数据集将会被定义在 `{}_datasets` 变量当中，例如下面 `ChineseUniversal/CLUE_afqmc/CLUE_afqmc_gen_db509b.py` 中的 `afqmc_datasets`。 + +```python +afqmc_datasets = [ + dict( + abbr="afqmc-dev", + type=AFQMCDataset_V2, + path="./data/CLUE/AFQMC/dev.json", + reader_cfg=afqmc_reader_cfg, + infer_cfg=afqmc_infer_cfg, + eval_cfg=afqmc_eval_cfg, + ), +] +``` + +以及 `ChineseUniversal/CLUE_cmnli/CLUE_cmnli_ppl_b78ad4.py` 中的 `afqmc_datasets`。 + +```python +cmnli_datasets = [ + dict( + type=HFDataset, + abbr='cmnli', + path='json', + split='train', + data_files='./data/CLUE/cmnli/cmnli_public/dev.json', + reader_cfg=cmnli_reader_cfg, + infer_cfg=cmnli_infer_cfg, + eval_cfg=cmnli_eval_cfg) +] +``` + +以上述两个数据集为例， 如果用户想同时评测这两个数据集，可以在 `configs` 目录下新建一个配置文件，我们使用 `mmengine` 配置中直接import的机制来构建数据集部分的参数，如下所示： + +```python +from mmengine.config import read_base + +with read_base(): + from .datasets.CLUE_afqmc.CLUE_afqmc_gen_db509b import afqmc_datasets + from .datasets.CLUE_cmnli.CLUE_cmnli_ppl_b78ad4 import cmnli_datasets + +datasets = [] +datasets += afqmc_datasets +datasets += cmnli_datasets +``` + +用户可以根据需要，选择不同能力不同数据集以及不同评测方式的配置文件来构建评测脚本中数据集的部分。 + +有关如何启动评测任务，以及如何评测自建数据集可以参考相关文档。 diff --git a/opencompass/datasets/GaokaoBench.py b/opencompass/datasets/GaokaoBench.py new file mode 100644 index 00000000..08c411e4 --- /dev/null +++ b/opencompass/datasets/GaokaoBench.py @@ -0,0 +1,132 @@ +import json +import re + +from datasets import Dataset + +from opencompass.openicl.icl_evaluator import BaseEvaluator +from opencompass.registry import ICL_EVALUATORS, LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class GaokaoBenchDataset(BaseDataset): + + @staticmethod + def load(path: str): + with open(path) as f: + data = json.load(f) + return Dataset.from_list(data['example']) + + +valid_gaokao_bench_question_types = [ + 'single_choice', 'multi_choice', 'multi_question_choice', + 'five_out_of_seven', 'cloze', 'subjective', 'correction' +] + + +class GaokaoBenchEvaluator(BaseEvaluator): + + def __init__(self, question_type) -> None: + super().__init__() + assert question_type in valid_gaokao_bench_question_types + self.question_type = question_type + + def do_predictions_postprocess(self, model_output, answer_lenth=None): + if self.question_type == 'single_choice': + model_answer = [] + temp = re.findall(r'[A-D]', model_output[::-1]) + if len(temp) != 0: + model_answer.append(temp[0]) + + elif self.question_type == 'multi_question_choice': + model_answer = [] + temp = re.findall(r'【答案】\s*[:：]*\s*[A-Z]', model_output) + + if len(temp) == answer_lenth: + for t in temp: + model_answer.append(re.findall(r'[A-Z]', t)[0]) + else: + temp = re.findall(r'[A-Z]', model_output) + if len(temp) > 0: + for k in range(min(len(temp), answer_lenth)): + model_answer.append(temp[k]) + + elif self.question_type == 'multi_choice': + model_answer = [] + answer = '' + content = re.sub(r'\s+', '', model_output) + answer_index = content.find('【答案】') + if answer_index > 0: + temp = content[answer_index:] + if len(re.findall(r'[A-D]', temp)) > 0: + for t in re.findall(r'[A-D]', temp): + answer += t + else: + temp = content[-10:] + if len(re.findall(r'[A-D]', temp)) > 0: + for t in re.findall(r'[A-D]', temp): + answer += t + if len(answer) != 0: + model_answer.append(answer) + + elif self.question_type == 'five_out_of_seven': + model_answer = [] + temp = re.findall(r'[A-G]', model_output) + if len(temp) > 0: + for k in range(min(5, len(temp))): + model_answer.append(temp[k]) + + return model_answer + + def ensure_same_length(self, pred, refr): + if len(pred) == len(refr): + return pred + return ['Z'] * len(refr) + + def score(self, predictions, references): + if self.question_type not in [ + 'single_choice', 'multi_choice', 'multi_question_choice', + 'five_out_of_seven' + ]: + return {'score': 0} + elif self.question_type == 'multi_choice': + correct_score, total_score = 0, 0 + for pred, refr in zip(predictions, references): + pred = self.do_predictions_postprocess(pred) + pred = self.ensure_same_length(pred, refr) + for p, r in zip(pred, refr): + if p == r: + correct_score += 2 + else: + for i in p: + if i not in r: + break + else: + correct_score += 1 + total_score += 2 + return {'score': correct_score / total_score * 100} + else: + correct_score, total_score = 0, 0 + for pred, refr in zip(predictions, references): + if self.question_type == 'multi_question_choice': + pred = self.do_predictions_postprocess(pred, len(refr)) + else: + pred = self.do_predictions_postprocess(pred) + pred = self.ensure_same_length(pred, refr) + for p, r in zip(pred, refr): + if p == r: + correct_score += 1 + total_score += 1 + return {'score': correct_score / total_score * 100} + + +for question_type in valid_gaokao_bench_question_types: + # fix classic closure problem + def _gaokao_register(question_type): + ICL_EVALUATORS.register_module( + name='GaokaoBenchEvaluator' + '_' + question_type, + module=lambda *args, **kwargs: GaokaoBenchEvaluator( + question_type=question_type, *args, **kwargs)) + + _gaokao_register(question_type) diff --git a/opencompass/datasets/afqmcd.py b/opencompass/datasets/afqmcd.py new file mode 100644 index 00000000..11a15ab2 --- /dev/null +++ b/opencompass/datasets/afqmcd.py @@ -0,0 +1,21 @@ +import json + +from datasets import Dataset + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class AFQMCDataset_V2(BaseDataset): + + @staticmethod + def load(path): + data = [] + with open(path, 'r') as f: + for line in f: + line = json.loads(line) + line['label'] = 'AB'[int(line['label'])] + data.append(line) + return Dataset.from_list(data) diff --git a/opencompass/datasets/ax.py b/opencompass/datasets/ax.py new file mode 100644 index 00000000..38656864 --- /dev/null +++ b/opencompass/datasets/ax.py @@ -0,0 +1,24 @@ +import json + +from datasets import Dataset + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class AXDataset_V2(BaseDataset): + + @staticmethod + def load(path: str): + dataset = [] + with open(path, 'r') as f: + for line in f: + line = json.loads(line) + line['label'] = { + 'entailment': 'A', + 'not_entailment': 'B' + }[line['label']] + dataset.append(line) + return Dataset.from_list(dataset) diff --git a/opencompass/datasets/base.py b/opencompass/datasets/base.py new file mode 100644 index 00000000..5412ef4c --- /dev/null +++ b/opencompass/datasets/base.py @@ -0,0 +1,28 @@ +from abc import abstractstaticmethod +from typing import Dict, Optional, Union + +from datasets import Dataset, DatasetDict + +from opencompass.openicl import DatasetReader + + +class BaseDataset: + + def __init__(self, reader_cfg: Optional[Dict] = {}, **kwargs): + self.dataset = self.load(**kwargs) + self._init_reader(**reader_cfg) + + def _init_reader(self, **kwargs): + self.reader = DatasetReader(self.dataset, **kwargs) + + @property + def train(self): + return self.reader.dataset['train'] + + @property + def test(self): + return self.reader.dataset['test'] + + @abstractstaticmethod + def load(**kwargs) -> Union[Dataset, DatasetDict]: + pass diff --git a/opencompass/datasets/bbh.py b/opencompass/datasets/bbh.py new file mode 100644 index 00000000..e803ca9e --- /dev/null +++ b/opencompass/datasets/bbh.py @@ -0,0 +1,71 @@ +import json +import os.path as osp +import re + +from datasets import Dataset + +from opencompass.openicl.icl_evaluator import BaseEvaluator +from opencompass.registry import (ICL_EVALUATORS, LOAD_DATASET, + TEXT_POSTPROCESSORS) + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class BBHDataset(BaseDataset): + + @staticmethod + def load(path: str, name: str): + with open(osp.join(path, f'{name}.json'), 'r') as f: + data = json.load(f)['examples'] + dataset = Dataset.from_list(data) + return dataset + + +@TEXT_POSTPROCESSORS.register_module('bbh-mcq') +def bbh_mcq_postprocess(text: str) -> str: + ans = text + ans_line = ans.split('answer is ') + if len(ans_line) != 1: + ans = ans_line[1].strip() + match = re.search(r'\(([A-Z])\)*', ans) + if match: + return match.group(1) + match = re.search(r'([A-Z])', ans) + if match: + return match.group(1) + return ans + + +@TEXT_POSTPROCESSORS.register_module('bbh-freeform') +def bbh_freeform_postprocess(text: str) -> str: + ans = text + ans_line = ans.split('answer is ') + if len(ans_line) != 1: + ans = ans_line[1].strip() + ans = ans.split('\n')[0] + if ans.endswith('.'): + ans = ans[:-1] + return ans + + +@ICL_EVALUATORS.register_module() +class BBHEvaluator(BaseEvaluator): + + def score(self, predictions, references): + if len(predictions) != len(references): + return { + 'error': 'predictions and references have different ' + 'length' + } + + predictions = [bbh_freeform_postprocess(pred) for pred in predictions] + + cnt = 0 + for pred, ref in zip(predictions, references): + if pred == ref: + cnt += 1 + + score = cnt / len(predictions) * 100 + + return {'score': score} diff --git a/opencompass/datasets/boolq.py b/opencompass/datasets/boolq.py new file mode 100644 index 00000000..d0eb576f --- /dev/null +++ b/opencompass/datasets/boolq.py @@ -0,0 +1,41 @@ +import json + +from datasets import Dataset, load_dataset + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class