[Feat] Update math/agent (#716)

* minor add

* minor add

* minor fix

configs/datasets/gsm8k/gsm8k_agent_gen_be1606.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import AgentInferencer
+from opencompass.datasets import (
+    GSM8KDataset,
+    gsm8k_postprocess,
+    gsm8k_dataset_postprocess,
+    Gsm8kAgentEvaluator,
+)
+
+gsm8k_reader_cfg = dict(input_columns=["question"], output_column="answer")
+
+gsm8k_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                # # ################################### NEW SHOT ###################################
+                dict(role='HUMAN', prompt='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?'),
+                dict(role='BOT', prompt='Tool:PythonInterpreter\nTool Input:def solution():\n    mark_pointers_2 = 25 * 2\n    mark_pointers_3 = 8 * 3\n    mark_free_throws = 10 * 1\n    mark_points_scored = mark_pointers_2 + mark_pointers_3 + mark_free_throws\n    opponents_pointers_2 = mark_pointers_2 * 2\n    opponents_pointers_3 = mark_pointers_3 / 2\n    opponents_free_throws = mark_free_throws / 2\n    opponents_points_scored = opponents_pointers_2 + opponents_pointers_3 + opponents_free_throws\n    total_points_scored = mark_points_scored + opponents_points_scored\n    result = total_points_scored\n    return result'),
+                dict(role='SYSTEM', prompt='Response:201'),
+                dict(role='BOT', prompt='Thought: According to the response, I got the answer\nFinalAnswer: 201'),
+
+                dict(role='HUMAN', prompt='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?'),
+                dict(role='BOT', prompt='Tool:PythonInterpreter\nTool Input:def solution():\n    marbles = 60\n    num_increased_marbles = marbles * 2 / 5\n    num_total_marbles = marbles + num_increased_marbles\n    frisbees = marbles / 2\n    num_increased_frisbees = frisbees * 2 / 5\n    num_total_frisbees = frisbees + num_increased_frisbees\n    deck_cards = frisbees - 20\n    num_increased_deck_cards = deck_cards * 2 / 5\n    num_total_deck_cards = deck_cards + num_increased_deck_cards\n    num_total = num_total_marbles + num_total_frisbees + num_total_deck_cards\n    result = num_total\n    return result'),
+                dict(role='SYSTEM', prompt='Response:140'),
+                dict(role='BOT', prompt='Thought: According to the response, I got the answer\nFinalAnswer: 140'),
+
+                dict(role='HUMAN', prompt='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?'),
+                dict(role='BOT', prompt="""Tool:PythonInterpreter\nTool Input:def solution():\n    num_fruits_per_first_three_basket = 9 + 15 + 14\n    num_fruits_first_three_basket = num_fruits_per_first_three_basket * 3\n    num_apple_fourth_basket = 9 - 2\n    num_orange_fourth_basket = 15 - 2\n    num_banana_fourth_basket = 14 - 2\n    num_fruits_fourth_basket = num_apple_fourth_basket + num_orange_fourth_basket + num_banana_fourth_basket\n    num_fruits_total = num_fruits_first_three_basket + num_fruits_fourth_basket\n    result = num_fruits_total\n    return result"""),
+                dict(role='SYSTEM', prompt='Response:146'),
+                dict(role='BOT', prompt='Thought: According to the response, I got the answer\nFinalAnswer: 146'),
+
+                dict(role='HUMAN', prompt='{question}'),
+            ])),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=AgentInferencer),
+)
+
+gsm8k_eval_cfg = dict(
+    evaluator=dict(type=Gsm8kAgentEvaluator),
+    pred_postprocessor=dict(type=gsm8k_postprocess),
+    dataset_postprocessor=dict(type=gsm8k_dataset_postprocess),
+)
+
+gsm8k_datasets = [
+    dict(
+        abbr='gsm8k-agent',
+        type=GSM8KDataset,
+        path='./data/gsm8k',
+        reader_cfg=gsm8k_reader_cfg,
+        infer_cfg=gsm8k_infer_cfg,
+        eval_cfg=gsm8k_eval_cfg,
+    )
+]

