EffiBench-X: A Multi-Language Benchmark for Measuring Efficiency of LLM-Generated Code
Paper • 2505.13004 • Published
id stringlengths 3 12 | title stringlengths 3 66 | title_slug stringlengths 3 66 | description stringlengths 39 25.4k | description_md stringlengths 39 4.82k | difficulty stringclasses 113
values | tags listlengths 0 9 | source stringclasses 5
values | url stringlengths 37 96 | type stringclasses 2
values | release_timestamp int64 1.7B 1.73B ⌀ | release_date stringlengths 19 19 ⌀ | time_limit_nanos int64 1B 9B ⌀ | memory_limit_bytes int64 256M 2.1B ⌀ | starter_code dict | solutions dict | test_case_generator stringlengths 521 16.9k | evaluator stringlengths 200 5.31k | generated_tests stringlengths 3.2k 359M | test_runners dict |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1444 | Yokohama Phenomena | yokohama-phenomena | Yokohama Phenomena
Do you know about Yokohama Phenomena? The phenomenon takes place when three programmers, sitting around a table, hold a single pen together above a board. A grid of squares is drawn on the board, with each square marked with a single letter. Although none of the participants purposely moves the pen,... | Yokohama Phenomena
Do you know about Yokohama Phenomena? The phenomenon takes place when three programmers, sitting around a table, hold a single pen together above a board. A grid of squares is drawn on the board, with each square marked with a single letter. Although none of the participants purposely moves the pen,... | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1444?year=2023 | io | null | null | 2,000,000,000 | 2,097,152,000 | null | {
"cpp": {
"code": "#include <iostream>\n#include <vector>\n#include <string>\n\nusing namespace std;\n\nint n, m;\nvector<string> grid;\nstring target = \"YOKOHAMA\";\nint target_len = 8;\nint directions[4][2] = {{0, 1}, {1, 0}, {0, -1}, {-1, 0}};\n\nbool isValid(int x, int y) {\n return x >= 0 && x < n && y ... | def generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:
import random
# collections is not strictly needed for the current DFS implementation
# import collections
random.seed(seed)
# --- Solver (adapted from problem description) ---
# Memoization cache for the DFS. Must be rese... | def evaluate(expected_output: str, program_output: str) -> bool:
import math # Required by instructions, though not used for this integer-only problem.
# Kept for adherence, can be removed if strictly not needed.
# --- Deserialization logic (must be independent) ---
# This problem's output ... | [{"input": "2 4\nYOHA\nOKAM", "output": "8"}, {"input": "3 4\nYOKH\nOKHA\nKHAM", "output": "0"}, {"input": "3 6\nMAYOHA\nAHOKAM\nMAYOHA", "output": "80"}, {"input": "1 1\nA", "output": "0"}, {"input": "1 1\nH", "output": "0"}, {"input": "1 1\nK", "output": "0"}, {"input": "1 1\nM", "output": "0"}, {"input": "1 1\nO", "... | null | |
1447 | Nested Repetition Compression | nested-repetition-compression | "Nested Repetition Compression\n\nYou have a number of strings of lowercase letters to be sent in e-(...TRUNCATED) | "Nested Repetition Compression\n\nYou have a number of strings of lowercase letters to be sent in e-(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1447?year=2023 | io | null | null | 2,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"#include <bits/stdc++.h>\nusing namespace std;\n#if __has_include(<atcoder/all>)\n#i(...TRUNCATED) | "def generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n import random\n from(...TRUNCATED) | "def evaluate(expected_output: str, program_output: str) -> bool:\n # No external library imports(...TRUNCATED) | "[{\"input\": \"abababaaaaa\", \"output\": \"3(ab)5(a)\"}, {\"input\": \"abababcaaaaaabababcaaaaa\",(...TRUNCATED) | null | |
1448 | Chayas | chayas | "Chayas\n\nOnce upon a time, there were a number of chayas (teahouses) along one side of an east-wes(...TRUNCATED) | "Chayas\n\nOnce upon a time, there were a number of chayas (teahouses) along one side of an east-wes(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1448?year=2023 | io | null | null | 8,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"#include<bits/stdc++.h>\nusing namespace std;\nusing ll = long long;\nusing vll = ve(...TRUNCATED) | "def generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n import random\n impo(...TRUNCATED) | "def evaluate(expected_output: str, program_output: str) -> bool:\n \"\"\"\n Determine if a su(...TRUNCATED) | "[{\"input\": \"5 4\\n1 2 4\\n2 3 5\\n3 2 4\\n1 3 2\\n\", \"output\": \"4\"}, {\"input\": \"4 2\\n3 (...TRUNCATED) | null | |