BoolQDataset(BaseDataset): + + @staticmethod + def load(**kwargs): + + dataset = load_dataset(**kwargs) + + def preprocess(example): + if example['label'] == 'true': + example['answer'] = 1 + else: + example['answer'] = 0 + + return example + + dataset = dataset.map(preprocess) + return dataset + + +@LOAD_DATASET.register_module() +class BoolQDataset_V2(BaseDataset): + + @staticmethod + def load(path): + dataset = [] + with open(path, 'r') as f: + for line in f: + line = json.loads(line) + line['label'] = {'true': 'A', 'false': 'B'}[line['label']] + dataset.append(line) + return Dataset.from_list(dataset) diff --git a/opencompass/datasets/bustum.py b/opencompass/datasets/bustum.py new file mode 100644 index 00000000..d145f4f9 --- /dev/null +++ b/opencompass/datasets/bustum.py @@ -0,0 +1,21 @@ +import json + +from datasets import Dataset + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class bustumDataset_V2(BaseDataset): + + @staticmethod + def load(path): + data = [] + with open(path, 'r') as f: + for line in f: + line = json.loads(line) + line['label'] = 'AB'[int(line['label'])] + data.append(line) + return Dataset.from_list(data) diff --git a/opencompass/datasets/cmrc.py b/opencompass/datasets/cmrc.py new file mode 100644 index 00000000..ccd83494 --- /dev/null +++ b/opencompass/datasets/cmrc.py @@ -0,0 +1,40 @@ +import json + +from datasets import Dataset + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class CMRCDataset(BaseDataset): + + @staticmethod + def load(path: str): + with open(path) as f: + data = json.load(f) + # 将原始数据转换为所需的格式 + rows = [] + for index, paragraphs in enumerate(data['data']): + for paragraph in paragraphs['paragraphs']: + + context = paragraph['context'] + + for question in paragraph['qas']: + answers = question['answers'] + unique_answers = list(set([a['text'] for a in answers])) + rows.append({ + 'context': context, + 'question': question['question'], + 'answers': unique_answers + }) + + # 创建 Dataset + dataset = Dataset.from_dict({ + 'context': [row['context'] for row in rows], + 'question': [row['question'] for row in rows], + 'answers': [row['answers'] for row in rows] + }) + + return dataset diff --git a/opencompass/datasets/govrepcrs.py b/opencompass/datasets/govrepcrs.py new file mode 100644 index 00000000..4356eda6 --- /dev/null +++ b/opencompass/datasets/govrepcrs.py @@ -0,0 +1,37 @@ +from datasets import Dataset, DatasetDict + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class GovRepcrsDataset(BaseDataset): + + @staticmethod + def load(path: str): + import json + import os + dataset_dict = DatasetDict() + splits = ['train', 'valid', 'test'] + dataset_lists = {x: [] for x in splits} + for split in splits: + split_fp = os.path.join(path, 'gov-report', 'split_ids', + 'crs_' + split + '.ids') + with open(split_fp, 'r') as f: + for line in f.readlines(): + xpath = os.path.join(path, 'gov-report', 'crs', + line.strip() + '.json') + with open(xpath, 'r') as df: + data = json.load(df) + content = data['title'] + '\n' + '\n'.join( + [(x['section_title'] if x['section_title'] else '') + + '\n' + '\n'.join(x['paragraphs']) + for x in data['reports']['subsections']]) + summary = '\n'.join(data['summary']) + dataset_lists[split].append({ + 'content': content, + 'summary': summary, + }) + dataset_dict[split] = Dataset.from_list(dataset_lists[split]) + return dataset_dict diff --git a/opencompass/datasets/iwslt2017.py b/opencompass/datasets/iwslt2017.py new file mode 100644 index 00000000..21a5f2d5 --- /dev/null +++ b/opencompass/datasets/iwslt2017.py @@ -0,0 +1,16 @@ +from datasets import load_dataset + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class IWSLT2017Dataset(BaseDataset): + + @staticmethod + def load(**kwargs): + dataset = load_dataset(**kwargs) + dataset = dataset.map(lambda example: example['translation'] + ).remove_columns('translation') + return dataset diff --git a/opencompass/datasets/narrativeqa.py b/opencompass/datasets/narrativeqa.py new file mode 100644 index 00000000..4cdbe1c6 --- /dev/null +++ b/opencompass/datasets/narrativeqa.py @@ -0,0 +1,43 @@ +from datasets import Dataset, DatasetDict + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class NarrativeQADataset(BaseDataset): + + @staticmethod + def load(path: str): + import csv + import os + dataset_dict = DatasetDict() + splits = ['train', 'valid', 'test'] + dataset_lists = {x: [] for x in splits} + with open(os.path.join(path, 'qaps.csv'), 'r') as f: + reader = csv.reader(f, delimiter=',') + for row in reader: + if row[1] == 'set': + continue + split = row[1] # set + answers = [row[3], row[4]] # row['answer1'], row['answer2'] + question = row[2] # question + x_path = os.path.join(path, 'tmp', + row[0] + '.content') # document_id + + try: + with open(x_path, 'r', encoding='utf-8') as f: + evidence = f.read(100000) + except: # noqa: E722 + continue + dataset_lists[split].append({ + 'answer': answers, + 'question': question, + 'evidence': evidence, + }) + + for split in splits: + dataset_dict[split] = Dataset.from_list(dataset_lists[split]) + + return dataset_dict diff --git a/opencompass/datasets/wsc.py b/opencompass/datasets/wsc.py new file mode 100644 index 00000000..398ce4d7 --- /dev/null +++ b/opencompass/datasets/wsc.py @@ -0,0 +1,58 @@ +import json + +from datasets import Dataset, load_dataset + +from opencompass.registry import LOAD_DATASET + +from .base import BaseDataset + + +@LOAD_DATASET.register_module() +class WSCDataset(BaseDataset): + + @staticmethod + def load(**kwargs): + + dataset = load_dataset(**kwargs) + + def preprocess(example): + text_list = example['text'].split(' ') + assert ' ' not in example['target']['span2_text'] + # span1 may have 1 or more than 1 words + # span2 is the pronoun and has only 1 word + text_list[example['target'] + ['span2_index']] = example['target']['span1_text'] + example['new_text'] = ' '.join(text_list) + if example['label'] == 'true': + example['answer'] = 1 + else: + example['answer'] = 0 + example['span1'] = example['target']['span1_text'] + example['span2'] = example['target']['span2_text'] + del example['target'] + return example + + dataset = dataset.map(preprocess) + return dataset + + +@LOAD_DATASET.register_module() +class WSCDataset_V2(BaseDataset): + + @staticmethod + def load(path): + data = [] + with open(path, 'r') as f: + for line in f: + line = json.loads(line) + item = { + 'span1': line['target']['span1_text'], + 'span2': line['target']['span2_text'], + 'text': line['text'], + 'label': { + 'true': 'A', + 'false': 'B' + }[line['label']], + } + data.append(item) + return Dataset.from_list(data) diff --git a/opencompass/models/openai_api.py b/opencompass/models/openai_api.py new file mode 100644 index 00000000..cc5e1f8d --- /dev/null +++ b/opencompass/models/openai_api.py @@ -0,0 +1,154 @@ +import os +from concurrent.futures import ThreadPoolExecutor +from typing import Dict, List, Optional, Union + +from opencompass.registry import MODELS +from opencompass.utils.prompt import PromptList + +from .base_api import BaseAPIModel + +PromptType = Union[PromptList, str] + + +@MODELS.register_module() +class OpenAI(BaseAPIModel): + """Model wrapper around OpenAI's models. + + Args: + path (str): The name of OpenAI's model. + max_seq_len (int): The maximum allowed sequence length of a model. + Note that the length of prompt + generated tokens shall not exceed + this value. Defaults to 2048. + query_per_second (int): The maximum queries allowed per second + between two consecutive calls of the API. Defaults to 1. + retry (int): Number of retires if the API call fails. Defaults to 2. + key (str): OpenAI key. In particular, when it is set to "ENV", the key + will be fetched from the environment variable $OPENAI_API_KEY, as + how openai defaults to be. Defaults to 'ENV' + meta_template (Dict, optional): The model's meta prompt + template if needed, in case the requirement of injecting or + wrapping of any meta instructions. + openai_api_base (str): The base url of OpenAI's API. Defaults to + 'https://api.openai.com/v1'. + """ + + is_api: bool = True + + def __init__(self, + path: str, + max_seq_len: int = 2048, + query_per_second: int = 1, + retry: int = 2, + key: str = 'ENV', + meta_template: Optional[Dict] = None, + openai_api_base: str = 'https://api.openai.com/v1'): + super().