configs/datasets/gsm8k/gsm8k_gen_d6de81.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import GSM8KDataset, gsm8k_postprocess, gsm8k_dataset_postprocess, Gsm8kEvaluator
+
+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: 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=Gsm8kEvaluator),
+                      pred_postprocessor=dict(type=gsm8k_postprocess),
+                      dataset_postprocessor=dict(type=gsm8k_dataset_postprocess))
+
+gsm8k_datasets = [
+    dict(
+        abbr='gsm8k',
+        type=GSM8KDataset,
+        path='./data/gsm8k',
+        reader_cfg=gsm8k_reader_cfg,
+        infer_cfg=gsm8k_infer_cfg,
+        eval_cfg=gsm8k_eval_cfg)
+]

configs/datasets/math/math_agent_gen_af2293.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import AgentInferencer
+from opencompass.datasets import (
+    MATHDataset, MATHAgentEvaluator, math_postprocess
+)
+# use pal format but not perform well
+math_reader_cfg = dict(input_columns=['problem'], output_column='solution')
+
+math_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(
+            round=[
+                # # ################################### NEW SHOT ###################################
+                dict(role='HUMAN', prompt='Find the coefficient of $x^3$ when $3(x^2 - x^3+x) +3(x +2x^3- 3x^2 + 3x^5+x^3) -5(1+x-4x^3 - x^2)$ is simplifie.'),
+                dict(role='BOT', prompt="""Tool:PythonInterpreter\nTool Input:from sympy import symbols, simplify
+
+def solution():
+    x = symbols('x')
+    expr = 3*(x**2 - x**3 + x) + 3*(x + 2*x**3 - 3*x**2 + 3*x**5 + x**3) - 5*(1 + x - 4*x**3 - x**2)
+    simplified_expr = simplify(expr)
+
+    x3_coefficient = simplified_expr.as_coefficients_dict()[x**3]
+    result = x3_coefficient
+    return result"""),
+                dict(role='SYSTEM', prompt='Response:26'),
+                dict(role='BOT', prompt='FinalAnswer: The final answer is $26$. I hope it is correct.'),
+                dict(role='HUMAN', prompt='The surface area of a sphere with radius $r$ is $4\pi r^2$. Including the area of its circular base, what is the total surface area of a hemisphere with radius 6 cm? Express your answer in terms of $\pi$.'),
+                dict(role='BOT', prompt="""Tool:PythonInterpreter\nTool Input:import math
+
+def solution():
+    radius = 6
+
+    # Surface area of the hemisphere
+    hemisphere_area = 2 * math.pi * radius**2
+
+    # Area of the circular base
+    base_area = math.pi * radius**2
+
+    # Total surface area
+    total_surface_area = hemisphere_area + base_area
+
+    # Formatting the result in LaTeX
+    result = r'{}\pi'.format(total_surface_area / math.pi)
+    return result"""),
+                dict(role='SYSTEM', prompt='Response:108.0\\pi'),
+                dict(role='BOT', prompt='FinalAnswer: The final answer is $108.0\pi$. I hope it is correct.'),
+                dict(role='HUMAN', prompt='Monica tosses a fair 6-sided die.  If the roll is a prime number, then she wins that amount of dollars (so that, for example, if she rolls 3, then she wins 3 dollars).  If the roll is composite, she wins nothing. Otherwise, she loses 3 dollars. What is the expected value of her winnings on one die toss? Express your answer as a dollar value to the nearest cent.'),
+                dict(role='BOT', prompt="""Tool:PythonInterpreter\nTool Input:def solution():
+    # Probabilities of each outcome
+    prime_prob = 1 / 6
+    composite_prob = 1 / 3
+    otherwise_prob = 1 / 6
+
+    # Expected value of each outcome
+    prime_expected_value = (2 * prime_prob) + (3 * prime_prob) + (5 * prime_prob)
+    composite_expected_value = 0 * composite_prob
+    otherwise_expected_value = -3 * otherwise_prob
+
+    # Total expected value
+    total_expected_value = prime_expected_value + composite_expected_value + otherwise_expected_value
+
+    # Dollar value to the nearest cent
+    result = "{:.2f}".format(total_expected_value)
+    return result"""),
+                dict(role='SYSTEM', prompt='Response:1.17'),
+                dict(role='BOT', prompt='FinalAnswer: The final answer is $1.17$. I hope it is correct.'),
+                dict(role='HUMAN', prompt="Given $\mathbf{a} = \\begin{pmatrix} -7 \\ 0 \\ 1 \end{pmatrix}$ and $\mathbf{b} = \\begin{pmatrix} 4 \\ 2 \\ -1 \end{pmatrix},$ find $\mathbf{a} - 3 \mathbf{b}.$"),
+                dict(role='BOT', prompt="""Tool:PythonInterpreter\nTool Input:import numpy as np
+
+def solution()
+    a = np.array([-7, 0, 1])
+    b = np.array([4, 2, -1])
+
+    result = a - 3 * b
+
+    result = r'\\begin{{pmatrix}} {} \ {} \ {} \end{{pmatrix}}'.format(result[0], result[1], result[2])
+    return result"""),
+                dict(role='SYSTEM', prompt='Response:\\begin{pmatrix} -19 \\ -6 \\ 4 \\end{pmatrix}'),
+                dict(role='BOT', prompt='FinalAnswer: The final answer is $\\begin{pmatrix} -19 \\ -6 \\ 4 \\end{pmatrix}$. I hope it is correct.'),
+                dict(role='HUMAN', prompt='{problem}'),
+            ])),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=AgentInferencer),
+)
+
+math_eval_cfg = dict(
+    evaluator=dict(type=MATHAgentEvaluator),
+    pred_postprocessor=dict(type=math_postprocess),
+)
+
+math_datasets = [
+    dict(
+        abbr='math-agent',
+        type=MATHDataset,
+        path='./data/math/math.json',
+        reader_cfg=math_reader_cfg,
+        infer_cfg=math_infer_cfg,
+        eval_cfg=math_eval_cfg,
+    )
+]