1449 | Color Inversion on a Huge Chessboard | color-inversion-on-a-huge-chessboard | "Color Inversion on a Huge Chessboard\n\nYou are given a set of square cells arranged in a chessboar(...TRUNCATED) | "Color Inversion on a Huge Chessboard\n\nYou are given a set of square cells arranged in a chessboar(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1449?year=2023 | io | null | null | 4,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"#include <bits/stdc++.h>\n\n// NOLINTBEGIN\n// clang-format off\n// DO NOT REMOVE TH(...TRUNCATED) | "import random\n\ndef generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n random(...TRUNCATED) | "import math # Required by template, not used for this problem's integer outputs\n\ndef evaluate(exp(...TRUNCATED) | "[{\"input\": \"159 3\\nROW 7\\nROW 88\\nCOLUMN 46\\n\", \"output\": \"24963\\n24645\\n24335\\n\"}, (...TRUNCATED) | null | |
1455 | Ribbon on the Christmas Present | ribbon-on-the-christmas-present | "Ribbon on the Christmas Present\n\nYou are preparing a ribbon to decorate the Christmas present box(...TRUNCATED) | "Ribbon on the Christmas Present\n\nYou are preparing a ribbon to decorate the Christmas present box(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1455?year=2024 | io | null | null | 2,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"#include<bits/stdc++.h>\n#define rep(i,j,n) for(ll i=j;i<(ll)(n);i++)\n#define rrep((...TRUNCATED) | "def generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n import random\n impo(...TRUNCATED) | "def evaluate(expected_output: str, program_output: str) -> bool:\n \n # Helper for deserializ(...TRUNCATED) | "[{\"input\": \"6\\n50 100 50 50 100 50\", \"output\": \"3\\n\"}, {\"input\": \"5\\n1 2 3 2 1\", \"o(...TRUNCATED) | null | |
1456 | The Sparsest Number in Between | the-sparsest-number-in-between | "The Sparsest Number in Between\n\nYou are given a pair of positive integers $a$ and $b$ ($a \\leq b(...TRUNCATED) | "The Sparsest Number in Between\n\nYou are given a pair of positive integers $a$ and $b$ ($a \\leq b(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1456?year=2024 | io | null | null | 2,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"#include <iostream>\n#include <climits>\n\nusing namespace std;\n\nlong long countOn(...TRUNCATED) | "import random\n\n# Task 1: Test Case Generator\ndef generate_test_cases(num_cases: int, seed: int =(...TRUNCATED) | "import math # Not strictly needed for this problem, but good practice for the template\n\n# Task 2:(...TRUNCATED) | "[{\"input\": \"10 13\", \"output\": \"10\"}, {\"input\": \"11 15\", \"output\": \"12\"}, {\"input\"(...TRUNCATED) | null | |
1459 | E-Circuit Is Now on Sale! | e-circuit-is-now-on-sale! | "E-Circuit Is Now on Sale!\n\nAre you looking for math education tools for your children? Then, why (...TRUNCATED) | "E-Circuit Is Now on Sale!\n\nAre you looking for math education tools for your children? Then, why (...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1459?year=2024 | io | null | null | 2,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"// AOJ #1459\n// E-Circuit Is Now on Sale! 2025.1.14\n\n#include <bits/stdc++.h>\nus(...TRUNCATED) | "def generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n import random\n impo(...TRUNCATED) | "def evaluate(expected_output: str, program_output: str) -> bool:\n # Deserialization for output:(...TRUNCATED) | "[{\"input\": \"6 8\\n3.......\\n#....P..\\n#....#.2\\n#.###*#+\\n##-....#\\n..1...4#\", \"output\":(...TRUNCATED) | null | |
1460 | The Farthest Point | the-farthest-point | "The Farthest Point\n\nAnant is on one of the vertices, say the starting vertex, of a rectangular cu(...TRUNCATED) | "The Farthest Point\n\nAnant is on one of the vertices, say the starting vertex, of a rectangular cu(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1460?year=2024 | io | null | null | 2,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"#include <bits/stdc++.h>\nusing namespace std;\n\n\n\n//----------------------------(...TRUNCATED) | "def generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n import math\n import(...TRUNCATED) | "def evaluate(expected_output: str, program_output: str) -> bool:\n import math\n\n # Deserial(...TRUNCATED) | "[{\"input\": \"11 20 10\", \"output\": \"29.09682113221305\"}, {\"input\": \"84 51 41\", \"output\"(...TRUNCATED) | null | |