__init__(path=path, + max_seq_len=max_seq_len, + meta_template=meta_template, + query_per_second=query_per_second, + retry=retry) + import openai + import tiktoken + self.openai = openai + self.tiktoken = tiktoken + + self.openai.api_key = os.getenv( + 'OPENAI_API_KEY') if key == 'ENV' else key + self.openai.api_rase = openai_api_base + + def generate( + self, + inputs: List[str or PromptList], + max_out_len: int = 512, + temperature: float = 0.7, + ) -> List[str]: + """Generate results given a list of inputs. + + Args: + inputs (List[str or PromptList]): A list of strings or PromptDicts. + The PromptDict should be organized in OpenCompass' + API format. + max_out_len (int): The maximum length of the output. + temperature (float): What sampling temperature to use, + between 0 and 2. Higher values like 0.8 will make the output + more random, while lower values like 0.2 will make it more + focused and deterministic. Defaults to 0.7. + + Returns: + List[str]: A list of generated strings. + """ + with ThreadPoolExecutor() as executor: + results = list( + executor.map(self._generate, inputs, + [max_out_len] * len(inputs), + [temperature] * len(inputs))) + return results + + def _generate(self, input: str or PromptList, max_out_len: int, + temperature: float) -> str: + """Generate results given a list of inputs. + + Args: + inputs (str or PromptList): A string or PromptDict. + The PromptDict should be organized in OpenCompass' + API format. + max_out_len (int): The maximum length of the output. + temperature (float): What sampling temperature to use, + between 0 and 2. Higher values like 0.8 will make the output + more random, while lower values like 0.2 will make it more + focused and deterministic. + + Returns: + str: The generated string. + """ + assert isinstance(input, (str, PromptList)) + + # max num token for gpt-3.5-turbo is 4097 + max_out_len = min(max_out_len, 4000 - self.get_token_len(str(input))) + + if isinstance(input, str): + messages = [{'role': 'user', 'content': input}] + else: + messages = [] + for item in input: + msg = {'content': item['prompt']} + if item['role'] == 'HUMAN': + msg['role'] = 'user' + elif item['role'] == 'BOT': + msg['role'] = 'assistant' + elif item['role'] == 'SYSTEM': + msg['role'] = 'system' + messages.append(msg) + + max_num_retries = 0 + while max_num_retries < self.retry: + self.wait() + try: + response = self.openai.ChatCompletion.create( + model=self.path, + messages=messages, + max_tokens=max_out_len, + n=1, + stop=None, + temperature=temperature, + ) + except self.openai.error.RateLimitError: + max_num_retries -= 1 + max_num_retries += 1 + + result = response.choices[0].message.content.strip() + return result + + def get_token_len(self, prompt: str) -> int: + """Get lengths of the tokenized string. Only English and Chinese + characters are counted for now. Users are encouraged to override this + method if more accurate length is needed. + + Args: + prompt (str): Input string. + + Returns: + int: Length of the input tokens + """ + enc = self.tiktoken.encoding_for_model(self.path) + return len(enc.encode(prompt)) diff --git a/opencompass/openicl/icl_inferencer/icl_base_inferencer.py b/opencompass/openicl/icl_inferencer/icl_base_inferencer.py new file mode 100644 index 00000000..70f80dfe --- /dev/null +++ b/opencompass/openicl/icl_inferencer/icl_base_inferencer.py @@ -0,0 +1,162 @@ +"""Basic Inferencer.""" +import json +import os +from pathlib import Path +from typing import List, Optional + +import numpy as np +from mmengine.dist import is_main_process +from torch.utils.data import DataLoader + +from ..icl_prompt_template import PromptTemplate +from ..icl_retriever import BaseRetriever + + +class BaseInferencer: + """Base Inferencer class for all evaluation Inferencer. + + Attributes: + model (:obj:`BaseModel`, optional): The module to inference. + max_model_token_num (:obj:`int`, optional): Maximum number of + tokenized words allowed by the LM. + batch_size (:obj:`int`, optional): Batch size for the + :obj:`DataLoader`. + output_json_filepath (:obj:`str`, optional): File path for output + `JSON` file. + output_json_filename (:obj:`str`, optional): File name for output + `JSON` file. + api_name (:obj:`str`, optional): Name of API service. + call_api (:obj:`bool`): If ``True``, an API for LM models will be used, + determined by :obj:`api_name`. + """ + model = None + + def __init__( + self, + model, + max_seq_len: Optional[int] = None, + batch_size: Optional[int] = 1, + output_json_filepath: Optional[str] = './icl_inference_output', + output_json_filename: Optional[str] = 'predictions', + **kwargs, + ) -> None: + self.model = model + + self.max_seq_len = max_seq_len + self.batch_size = batch_size + self.output_json_filepath = output_json_filepath + self.output_json_filename = output_json_filename + self.is_main_process = is_main_process() + if not os.path.exists(self.output_json_filepath): + os.makedirs(self.output_json_filepath) + + def inference(self, + retriever: BaseRetriever, + ice_template: Optional[PromptTemplate] = None, + prompt_template: Optional[PromptTemplate] = None, + output_json_filepath: Optional[str] = None, + output_json_filename: Optional[str] = None) -> List: + """Perform In-Context Inference given a retriever and optional + templates. + + Args: + retriever (:obj:`BaseRetriever`): An instance of a Retriever class + that will be used to retrieve in-context examples + ice_template (:obj:`PromptTemplate`, optional): A template for + generating the in-context examples prompt. Defaults to None. + prompt_template (:obj:`PromptTemplate`, optional): A template for + generating the final prompt. Defaults to None. + output_json_filepath (:obj:`str`, optional): The file path to save + the results as a `JSON` file. Defaults to None. + output_json_filename (:obj:`str`, optional): The file name to save + the results as a `JSON` file. Defaults to None. + + Raises: + NotImplementedError: If the function is not implemented in the + subclass. + + Returns: + :obj:`List:` A list of string, each representing the results of one + inference. + """ + raise NotImplementedError("Method hasn't been implemented yet") + + @staticmethod + def get_dataloader(datalist: List[List], batch_size: int) -> DataLoader: + """Return a dataloader of the input data list.""" + dataloader = DataLoader(datalist, + batch_size=batch_size, + collate_fn=lambda x: x) + return dataloader + + +def dump_results_dict(results_dict, filename): + with open(filename, 'w', encoding='utf-8') as json_file: + json.dump(results_dict, json_file, indent=4, ensure_ascii=False) + + +class GenInferencerOutputHandler: + origin_prompt_dict = {} + output_dict = {} + prediction_dict = {} + results_dict = {} + + def __init__(self) -> None: + self.results_dict = {} + + def write_to_json(self, save_dir: str, filename: str): + """Dump the result to a json file.""" + dump_results_dict(self.results_dict, Path(save_dir) / filename) + + def save_results(self, origin_prompt, prediction, idx): + self.results_dict[str(idx)] = { + 'origin_prompt': origin_prompt, + 'prediction': prediction, + } + + +class PPLInferencerOutputHandler: + results_dict = {} + + def __init__(self) -> None: + self.results_dict = {} + + def write_to_json(self, save_dir: str, filename: str): + """Dump the result to a json file.""" + dump_results_dict(self.results_dict, Path(save_dir) / filename) + + def save_ice(self, ice): + for idx, example in enumerate(ice): + if str(idx) not in self.results_dict.keys(): + self.results_dict[str(idx)] = {} + self.results_dict[str(idx)]['in-context examples'] = example + + def save_predictions(self, predictions): + for idx, prediction in enumerate(predictions): + if str(idx) not in self.results_dict.keys(): + self.results_dict[str(idx)] = {} + self.results_dict[str(idx)]['prediction'] = prediction + + def save_prompt_and_ppl(self, label, input, prompt, ppl, idx): + if str(idx) not in self.results_dict.keys(): + self.results_dict[str(idx)] = {} + if 'label: ' + str(label) not in self.results_dict[str(idx)].keys(): + self.results_dict[str(idx)]['label: ' + str(label)] = {} + self.results_dict[str(idx)]['label: ' + + str(label)]['testing input'] = input + self.results_dict[str(idx)]['label: ' + str(label)]['prompt'] = prompt + self.results_dict[str(idx)]['label: ' + str(label)]['PPL'] = ppl + + def save_prompt_and_condprob(self, input, prompt, cond_prob, idx, choices): + if str(idx) not in self.results_dict.keys(): + self.results_dict[str(idx)] = {} + # TODO: + # for single token situation, the input will always be yes currently + self.results_dict[str(idx)]['testing input'] = input + self.results_dict[str(idx)]['prompt'] = prompt + # TODO: hard code here + self.results_dict[str(idx)]['choices'] = choices + # For calculate auc scores, set scores as prediction + self.results_dict[str(idx)]['prediction'] = cond_prob + # set pred label in case needed + self.results_dict[str(idx)]['pred_label'] = int(np.argmax(cond_prob)) diff --git a/opencompass/openicl/icl_retriever/icl_mdl_retriever.py b/opencompass/openicl/icl_retriever/icl_mdl_retriever.py new file mode 100644 index 00000000..1bd32463 --- /dev/null +++ b/opencompass/openicl/icl_retriever/icl_mdl_retriever.py @@ -0,0 +1,186 @@ +"""MDL Retriever.""" + +from typing import List, Optional + +import numpy as np +import torch +import tqdm +from transformers import AutoModelForCausalLM + +from opencompass.openicl import PromptTemplate +from opencompass.openicl.icl_retriever.icl_topk_retriever import TopkRetriever +from opencompass.openicl.utils.logging import get_logger +from opencompass.registry import ICL_PROMPT_TEMPLATES, ICL_RETRIEVERS + +logger = get_logger(__name__) + + +@ICL_RETRIEVERS.register_module() +class MDLRetriever(TopkRetriever): + """MDL Retriever, subclass of `TopkRetriever`. MDL is a abbreviation of + Minimum Description Length, specially designed for ppl evaluation. You may + refer to the paper for more details: https://arxiv.org/pdf/2212.10375.pdf. + + Args: + dataset (`BaseDataset`): Any BaseDataset instances. + Attributes of ``reader``, ``train`` and ``test`` will be used. + ice_separator (`Optional[str]`): The separator between each in-context + example template when origin `PromptTemplate` is provided. Defaults + to '\n'. + ice_eos_token (`Optional[str]`): The end of sentence token for + in-context example template when origin `PromptTemplate` is + provided. Defaults to '\n'. + ice_num (`Optional[int]`): The number of in-context example template + when origin `PromptTemplate` is provided. Defaults to 1. + sentence_transformers_model_name (`Optional[str]`): The name of the + sentence transformers model. Defaults to 'all-mpnet-base-v2'. + tokenizer_name (`Optional[str]`): The name of the tokenizer. Defaults + to 'gpt2-xl'. + batch_size (`Optional[int]`): The batch size for the dataloader. + Defaults to 1. + candidate_num (`Optional[int]`): The number of candidates to retrieve + for each example. Defaults to 1. + ce_model_name (`Optional[str]`): The name of the model for calculating + MDL. Defaults to 'gpt2-xl'. + select_time (`Optional[int]`): The number of times to select MDL. + Defaults to 5. + ice_template (`Optional[PromptTemplate]`): The template for in-context + example. Defaults to None. + prompt_template (`Optional[PromptTemplate]`): The template for prompt. + Defaults to None. + labels (`Optional[List]`): The labels for calculating MDL. Defaults to + None. + seed (`Optional[int]`): The seed for random. Defaults to 1. + """ + metric_model = None + + def __init__(self, + dataset, + ice_separator: Optional[str] = '\n', + ice_eos_token: Optional[str] = '\n', + ice_num: Optional[int] = 1, + sentence_transformers_model_name: Optional[ + str] = 'all-mpnet-base-v2', + tokenizer_name: Optional[str] = 'gpt2-xl', + batch_size: Optional[int] = 1, + candidate_num: Optional[int] = 1, + ce_model_name: Optional[str] = 'gpt2-xl', + select_time: Optional[int] = 5, + ice_template: Optional[PromptTemplate] = None, + prompt_template: Optional[PromptTemplate] = None, + labels: Optional[List] = None, + seed: Optional[int] = 1) -> None: + super().__init__(dataset, ice_separator, ice_eos_token, ice_num, + sentence_transformers_model_name, tokenizer_name, + batch_size) + self.ce_model_name = ce_model_name + self.candidate_num = candidate_num + self.select_time = select_time + self.ice_template = ICL_PROMPT_TEMPLATES.build(ice_template) + if prompt_template is not None: + self.prompt_template = ICL_PROMPT_TEMPLATES.build(prompt_template) + else: + self.prompt_template = None + self.labels = labels + self.seed = seed + + def topk_search(self): + np.random.seed(self.seed) + res_list = self.forward(self.dataloader) + rtr_idx_list = [[] for _ in range(len(res_list))] + + logger.info('Retrieving data for test set...') + for entry in tqdm.tqdm(res_list, disable=not self.is_main_process): + idx = entry['metadata']['id'] + embed = np.expand_dims(entry['embed'], axis=0) + near_ids = self.index.search( + embed, min(self.candidate_num, + len(self.index_ds)))[1][0].tolist() + candidates = [] + mdl_scores = [] + for j in range(self.select_time): + if j == 0: + rand_idx_list = near_ids[:self.ice_num] + else: + rand_idx_list = np.random.choice(near_ids, + self.ice_num, + replace=False) + rand_idx_list = [int(i) for i in rand_idx_list] + candidates.append(rand_idx_list) + + ice = self.generate_ice(rand_idx_list, + ice_template=self.ice_template) + ice = str(ice) + mask_length = len( + self.tokenizer(ice + self.ice_eos_token, + verbose=False)['input_ids']) + if self.labels is None: + labels = self.get_labels(self.ice_template, + self.prompt_template) + else: + labels = self.labels + prompt_list = [] + for label in labels: + prompt = self.generate_label_prompt( + idx, ice, label, self.ice_template, + self.prompt_template) + prompt = str(prompt) + prompt_list.append(prompt) + loss_list = self.cal_ce(prompt_list, mask_length=mask_length) + probs = np.exp(-np.array(loss_list)) + normalized_probs = probs / probs.sum(0, keepdims=True) + neg_entropy = -entropy(normalized_probs, label_dim=0) + mdl_scores.append(neg_entropy) + + rtr_idx_list[idx] = candidates[mdl_scores.index(max(mdl_scores))] + rtr_idx_list[idx] = [int(i) for i in rtr_idx_list[idx]] + + return rtr_idx_list + + def retrieve(self): + """Retrieve the in-context example index for each test example.""" + return self.topk_search() + + def cal_ce(self, input_texts: List[str], mask_length=None): + if self.metric_model is None: + logger.info( + f'Load model {self.ce_model_name} for calculating MDL...') + self.metric_model = AutoModelForCausalLM.from_pretrained( + self.ce_model_name) + self.metric_model.to(self.device) + inputs = self.tokenizer(input_texts, + padding=True, + return_tensors='pt', + truncation=True) + inputs = {k: v.to(self.device) for k, v in inputs.items()} + outputs = self.metric_model(**inputs) + + shift_logits = outputs.logits[..., :-1, :].contiguous() + shift_labels = inputs['input_ids'][..., 1:].contiguous() + + loss_fct = torch.nn.CrossEntropyLoss( + reduction='none', ignore_index=self.tokenizer.pad_token_id) + shift_logits = shift_logits.view(-1, shift_logits.size(-1)) + loss = loss_fct(shift_logits, + shift_labels.view(-1)).view(shift_labels.size()) + if mask_length is not None: + mask = torch.cat([ + torch.zeros([loss.shape[0], mask_length], dtype=torch.float), + torch.ones([loss.shape[0], loss.shape[-1] - mask_length], + dtype=torch.float) + ], -1) + mask = mask.to(self.device) + loss = torch.mul(mask, loss) + + lens = (inputs['input_ids'] != + self.tokenizer.pad_token_id).sum(-1).cpu().numpy() + if mask_length is not None: + lens -= mask_length + ce_loss = loss.sum(-1).cpu().detach().numpy() / lens + return ce_loss + + +def entropy(probs: np.array, label_dim: int = 0, mask=None): + if mask is None: + return -(probs * np.log(probs)).sum(label_dim) + return -(mask * probs * np.log(probs)).sum(label_dim) diff --git a/opencompass/openicl/icl_retriever/icl_votek_retriever.py b/opencompass/openicl/icl_retriever/icl_votek_retriever.py new file mode 100644 index 00000000..ab81ad0e --- /dev/null +++ b/opencompass/openicl/icl_retriever/icl_votek_retriever.py @@ -0,0 +1,99 @@ +"""Votek Retriever.""" + +import json +import os +import random +from collections import defaultdict +from typing import Optional + +import numpy as np +from sklearn.metrics.pairwise import cosine_similarity + +from opencompass.openicl.icl_retriever.icl_topk_retriever import TopkRetriever + + +class VotekRetriever(TopkRetriever): + """Vote-k In-context Learning Retriever, subclass of `TopkRetriever`. + + **WARNING**: This class has not been tested thoroughly. Please use it with + caution. + """ + + def __init__(self, + dataset, + ice_separator: Optional[str] = '\n', + ice_eos_token: Optional[str] = '\n', + ice_num: Optional[int] = 1, + sentence_transformers_model_name: Optional[ + str] = 'all-mpnet-base-v2', + tokenizer_name: Optional[str] = 'gpt2-xl', + batch_size: Optional[int] = 1, + votek_k: Optional[int] = 3) -> None: + super().__init__(dataset, ice_separator, ice_eos_token, ice_num, + sentence_transformers_model_name, tokenizer_name, + batch_size) + self.votek_k = votek_k + + def votek_select(self, + embeddings=None, + select_num=None, + k=None, + overlap_threshold=None, + vote_file=None): + n = len(embeddings) + if vote_file is not None and os.path.isfile(vote_file): + with open(vote_file) as f: + vote_stat = json.load(f) + else: + vote_stat = defaultdict(list) + + for i in range(n): + cur_emb = embeddings[i].reshape(1, -1) + cur_scores = np.sum(cosine_similarity(embeddings, cur_emb), + axis=1) + sorted_indices = np.argsort(cur_scores).tolist()[-k - 1:-1] + for idx in sorted_indices: + if idx != i: + vote_stat[idx].append(i) + + if vote_file is not None: + with open(vote_file, 'w', encoding='utf-8') as f: + json.dump(vote_stat, f) + votes = sorted(vote_stat.items(), + key=lambda x: len(x[1]), + reverse=True) + j = 0 + selected_indices = [] + while len(selected_indices) < select_num and j < len(votes): + candidate_set = set(votes[j][1]) + flag = True + for pre in range(j): + cur_set = set(votes[pre][1]) + if len(candidate_set.intersection( + cur_set)) >= overlap_threshold * len(candidate_set): + flag = False + break + if not flag: + j += 1 + continue + selected_indices.append(int(votes[j][0])) + j += 1 + if len(selected_indices) < select_num: + unselected_indices = [] + cur_num = len(selected_indices) + for i in range(n): + if i not in selected_indices: + unselected_indices.append(i) + selected_indices += random.sample(unselected_indices, + select_num - cur_num) + return selected_indices + + def vote_k_search(self): + vote_k_idxs = self.votek_select(embeddings=self.embed_list, + select_num=self.ice_num, + k=self.votek_k, + overlap_threshold=1) + return [vote_k_idxs[:] for _ in range(len(self.test_ds))] + + def retrieve(self): + return self.vote_k_search() diff --git a/opencompass/openicl/utils/api_service.py b/opencompass/openicl/utils/api_service.py new file mode 100644 index 00000000..f1ee4aac --- /dev/null +++ b/opencompass/openicl/utils/api_service.py @@ -0,0 +1,104 @@ +import json +import os +import time + +import numpy as np +import openai +import requests + +OPENICL_API_NAME_LIST = ['opt-175b', 'gpt3'] +OPENICL_API_PARAMETER_DICT = { + 'opt-175b': ['URL', 'headers'], + 'gpt3': [ + 'engine', 'temperature', 'max_tokens', 'top_p', 'frequency_penalty', + 'presence_penalty', 'sleep_time' + ] +} +OPENICL_API_REQUEST_CONFIG = { + 'opt-175b': { + 'URL': '', # http://xxx/completions or http://xxx/generate + 'headers': { + 'Content-Type': 'application/json; charset=UTF-8' + } + }, + 'gpt3': { + 'engine': 'text-davinci-003', + 'temperature': 0, + 'max_tokens': 256, + 'top_p': 1.0, + 'frequency_penalty': 0.0, + 'presence_penalty': 0.0, + 'sleep_time': 3 + } +} +PROXIES = {'https': '', 'http': ''} + + +def is_api_available(api_name): + if api_name is None: + return False + return True if api_name in OPENICL_API_NAME_LIST else False + + +def update_openicl_api_request_config(api_name, **kwargs): + if api_name is None or not is_api_available(api_name): + return + + parameter_list = OPENICL_API_PARAMETER_DICT[api_name] + for parameter in parameter_list: + if parameter in kwargs.keys(): + OPENICL_API_REQUEST_CONFIG[api_name][parameter] = kwargs[parameter] + + +def api_get_ppl(api_name, input_texts): + if api_name == 'opt-175b': + pyload = {'prompt': input_texts, 'max_tokens': 0, 'echo': True} + response = json.loads( + requests.post( + OPENICL_API_REQUEST_CONFIG[api_name]['URL'], + data=json.dumps(pyload), + headers=OPENICL_API_REQUEST_CONFIG[api_name]['headers'], + proxies=PROXIES).text) + lens = np.array( + [len(r['logprobs']['tokens']) for r in response['choices']]) + ce_loss = np.array([ + -sum(r['logprobs']['token_logprobs']) for r in response['choices'] + ]) + return ce_loss / lens + + if api_name == 'gpt3': + raise NotImplementedError("GPT-3 API doesn't support PPL calculation") + + +def api_get_tokens(api_name, input_texts): + length_list = [len(text) for text in input_texts] + + if api_name == 'opt-175b': + pyload = {'prompt': input_texts, 'max_tokens': 100, 'echo': True} + response = json.loads( + requests.post( + OPENICL_API_REQUEST_CONFIG[api_name]['URL'], + data=json.dumps(pyload), + headers=OPENICL_API_REQUEST_CONFIG[api_name]['headers'], + proxies=PROXIES).text) + return [r['text'] for r in response['choices']], [ + r['text'][length:] + for r, length in zip(response['choices'], length_list) + ] + + if api_name == 'gpt3': + openai.api_key = os.getenv('OPENAI_API_KEY') + response = openai.Completion.create( + engine=OPENICL_API_REQUEST_CONFIG['gpt3']['engine'], + prompt=input_texts, + temperature=OPENICL_API_REQUEST_CONFIG['gpt3']['temperature'], + max_tokens=OPENICL_API_REQUEST_CONFIG['gpt3']['max_tokens'], + top_p=OPENICL_API_REQUEST_CONFIG['gpt3']['top_p'], + frequency_penalty=OPENICL_API_REQUEST_CONFIG['gpt3'] + ['frequency_penalty'], + presence_penalty=OPENICL_API_REQUEST_CONFIG['gpt3'] + ['presence_penalty']) + time.sleep(OPENICL_API_REQUEST_CONFIG['gpt3']['sleep_time']) + return [(input + r['text']) + for r, input in zip(response['choices'], input_texts) + ], [r['text'] for r in response['choices']] diff --git a/opencompass/partitioners/naive.py b/opencompass/partitioners/naive.py new file mode 100644 index 00000000..92447d76 --- /dev/null +++ b/opencompass/partitioners/naive.py @@ -0,0 +1,60 @@ +import os.path as osp +from typing import Dict, List + +from mmengine.config import Config, ConfigDict + +from opencompass.registry import PARTITIONERS +from opencompass.utils import get_infer_output_path + +from .base import BasePartitioner + + +@PARTITIONERS.register_module() +class NaivePartitioner(BasePartitioner): + """Naive task partitioner. This partitioner will generate a task for each + model-dataset pair. + + Args: + config (ConfigDict): The full config dict. + """ + + def partition(self, models: List[ConfigDict], datasets: List[ConfigDict], + work_dir: str, out_dir: str) -> List[Dict]: + """Partition model-dataset pairs into tasks. Each task is defined as a + dict and will run independently as a unit. Its structure is as + follows: + + .. code-block:: python + + { + 'models': [], # a list of model configs + 'datasets': [[]], # a nested list of dataset configs, each + list corresponds to a model + 'work_dir': '', # the work dir + } + + Args: + models (List[ConfigDict]): A list of model configs. + datasets (List[ConfigDict]): A list of dataset configs. + work_dir (str): The work dir for the task. + out_dir (str): The full output path for the task, intended for + Partitioners to check whether the task is finished via the + existency of result file in this directory. + + Returns: + List[Dict]: A list of tasks. + """ + + tasks = [] + for model in models: + for dataset in datasets: + filename = get_infer_output_path(model, dataset, out_dir) + if osp.exists(filename): + continue + task = Config({ + 'models': [model], + 'datasets': [[dataset]], + 'work_dir': work_dir + }) + tasks.append(task) + return tasks diff --git a/opencompass/partitioners/size.py b/opencompass/partitioners/size.py new file mode 100644 index 00000000..b1eb4eda --- /dev/null +++ b/opencompass/partitioners/size.py @@ -0,0 +1,187 @@ +import copy +import math +import os.path as osp +from typing import List, Tuple, Union + +import mmengine +from mmengine.config import Config, ConfigDict + +from opencompass.registry import PARTITIONERS +from opencompass.utils import (build_dataset_from_cfg, dataset_abbr_from_cfg, + get_infer_output_path) + +from .base import BasePartitioner + + +@PARTITIONERS.register_module() +class SizePartitioner(BasePartitioner): + """Task partitioner based on the size of the dataset (with some rough + expansion as an estimation of computational cost). + + Args: + out_dir (str): The output directory of tasks. + max_task_size (int): The maximum size of a task. + gen_task_coef (int): The dataset cost measurement coefficient for + generation tasks. + dataset_size_path (str): The path to the dataset size cache file. + """ + + def __init__(self, + out_dir: str, + max_task_size: int = 2000, + gen_task_coef: int = 20, + dataset_size_path: str = '.cache/dataset_size.json'): + super().__init__(out_dir) + self.max_task_size = max_task_size + self.gen_task_coef = gen_task_coef + self.dataset_size_path = dataset_size_path + + def partition(self, models: List[ConfigDict], datasets: List[ConfigDict], + work_dir: str, out_dir: str) -> List[ConfigDict]: + """Partition model-dataset pairs into tasks. Each task is defined as a + dict and will run independently as a unit. Its structure is as + follows: + + .. code-block:: python + + { + 'models': [], # a list of model configs + 'datasets': [[]], # a nested list of dataset configs, each + list corresponds to a model + 'work_dir': '', # the work dir + } + + Args: + models (List[ConfigDict]): A list of model configs. + datasets (List[ConfigDict]): A list of dataset configs. + work_dir (str): The work dir for the task. + out_dir (str): The full output path for the task, intended for + Partitioners to check whether the task is finished via the + existency of result file in this directory. + + Returns: + List[ConfigDict]: A list of tasks. + """ + + datasets = sorted(datasets, + key=lambda x: self.get_cost(x), + reverse=True) + tasks = [] + for model in models: + task = Config({ + 'models': [model], + 'datasets': [[]], + 'work_dir': work_dir + }) + num_data = 0 + for dataset in datasets: + filename = get_infer_output_path(model, dataset, out_dir) + root, ext = osp.splitext(filename) + # skip the task if the task output exists + if osp.exists(filename): + continue + dataset_size = self.get_cost(dataset) + if dataset_size > self.max_task_size: + dataset_splits = self.split_dataset(dataset) + for i, dataset_split in enumerate(dataset_splits): + # skip the task it the task output exists + if not osp.exists(f'{root}_{i}{ext}'): + tasks.append( + Config({ + 'models': [model], + 'datasets': [[dataset_split]], + 'work_dir': work_dir + })) + else: + if num_data + dataset_size > self.max_task_size: + tasks.append(task) + task = Config({ + 'models': [model], + 'datasets': [[]], + 'work_dir': work_dir + }) + num_data = 0 + task['datasets'][0].append(dataset) + num_data = num_data + dataset_size + if task['datasets'][0]: + tasks.append(task) + + return tasks + + @property + def dataset_size(self): + if not hasattr(self, '_dataset_size'): + if osp.exists(self.dataset_size_path): + self._dataset_size = mmengine.load(self.dataset_size_path) + else: + self._dataset_size = {} + return self._dataset_size + + def split_dataset(self, dataset_cfg: ConfigDict) -> List[ConfigDict]: + """Split dataset into several parts.""" + dataset_size, num_repeats = self.get_cost(dataset_cfg, + get_raw_factors=True) + split_configs = [] + abbr = dataset_abbr_from_cfg(dataset_cfg) + step = self.max_task_size // num_repeats + # evenly distribute the task + step = math.ceil(dataset_size / math.ceil(dataset_size / step)) + for part, i in enumerate(range(0, dataset_size, step)): + cfg = copy.deepcopy(dataset_cfg) + cfg['abbr'] = abbr + f'_{part}' + test_range = cfg['reader_cfg'].get('test_range', '') + cfg['reader_cfg']['test_range'] = f'{test_range}[{i}:{i+step}]' + split_configs.append(cfg) + return