configs/datasets/math/math_gen_1ed9c2.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import MATHDataset, MATHEvaluator, math_postprocess
+
+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 coefficient of $x^3$ when $3(x^2 - x^3+x) +3(x +2x^3- 3x^2 + 3x^5+x^3) -5(1+x-4x^3 - x^2)$ is simplified.\nSolution:"
+            ),
+            dict(
+                role="BOT",
+                prompt=
+                "Combine like terms to simplify the expression. The coefficient of $x^3$ is calculated as $$(-3+2\cdot(2+1))+(-5)\cdot(-4))$ = 26$. Thus, the coefficient of $x^3$ is $\\boxed{26}$.\nFinal Answer: The final answer is $26$. I hope it is correct.\n"
+            ),
+            dict(
+                role="HUMAN",
+                prompt=
+                "Problem:\nThe surface area of a sphere with radius $r$ is $4\pi r^2$. Including the area of its circular base, what is the total surface area of a hemisphere with radius 6 cm? Express your answer in terms of $\pi$.\nSolution:"
+            ),
+            dict(
+                role="BOT",
+                prompt=
+                "The surface area of a hemisphere (not including the base) is half that of a sphere, so it is $2\pi r^2$. The area of the base is $\pi r^2$. Therefore, for a hemisphere with radius 6 cm, the total surface area is $2\pi (6)^2 + \pi (6)^2 = 108\pi$ square cm.\nFinal Answer: The final answer is $108\pi$ square cm. I hope it is correct.\n"
+            ),
+            dict(
+                role="HUMAN",
+                prompt=
+                "Problem:\nMonica tosses a fair 6-sided die. If the roll is a prime number, then she wins that amount of dollars (so that, for example, if she rolls 3, then she wins 3 dollars). If the roll is composite, she wins nothing. Otherwise, she loses 3 dollars. What is the expected value of her winnings on one die toss? Express your answer as a dollar value to the nearest cent.\nSolution:"
+            ),
+            dict(
+                role="BOT",
+                prompt=
+                "The prime numbers rolled could be 2, 3, or 5, and each has a 1/6 chance of being rolled. The composite number 4 or 6 has a 2/6 chance of being rolled, but it results in $0 win. The remaining non-prime and non-composite number is 1 , and it results in a loss of $3, with a 1/6 chance. So, the expected winnings are $(2+3+5)(1/6)+0(2/6)+(-3)(1/6) = \$1.17$.\nFinal Answer: The final answer is $\$1.17$. I hope it is correct.\n"
+            ),
+            dict(
+                role="HUMAN",
+                prompt=
+                "Problem:\nGiven $\mathbf{a} = \\begin{pmatrix} -7 \\ 0 \\ 1 \end{pmatrix}$ and $\mathbf{b} = \\begin{pmatrix} 4 \\ 2 \\ -1 \end{pmatrix},$ find $\mathbf{a} - 3 \mathbf{b}.$\nSolution:"
+            ),
+            dict(
+                role="BOT",
+                prompt=
+                "We find $3 \mathbf{b}$ first, which is $\\begin{pmatrix} 12 \\ 6 \\ -3 \end{pmatrix}$. Then we subtract this vector from $\mathbf{a}$. So, $\mathbf{a} - 3 \mathbf{b} = \\begin{pmatrix} -7 - 12 \\ 0 - 6 \\ 1 - (-3) \end{pmatrix} = \\begin{pmatrix} -19 \\ -6 \\ 4 \end{pmatrix}.$\nFinal Answer: The final answer is $\\begin{pmatrix} -19 \\ -6 \\ 4 \end{pmatrix}$. I hope it is correct.\n"
+            ),
+            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_postprocess))
+
+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)
+]