split_configs + + def get_cost(self, + dataset: ConfigDict, + get_raw_factors: bool = False) -> Union[int, Tuple[int, int]]: + """Get the computational cost of inferring on the dataset. + + Args: + dataset (ConfigDict): The dataset config. + get_raw_factors (bool): If True, the raw factors of computational + cost will be returned. + + Returns: + int or Tuple[int, int]: The size of the dataset. If get_raw_factors + is True, the number of repeats will also be returned. + """ + dataset_abbr = dataset_abbr_from_cfg(dataset) + + # If it's the PPL template, the dataset size will be multiplied by the + # number of labels + infer_cfg = dataset.infer_cfg + test_range = dataset.reader_cfg.get('test_range', '') + template = (infer_cfg.prompt_template.template if 'prompt_template' + in infer_cfg else infer_cfg.ice_template.template) + # If it's the Gen template, the dataset size will be multiplied by the + # self.gen_task_coef + factor = self.gen_task_coef + if isinstance(template, dict): + ctr = sum(key in template for key in ('begin', 'round', 'end')) + if ctr != len(template.keys()): + factor = len(template.keys()) + + if dataset_abbr in self.dataset_size: + actual_size = eval('len(range(self.dataset_size[dataset_abbr])' + f'{test_range})') + if get_raw_factors: + return actual_size, factor + return factor * actual_size + + dataset = build_dataset_from_cfg(dataset) + self.dataset_size[dataset_abbr] = len(dataset.test) + + mmengine.mkdir_or_exist('.cache/') + mmengine.dump(self.dataset_size, + self.dataset_size_path, + indent=4, + ensure_ascii=False) + + actual_size = eval('len(range(self.dataset_size[dataset_abbr])' + f'{test_range})') + if get_raw_factors: + return actual_size, factor + return factor * actual_size diff --git a/opencompass/registry.py b/opencompass/registry.py new file mode 100644 index 00000000..8f26e607 --- /dev/null +++ b/opencompass/registry.py @@ -0,0 +1,24 @@ +from mmengine.registry import Registry + +PARTITIONERS = Registry('partitioner', locations=['opencompass.partitioners']) +RUNNERS = Registry('runner', locations=['opencompass.runners']) +TASKS = Registry('task', locations=['opencompass.tasks']) +MODELS = Registry('model', locations=['opencompass.models']) +# TODO: LOAD_DATASET -> DATASETS +LOAD_DATASET = Registry('load_dataset', locations=['opencompass.datasets']) +TEXT_POSTPROCESSORS = Registry( + 'text_postprocessors', locations=['opencompass.utils.text_postprocessors']) +EVALUATORS = Registry('evaluators', locations=['opencompass.evaluators']) + +ICL_INFERENCERS = Registry('icl_inferencers', + locations=['opencompass.openicl.icl_inferencer']) +ICL_RETRIEVERS = Registry('icl_retrievers', + locations=['opencompass.openicl.icl_retriever']) +ICL_DATASET_READERS = Registry( + 'icl_dataset_readers', + locations=['opencompass.openicl.icl_dataset_reader']) +ICL_PROMPT_TEMPLATES = Registry( + 'icl_prompt_templates', + locations=['opencompass.openicl.icl_prompt_template']) +ICL_EVALUATORS = Registry('icl_evaluators', + locations=['opencompass.openicl.icl_evaluator']) diff --git a/opencompass/runners/base.py b/opencompass/runners/base.py new file mode 100644 index 00000000..54ad7b2e --- /dev/null +++ b/opencompass/runners/base.py @@ -0,0 +1,80 @@ +import getpass +from abc import abstractmethod +from typing import Any, Dict, List, Tuple + +from mmengine.config import ConfigDict, Config + +from opencompass.utils import LarkReporter, get_logger + + +class BaseRunner: + """Base class for all runners. A runner is responsible for launching + multiple tasks. + + Args: + task (ConfigDict): Task type config. + debug (bool): Whether to run in debug mode. + lark_bot_url (str): Lark bot url. + """ + + def __init__(self, + task: ConfigDict, + debug: bool = False, + lark_bot_url: str = None): + self.task_cfg = Config(task) + self.debug = debug + if lark_bot_url: + self.lark_reporter = LarkReporter(lark_bot_url) + else: + self.lark_reporter = None + + def __call__(self, tasks: List[Dict[str, Any]]): + """Launch multiple tasks and summarize the results. + + Args: + tasks (list[dict]): A list of task configs, usually generated by + Partitioner. + """ + status = self.launch(tasks) + self.summarize(status) + + @abstractmethod + def launch(self, tasks: List[Dict[str, Any]]) -> List[Tuple[str, int]]: + """Launch multiple tasks. + + Args: + tasks (list[dict]): A list of task configs, usually generated by + Partitioner. + + Returns: + list[tuple[str, int]]: A list of (task name, exit code). + """ + + def summarize(self, status: List[Tuple[str, int]]) -> None: + """Summarize the results of the tasks. + + Args: + status (list[tuple[str, int]]): A list of (task name, exit code). + """ + + failed_logs = [] + for _task, code in status: + if code != 0: + get_logger().error(f'{_task} failed with code {code}') + failed_logs.append(_task) + if self.lark_reporter: + num_succeeded = len(status) - len(failed_logs) + if len(failed_logs) > 0: + content = f'{getpass.getuser()} 的 ' + content += f'{self.task_cfg.type} 任务已完成，' + content += f'成功任务 {num_succeeded} 个，' + content += f'失败 {len(failed_logs)} 个。以下为失败的任务列表：' + content += '\n' + '\n'.join(failed_logs) + self.lark_reporter.post(title=f'悲报：您有{len(failed_logs)}个' + '任务炸了', + content=content) + else: + content = f'{getpass.getuser()} 的 ' + content += f'{self.task_cfg.type} 任务已完成，' + content += f'成功任务 {num_succeeded} 个。' + self.lark_reporter.post(title='喜报：全部任务完成', content=content) diff --git a/opencompass/runners/slurm.py b/opencompass/runners/slurm.py new file mode 100644 index 00000000..448f5b2c --- /dev/null +++ b/opencompass/runners/slurm.py @@ -0,0 +1,148 @@ +import inspect +import os +import os.path as osp +import random +import subprocess +import time +from typing import Any, Dict, List, Tuple + +import mmengine +from mmengine.config import ConfigDict +from mmengine.utils import track_parallel_progress + +from opencompass.registry import RUNNERS, TASKS +from opencompass.utils import get_logger + +from .base import BaseRunner + + +@RUNNERS.register_module() +class SlurmRunner(BaseRunner): + """Distributed runner based on Slurm. It will launch tasks in parallel + using `srun` command. + + Args: + task (ConfigDict): Task type config. + max_num_workers (int): Max number of workers to run in parallel. + Defaults to 32. + retry (int): Number of retries if the job failed. Defaults to 2. + partition (str): Slurm partition name. Defaults to None. + quotatype (str): Slurm quota type. Defaults to None. + debug (bool): Whether to run in debug mode. Defaults to False. + lark_bot_url (str): Lark bot url. Defaults to None. + """ + + def __init__(self, + task: ConfigDict, + max_num_workers: int = 32, + retry: int = 2, + partition: str = None, + quotatype: str = None, + debug: bool = False, + lark_bot_url: str = None): + super().__init__(task=task, debug=debug, lark_bot_url=lark_bot_url) + self.max_num_workers = max_num_workers + self.retry = retry + self.partition = partition + self.quotatype = quotatype + + def launch(self, tasks: List[Dict[str, Any]]) -> List[Tuple[str, int]]: + """Launch multiple tasks. + + Args: + tasks (list[dict]): A list of task configs, usually generated by + Partitioner. + + Returns: + list[tuple[str, int]]: A list of (task name, exit code). + """ + + if not self.debug: + status = track_parallel_progress(self._launch, + tasks, + nproc=self.max_num_workers, + keep_order=False) + else: + status = [self._launch(task, random_sleep=False) for task in tasks] + return status + + def _launch(self, task_cfg: ConfigDict, random_sleep: bool = True): + """Launch a single task. + + Args: + task_cfg (ConfigDict): Task config. + random_sleep (bool): Whether to sleep for a random time before + running the command. This avoids cluster error when launching + multiple tasks at the same time. Default: True. + + Returns: + tuple[str, int]: Task name and exit code. + """ + + task_type = self.task_cfg.type + if isinstance(self.task_cfg.type, str): + task_type = TASKS.get(task_type) + task = task_type(task_cfg) + num_gpus = task.num_gpus + task_name = task.name + script_path = inspect.getsourcefile(task_type) + + # Dump task config to file + mmengine.mkdir_or_exist('tmp/') + param_file = f'tmp/{os.getpid()}_params.py' + task_cfg.dump(param_file) + + # Build up slurm command + task_cmd_template = task.get_command_template() + task_cmd = task_cmd_template.replace('{SCRIPT_PATH}', + script_path).replace( + '{CFG_PATH}', param_file) + cmd = 'srun' + if self.partition: + cmd += f' -p {self.partition}' + if self.quotatype: + cmd += f' --quotatype={self.quotatype}' + if num_gpus > 0: + cmd += f' --gres=gpu:{num_gpus}' + cmd += f" -N1 -J '{task_name[:512]}' {task_cmd}" + logger = get_logger() + logger.debug(f'Running command: {cmd}') + + # Run command with retry + if self.debug: + stdout = None + else: + out_path = task.get_log_path(file_extension='out') + mmengine.mkdir_or_exist(osp.split(out_path)[0]) + stdout = open(out_path, 'w', encoding='utf-8') + + if random_sleep: + time.sleep(random.randint(0, 10)) + result = subprocess.run(cmd, + shell=True, + text=True, + stdout=stdout, + stderr=stdout) + + retry = self.retry + output_paths = task.get_output_paths() + while self._job_failed(result.returncode, output_paths) and retry > 0: + retry -= 1 + if random_sleep: + time.sleep(random.randint(0, 10)) + result = subprocess.run(cmd, + shell=True, + text=True, + stdout=stdout, + stderr=stdout) + + if result.returncode != 0 and not self.debug: + logger.warning(f'task {task_name} fail, see\n{out_path}') + + # Clean up + os.remove(param_file) + return task_name, result.returncode + + def _job_failed(self, return_code: int, output_paths: List[str]) -> bool: + return return_code != 0 or not all( + osp.exists(output_path) for output_path in output_paths) diff --git a/opencompass/tasks/__init__.py b/opencompass/tasks/__init__.py new file mode 100644 index 00000000..308ea5d6 --- /dev/null +++ b/opencompass/tasks/__init__.py @@ -0,0 +1,2 @@ +from .openicl_eval import * # noqa: F401, F403 +from .openicl_infer import * # noqa: F401, F403 diff --git a/opencompass/tasks/base.py b/opencompass/tasks/base.py new file mode 100644 index 00000000..214c93b1 --- /dev/null +++ b/opencompass/tasks/base.py @@ -0,0 +1,87 @@ +import os +from abc import abstractmethod +from typing import List + +from mmengine.config import ConfigDict + +from opencompass.utils import get_infer_output_path, task_abbr_from_cfg + + +class BaseTask: + """Base class for all tasks. There are two ways to run the task: + 1. Directly by calling the `run` method. + 2. Calling the `get_command_template` method to get the command template, + and then run the command in the shell. + + Args: + cfg (ConfigDict): Config dict. + """ + + # The prefix of the task name. + name_prefix: str = None + # The subdirectory of the work directory to store the log files. + log_subdir: str = None + # The subdirectory of the work directory to store the output files. + output_subdir: str = None + + def __init__(self, cfg: ConfigDict): + self.cfg = cfg + self.model_cfgs = cfg['models'] + self.dataset_cfgs = cfg['datasets'] + self.work_dir = cfg['work_dir'] + + @abstractmethod + def run(self): + """Run the task.""" + + @abstractmethod + def get_command_template(self) -> str: + """Get the command template for the task. + + The command template should + contain the following placeholders: + 1. ``{SCRIPT_PATH}``: This placeholder will be replaced by the path to + the script file of the task. + 2. ``{CFG_PATH}`` This placeholder will be replaced by the + path to the config file of the task. + """ + + @property + def name(self) -> str: + return self.name_prefix + task_abbr_from_cfg( + { + 'models': self.model_cfgs, + 'datasets': self.dataset_cfgs + }) + + def __repr__(self) -> str: + return f'{self.__class__.__name__}({self.cfg})' + + def get_log_path(self, file_extension: str = 'json') -> str: + """Get the path to the log file. + + Args: + file_extension (str): The file extension of the log file. + Default: 'json'. + """ + return get_infer_output_path( + self.model_cfgs[0], self.dataset_cfgs[0][0], + os.path.join(self.work_dir, self.log_subdir), file_extension) + + def get_output_paths(self, file_extension: str = 'json') -> List[str]: + """Get the paths to the output files. Every file should exist if the + task succeeds. + + Args: + file_extension (str): The file extension of the output files. + Default: 'json'. + """ + output_paths = [] + for model, datasets in zip(self.model_cfgs, self.dataset_cfgs): + for dataset in datasets: + output_paths.append( + get_infer_output_path( + model, dataset, + os.path.join(self.work_dir, self.output_subdir), + file_extension)) + return output_paths diff --git a/opencompass/tasks/llm_eval.py b/opencompass/tasks/llm_eval.py new file mode 100644 index 00000000..2e1b5c85 --- /dev/null +++ b/opencompass/tasks/llm_eval.py @@ -0,0 +1,91 @@ +from collections import defaultdict +from typing import Dict, List + +import mmengine +from mmengine import ConfigDict, track_parallel_progress + +from opencompass.registry import EVALUATORS, MODELS +from opencompass.utils import dataset_abbr_from_cfg, model_abbr_from_cfg + + +@EVALUATORS.register_module() +class ModelEvaluator: + """TODO: Finish the implementation""" + + def __init__( + self, + config: ConfigDict, + ) -> None: + self.tasks = [] + self.cfg = config + self.parse_cfg(self.cfg.pop('evaluator', ConfigDict({}))) + self.dataset_abbrs = [ + dataset_abbr_from_cfg(d) for d in self.cfg['datasets'] + ] + self.model_abbrs = [model_abbr_from_cfg(m) for m in self.cfg['models']] + assert len(self.model_abbrs) > 1 + + def parse_cfg(self, cfg: ConfigDict): + # The judger + self.judger = MODELS.build(cfg['judger']) + # Maximum number of workers + self.max_num_workers = cfg.get('max_num_workers', 4) + + def evaluate(self): + model_scores = defaultdict(int) + all_partial_scores = track_parallel_progress( + self._evaluate_dataset, + self.dataset_abbrs, + nproc=self.max_num_workers, + keep_order=True) + for partial_scores in all_partial_scores: + for model_idx, score in partial_scores.items(): + model_scores[self.model_abbrs[model_idx]] += score + print(model_scores) + + def _load_dataset(self, dataset_abbr: str): + # for self. + original_datasets = [] + self.responses: List[List[str]] = [] + self.questions: List[str] = [] + for model_abbr in self.model_abbrs: + filename = f'output_model/{model_abbr}/{dataset_abbr}.json' + original_datasets.append(mmengine.load(filename)) + for key in original_datasets[-1].keys(): + self.questions.append(original_datasets[-1][key]['origin_prompt']) + responses = [] + for i in range(len(self.model_abbrs)): + responses.append(original_datasets[i][key]['prediction']) + self.responses.append(responses) + + def _evaluate_dataset(self, dataset_abbr: str): + self._load_dataset(dataset_abbr=dataset_abbr) + model_scores = defaultdict(int) + for question, responses in zip(self.questions, self.responses): + prompt = self._make_prompt(question, responses) + print(prompt) + output = self.judger.generate(prompt, + max_out_len=2 * + len(self.model_abbrs)) + model_scores = self._rank_models(output, model_scores) + return model_scores + + def _make_prompt(self, question: str, responses: List[str]) -> str: + prompt = ('Below are a question and a set of answers, each numbered by' + ' a digit. Please sort the answers from least to most ' + 'appropriate to the question. Only return the digit ' + 'seperated by a blank space. For example, when there are ' + 'three answers presented, you should say "1 0 2" when the ' + 'second answer is the best and the third is the worst.\n' + f'Q: {question}\n') + for i, response in enumerate(responses): + prompt += f'A{i + 1}: {response}\n' + return prompt + + def _rank_models(self, output: str, + model_scores: defaultdict) -> Dict[str, int]: + """Returns model ranking.""" + output = output.strip().split(' ') + for score, model_idx in enumerate(output): + model_scores[model_idx] += int(score) + return model_scores diff --git a/opencompass/tasks/openicl_eval.py b/opencompass/tasks/openicl_eval.py new file mode 100644 index 00000000..71e3fc06 --- /dev/null +++ b/opencompass/tasks/openicl_eval.py @@ -0,0 +1,178 @@ +import argparse +import os.path as osp +import time +from typing import Optional + +import mmengine +from mmengine.config import Config, ConfigDict +from mmengine.utils import mkdir_or_exist + +from opencompass.registry import (ICL_EVALUATORS, MODELS, TASKS, + TEXT_POSTPROCESSORS) +from opencompass.tasks.base import BaseTask +from opencompass.utils import (build_dataset_from_cfg, get_infer_output_path, + get_logger, task_abbr_from_cfg) + + +@TASKS.register_module(force=(__name__ == '__main__')) # A hack for script run +class OpenICLEvalTask(BaseTask): + """OpenICL Evaluation Task. + + This task is used to evaluate the metric between predictions and + references. + """ + + name_prefix = 'OpenICLEval' + log_subdir = 'logs/eval' + output_subdir = 'results' + + def __init__(self, cfg: ConfigDict): + super().__init__(cfg) + self.num_gpus = 0 + self.logger = get_logger() + + def get_command_template(self): + return 'python3 {SCRIPT_PATH} {CFG_PATH}' + + def run(self): + for model_cfg, dataset_cfgs in zip(self.model_cfgs, self.dataset_cfgs): + for dataset_cfg in dataset_cfgs: + self.model_cfg = model_cfg + self.dataset_cfg = dataset_cfg + + # Load Dataset + self.eval_cfg = self.dataset_cfg.get('eval_cfg') + self.output_column = dataset_cfg['reader_cfg']['output_column'] + + out_path = get_infer_output_path( + self.model_cfg, self.dataset_cfg, + osp.join(self.work_dir, 'results')) + if osp.exists(out_path): + continue + self._score() + + def _score(self): + test_set = build_dataset_from_cfg(self.dataset_cfg).test + # Postprocess dataset if necessary + if 'dataset_postprocessor' in self.eval_cfg: + TEXT_POSTPROCESSORS.get( + self.eval_cfg['dataset_postprocessor']['type']) + + def postprocess(sample): + s = sample[self.output_column] + proc = TEXT_POSTPROCESSORS.get( + self.eval_cfg['dataset_postprocessor']['type']) + sample[self.output_column] = proc(s) + return sample + + test_set = test_set.map(postprocess) + + # Load predictions + filename = get_infer_output_path( + self.model_cfg, self.dataset_cfg, + osp.join(self.work_dir, 'predictions')) + # in case the prediction is partial + root, ext = osp.splitext(filename) + partial_filename = root + '_0' + ext + + if not osp.exists(osp.realpath(filename)) and not osp.exists( + osp.realpath(partial_filename)): + result = {'error': 'No predictions found.'