configs/eval_chat_agent.py

@@ -9,8 +9,8 @@ from lagent import ReAct
 from lagent.agents.react import ReActProtocol
 
 with read_base():
-    from .datasets.gsm8k.gsm8k_agent_gen_3ac57d import gsm8k_datasets
-    from .datasets.math.math_agent_gen_861b4f import math_datasets
+    from .datasets.gsm8k.gsm8k_agent_gen_be1606 import gsm8k_datasets
+    from .datasets.math.math_agent_gen_af2293 import math_datasets
     from .datasets.MathBench.mathbench_agent_gen_568903 import mathbench_agent_datasets
     from .summarizers.math_agent import summarizer
 

configs/eval_chat_agent_baseline.py

+from mmengine.config import read_base
+from opencompass.models.openai_api import OpenAI
+from opencompass.partitioners import SizePartitioner
+from opencompass.runners import LocalRunner
+from opencompass.tasks import OpenICLInferTask
+
+with read_base():
+    from .datasets.gsm8k.gsm8k_gen_d6de81 import gsm8k_datasets
+    from .datasets.math.math_gen_1ed9c2 import math_datasets
+    from .datasets.MathBench.mathbench_gen import mathbench_datasets
+    from .summarizers.math_baseline import summarizer
+
+datasets = []
+datasets += gsm8k_datasets
+datasets += math_datasets
+datasets += mathbench_datasets
+
+
+models = [
+    dict(
+        abbr='gpt-3.5-react',
+        type=OpenAI,
+        path='gpt-3.5-turbo',
+        key='ENV',
+        query_per_second=1,
+        max_seq_len=4096,
+        batch_size=1,
+    ),
+]
+
+infer = dict(
+    partitioner=dict(type=SizePartitioner, max_task_size=1000),
+    runner=dict(
+        type=LocalRunner,
+        max_num_workers=16,
+        task=dict(type=OpenICLInferTask)),
+)
\ No newline at end of file

configs/summarizers/groups/mathbench_agent.py

+
+mathbench_agent_summary_groups = [
+    {
+        'name': 'mathbench-college-agent',
+        'subsets': [
+            ['mathbench-college-single_choice_cn-agent', 'acc_1'],
+            ['mathbench-college-cloze_en-agent', 'accuracy'],
+        ]
+    },
+    {
+        'name': 'mathbench-high-agent',
+        'subsets': [
+            ['mathbench-high-single_choice_cn-agent', 'acc_1'],
+            ['mathbench-high-single_choice_en-agent', 'acc_1'],
+        ]
+    },
+    {
+        'name': 'mathbench-middle-agent',
+        'subsets': [
+            ['mathbench-middle-single_choice_cn-agent', 'acc_1'],
+        ]
+    },
+    {
+        'name': 'mathbench-primary-agent',
+        'subsets': [
+            ['mathbench-primary-cloze_cn-agent', 'accuracy'],
+        ]
+    },
+    {
+        'name': 'mathbench-agent',
+        'subsets': [
+            'mathbench-college-agent',
+            'mathbench-high-agent',
+            'mathbench-middle-agent',
+            'mathbench-primary-agent',
+        ],
+    },
+    {
+        'name': 'mathbench-college-circular-agent',
+        'subsets': [
+            ['mathbench-college-single_choice_cn-agent', 'perf_4'],
+        ]
+    },
+    {
+        'name': 'mathbench-high-circular-agent',
+        'subsets': [
+            ['mathbench-high-single_choice_cn-agent', 'perf_4'],
+            ['mathbench-high-single_choice_en-agent', 'perf_4'],
+        ]
+    },
+    {
+        'name': 'mathbench-middle-circular-agent',
+        'subsets': [
+            ['mathbench-middle-single_choice_cn-agent', 'perf_4'],
+        ]
+    },
+    {
+        'name': 'mathbench-circular-agent',
+        'subsets': [
+            'mathbench-college-circular-agent',
+            'mathbench-high-circular-agent',
+            'mathbench-middle-circular-agent',
+        ],
+    },
+    {
+        'name': 'mathbench-circular-and-cloze-agent',
+        'subsets': [
+            'mathbench-high-circular-agent',
+            'mathbench-middle-circular-agent',
+            'mathbench-circular-agent',
+            'mathbench-college-cloze_en-agent',
+            'mathbench-primary-cloze_cn-agent',
+        ],
+    }
+]

configs/summarizers/math_agent.py

+from mmengine.config import read_base
+
+with read_base():
+    from .groups.mathbench_agent import mathbench_agent_summary_groups
+
 summarizer = dict(
     dataset_abbrs=[
         '######## GSM8K-Agent Accuracy ########', # category
@@ -11,17 +16,9 @@ summarizer = dict(
         ['math-agent', 'code_acc'],
         ['math-agent', 'action_pct'],
         '######## MathBench-Agent Accuracy ########', # category
-        ['mathbench-college-single_choice_cn-agent', 'acc_1'],
-        ['mathbench-college-cloze_en-agent', 'accuracy'],
-        ['mathbench-high-single_choice_cn-agent', 'acc_1'],
-        ['mathbench-high-single_choice_en-agent', 'acc_1'],
-        ['mathbench-middle-single_choice_cn-agent', 'acc_1'],
-        ['mathbench-primary-cloze_cn-agent', 'accuracy'],
-        '######## MathBench-Agent CircularEval ########', # category
-        ['mathbench-college-single_choice_cn-agent', 'perf_4'],
-        ['mathbench-high-single_choice_cn-agent', 'perf_4'],
-        ['mathbench-high-single_choice_en-agent', 'perf_4'],
-        ['mathbench-middle-single_choice_cn-agent', 'perf_4'],
+        'mathbench-agent',
+        'mathbench-circular-agent',
+        'mathbench-circular-and-cloze-agent',
     ],
     summary_groups=sum(
         [v for k, v in locals().items() if k.endswith("_summary_groups")], [])

configs/summarizers/math_baseline.py

+from mmengine.config import read_base
+
+with read_base():
+    from .groups.mathbench import mathbench_summary_groups
+
+summarizer = dict(
+    dataset_abbrs=[
+        '######## GSM8K Accuracy ########', # category
+        ['gsm8k', 'accuracy'],
+        '######## MATH Accuracy ########', # category
+        ['math', 'accuracy'],
+        '######## MathBench-Agent Accuracy ########', # category
+        'mathbench',
+        'mathbench-circular',
+        'mathbench-circular-and-cloze',
+    ],
+    summary_groups=sum(
+        [v for k, v in locals().items() if k.endswith("_summary_groups")], [])
+)