} + else: + if osp.exists(osp.realpath(filename)): + preds = mmengine.load(filename) + pred_strs = [ + preds[str(i)]['prediction'] for i in range(len(preds)) + ] + else: + filename = partial_filename + pred_strs = [] + i = 1 + while osp.exists(osp.realpath(filename)): + preds = mmengine.load(filename) + filename = root + f'_{i}' + ext + i += 1 + pred_strs += [ + preds[str(i)]['prediction'] for i in range(len(preds)) + ] + + if ('pred_role' in self.eval_cfg + and 'meta_template' in self.model_cfg + and not MODELS.get(self.model_cfg['type']).is_api): + # Create a prompt template for role config parsing + from opencompass.models.base import LMTemplateParser + parser = LMTemplateParser(self.model_cfg['meta_template']) + role = parser.roles[self.eval_cfg['pred_role']] + pred_strs = [ + self._extract_role_pred(pred, role.get('begin', None), + role.get('end', None)) + for pred in pred_strs + ] + + # Postprocess predictions if necessary + if 'pred_postprocessor' in self.eval_cfg: + proc = TEXT_POSTPROCESSORS.get( + self.eval_cfg['pred_postprocessor']['type']) + pred_strs = [proc(s) for s in pred_strs] + + icl_evaluator = ICL_EVALUATORS.build(self.eval_cfg['evaluator']) + result = icl_evaluator.score( + predictions=pred_strs, references=test_set[self.output_column]) + + if 'error' in result: + self.logger.error( + f'Task {task_abbr_from_cfg(self.cfg)}: {result["error"]}') + return + + # Save result + out_path = get_infer_output_path(self.model_cfg, self.dataset_cfg, + osp.join(self.work_dir, 'results')) + mkdir_or_exist(osp.split(out_path)[0]) + mmengine.dump(result, out_path) + + def _extract_role_pred(self, s: str, begin_str: Optional[str], + end_str: Optional[str]) -> str: + """Extract the role prediction from the full prediction string. The + role prediction may be the substring between the begin and end string. + + Args: + s (str): Full prediction string. + begin_str (str): The beginning string of the role + end_str (str): The ending string of the role. + + Returns: + str: The extracted role prediction. + """ + start = 0 + end = len(s) + + if begin_str: + begin_idx = s.find(begin_str) + if begin_idx != -1: + start = begin_idx + len(begin_str) + + if end_str: + # TODO: Support calling tokenizer for the accurate eos token + # and avoid such hardcode + end_idx = s.find(end_str[:1], start) + if end_idx != -1: + end = end_idx + + return s[start:end] + + +def parse_args(): + parser = argparse.ArgumentParser(description='Score Calculator') + parser.add_argument('config', help='Config file path') + args = parser.parse_args() + return args + + +if __name__ == '__main__': + args = parse_args() + cfg = Config.fromfile(args.config) + start_time = time.time() + inferencer = OpenICLEvalTask(cfg) + inferencer.run() + end_time = time.time() + get_logger().info(f'time elapsed: {end_time - start_time:.2f}s') diff --git a/opencompass/utils/abbr.py b/opencompass/utils/abbr.py new file mode 100644 index 00000000..52616c5b --- /dev/null +++ b/opencompass/utils/abbr.py @@ -0,0 +1,46 @@ +import os.path as osp +from typing import Dict + +from mmengine.config import ConfigDict + + +def model_abbr_from_cfg(cfg: ConfigDict) -> str: + """Generate model abbreviation from the model's confg.""" + if 'abbr' in cfg: + return cfg['abbr'] + model_abbr = cfg['type'] + '_' + '_'.join( + osp.realpath(cfg['path']).split('/')[-2:]) + model_abbr = model_abbr.replace('/', '_') + return model_abbr + + +def dataset_abbr_from_cfg(cfg: ConfigDict) -> str: + """Returns dataset abbreviation from the dataset's confg.""" + if 'abbr' in cfg: + return cfg['abbr'] + dataset_abbr = cfg['path'] + if 'name' in cfg: + dataset_abbr += '_' + cfg['name'] + dataset_abbr = dataset_abbr.replace('/', '_') + return dataset_abbr + + +def task_abbr_from_cfg(task: Dict) -> str: + """Returns task abbreviation from the task's confg.""" + return '[' + ','.join([ + f'{model_abbr_from_cfg(model)}/' + f'{dataset_abbr_from_cfg(dataset)}' + for i, model in enumerate(task['models']) + for dataset in task['datasets'][i] + ]) + ']' + + +def get_infer_output_path(model_cfg: ConfigDict, + dataset_cfg: ConfigDict, + root_path: str = None, + file_extension: str = 'json') -> str: + # TODO: Rename this func + assert root_path is not None, 'default root_path is not allowed any more' + model_abbr = model_abbr_from_cfg(model_cfg) + dataset_abbr = dataset_abbr_from_cfg(dataset_cfg) + return osp.join(root_path, model_abbr, f'{dataset_abbr}.{file_extension}') diff --git a/opencompass/utils/git.py b/opencompass/utils/git.py new file mode 100644 index 00000000..7b382c1b --- /dev/null +++ b/opencompass/utils/git.py @@ -0,0 +1,13 @@ +import subprocess + + +def get_git_root() -> str: + cmd = ['git', 'rev-parse', '--show-toplevel'] + result = subprocess.run(cmd, stdout=subprocess.PIPE, check=True) + return result.stdout.decode('utf-8').strip() + + +def get_latest_commit(branch: str) -> str: + cmd = ['git', 'rev-parse', branch] + result = subprocess.run(cmd, stdout=subprocess.PIPE, check=True) + return result.stdout.decode('utf-8').strip() diff --git a/requirements.txt b/requirements.txt new file mode 100644 index 00000000..77b80bbe --- /dev/null +++ b/requirements.txt @@ -0,0 +1,26 @@ +accelerate>=0.19.0 +datasets>=2.12.0 +evaluate>=0.3.0 +faiss_gpu==1.7.2 +nltk==3.8 +numpy==1.23.4 +openai==0.27.1 +rank_bm25==0.2.2 +requests==2.28.1 +scikit_learn==1.2.1 +sentence_transformers==2.2.2 +torch>=1.13.1 +tqdm==4.64.1 +transformers>=4.29.1 +openai +mmengine +jieba +pandas<2.0.0 +cpm_kernels +tokenizers>=0.13.3 +tabulate +fairscale +colossalai +tabulate +boto3 +tiktoken \ No newline at end of file diff --git a/setup.py b/setup.py new file mode 100644 index 00000000..9c44bfb9 --- /dev/null +++ b/setup.py @@ -0,0 +1,45 @@ +from setuptools import find_packages, setup +from setuptools.command.install import install + + +class DownloadNLTK(install): + + def run(self): + self.do_egg_install() + import nltk + nltk.download('punkt') + + +with open('README.md') as f: + readme = f.read() + + +def do_setup(): + setup( + name='opencompass', + version='0.5.0', + description='A comprehensive toolkit for large model evaluation', + # url="", + # author="", + long_description=readme, + long_description_content_type='text/markdown', + cmdclass={'download_nltk': DownloadNLTK}, + setup_requires=['nltk==3.8'], + python_requires='>=3.8.0', + packages=find_packages(exclude=[ + 'test*', + 'paper_test*', + ]), + keywords=['AI', 'NLP', 'in-context learning'], + classifiers=[ + 'Programming Language :: Python :: 3.8', + 'Programming Language :: Python :: 3.9', + 'Programming Language :: Python :: 3.10', + 'Intended Audience :: Developers', + 'Intended Audience :: Education', + 'Intended Audience :: Science/Research', + ]) + + +if __name__ == '__main__': + do_setup() diff --git a/tests/prompt/test_prompt_list.py b/tests/prompt/test_prompt_list.py new file mode 100644 index 00000000..83988581 --- /dev/null +++ b/tests/prompt/test_prompt_list.py @@ -0,0 +1,76 @@ +import unittest + +from opencompass.utils.prompt import PromptList + + +class TestPromptList(unittest.TestCase): + + def test_initialization(self): + pl = PromptList() + self.assertEqual(pl, []) + + pl = PromptList(['test', '123']) + self.assertEqual(pl, ['test', '123']) + + def test_format(self): + pl = PromptList(['hi {a}{b}', {'prompt': 'hey {a}!'}, '123']) + new_pl = pl.format(a=1, b=2, c=3) + self.assertEqual(new_pl, ['hi 12', {'prompt': 'hey 1!'}, '123']) + + new_pl = pl.format(b=2) + self.assertEqual(new_pl, ['hi {a}2', {'prompt': 'hey {a}!'}, '123']) + + new_pl = pl.format(d=1) + self.assertEqual(new_pl, ['hi {a}{b}', {'prompt': 'hey {a}!'}, '123']) + + def test_replace(self): + pl = PromptList(['hello world', {'prompt': 'hello world'}, '123']) + new_pl = pl.replace('world', 'there') + self.assertEqual(new_pl, + ['hello there', { + 'prompt': 'hello there' + }, '123']) + + new_pl = pl.replace('123', PromptList(['p', {'role': 'BOT'}])) + self.assertEqual( + new_pl, + ['hello world', { + 'prompt': 'hello world' + }, 'p', { + 'role': 'BOT' + }]) + + new_pl = pl.replace('2', PromptList(['p', {'role': 'BOT'}])) + self.assertEqual(new_pl, [ + 'hello world', { + 'prompt': 'hello world' + }, '1', 'p', { + 'role': 'BOT' + }, '3' + ]) + + with self.assertRaises(TypeError): + new_pl = pl.replace('world', PromptList(['new', 'world'])) + + def test_add(self): + pl = PromptList(['hello']) + new_pl = pl + ' world' + self.assertEqual(new_pl, ['hello', ' world']) + + pl2 = PromptList([' world']) + new_pl = pl + pl2 + self.assertEqual(new_pl, ['hello', ' world']) + + new_pl = 'hi, ' + pl + self.assertEqual(new_pl, ['hi, ', 'hello']) + + pl += '!' + self.assertEqual(pl, ['hello', '!']) + + def test_str(self): + pl = PromptList(['hello', ' world', {'prompt': '!'}]) + self.assertEqual(str(pl), 'hello world!') + + with self.assertRaises(TypeError): + pl = PromptList(['hello', ' world', 123]) + str(pl)