1464 | Mixing Solutions | mixing-solutions | "Mixing Solutions\n\nLet’s prepare for an experiment with the chemical Yokohama Yellow, or YY in s(...TRUNCATED) | "Mixing Solutions\n\nLet’s prepare for an experiment with the chemical Yokohama Yellow, or YY in s(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Regional/1464?year=2024 | io | null | null | 2,000,000,000 | 2,097,152,000 | null | {"cpp":{"code":"// O(n log n log M): yosupo さんの解法\n#include <bits/stdc++.h>\nusing namespa(...TRUNCATED) | "def generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n import random\n from(...TRUNCATED) | "def evaluate(expected_output: str, program_output: str) -> bool:\n from fractions import Fractio(...TRUNCATED) | "[{\"input\": \"3 10 5000\\n10 2000 3000\\n10 4000 6000\\n10 7000 8000\", \"output\": \"1 2\"}, {\"i(...TRUNCATED) | null | |
1664 | Which Team Should Receive the Sponsor Prize? | which-team-should-receive-the-sponsor-prize? | "Problem A\nWhich Team Should Receive the Sponsor Prize?\n\nIn ICQC (International Collegiate Quiz C(...TRUNCATED) | "Problem A\nWhich Team Should Receive the Sponsor Prize?\n\nIn ICQC (International Collegiate Quiz C(...TRUNCATED) | [] | aizu | https://onlinejudge.u-aizu.ac.jp/challenges/sources/ICPC/Prelim/1664?year=2023 | io | null | null | 8,000,000,000 | 262,144,000 | null | {"cpp":{"code":"#include <stdio.h>\n#include <stdlib.h>\n\n#define SEC 2023\n\nint main()\n{\n int (...TRUNCATED) | "import random\n\ndef generate_test_cases(num_cases: int, seed: int = 42) -> list[dict]:\n random(...TRUNCATED) | "import math # math.isclose is not needed for this problem as outputs are integers\n\ndef evaluate(e(...TRUNCATED) | "[{\"input\": \"5\\n8936 1425 2020 9675 6913\\n6\\n2023 3812 8280 9864 435 9196\\n0\\n\", \"output\"(...TRUNCATED) | null |
End of preview. Expand in Data Studio
Dataset Card for EffiBench-X
EffiBench-X is the first multi-language benchmark designed specifically to evaluate the efficiency of LLM-generated code across six programming languages: Python, C++, Java, JavaScript, Ruby, and Golang. The dataset comprises 623 competitive programming problems paired with human expert solutions as efficiency baselines.
Dataset Details
Dataset Description
EffiBench-X addresses critical limitations in existing code generation benchmarks by providing:
Multi-language evaluation across Python, C++, Java, JavaScript, Ruby, and Golang
Efficiency-focused metrics including execution time, memory peak, and memory integral
Human expert baselines for reliable efficiency comparison
Curated by: Yuhao Qing, Boyu Zhu, Mingzhe Du, Zhijiang Guo, Terry Yue Zhuo, Qianru Zhang, Jie M. Zhang, Heming Cui, Siu-Ming Yiu, Dong Huang, See-Kiong Ng, Luu Anh Tuan
Institutions: HKU, UCL, NTU, NUS, HKUST, Monash University, CSIRO's Data61, KCL
Language(s) (NLP): English
License: Apache License 2.0
Dataset Sources
- Repository: EffiBench-X (GitHub)
- Dataset: EffiBench/effibench-x
- Paper: arXiv:2505.13004
- Problem Sources:
Uses
Direct Use
- Benchmarking LLM code generation efficiency: Evaluate models on runtime performance, memory usage, and correctness across multiple languages
- Cross-language performance analysis: Compare model capabilities across different programming paradigms
- Model development: Train and fine-tune models for efficient code generation
- Research: Study efficiency gaps between LLM-generated and human expert code
Out-of-Scope Use
- Production deployment without validation: Solutions should be verified before production use
- Security-critical applications: The dataset focuses on algorithmic efficiency, not security
- Non-competitive programming domains: Problems are algorithmic in nature and may not represent all software engineering contexts
Dataset Structure
The dataset contains 623 problems categorized into:
- Functional problems: Implement specific functions/classes with I/O handled by test templates
- Standard I/O problems: Complete programs reading from stdin and writing to stdout
Key fields per record include:
id,title,title_slug,description,description_md,difficulty,tags,source,url,type- Limits:
time_limit_nanos,memory_limit_bytes - Code artifacts:
starter_code: language-keyed starter snippetssolutions: language-keyed canonical solutions (e.g., forcpp,golang,java,javascript,python3,ruby)test_case_generator: executable code string that programmatically produces testsevaluator: executable code string to determine pass/fail given expected vs. program outputgenerated_tests: serialized tests produced by the generatortest_runners: language-keyed runner templates for executing solutions
All problems are from competitive programming platforms.
Dataset Creation
Curation Rationale
Existing code generation benchmarks primarily focus on functional correctness with limited attention to efficiency, often restricted to Python. EffiBench-X addresses three critical limitations:
- Language diversity: Extends beyond Python to include statically-typed (C++, Java, Go) and dynamically-typed languages (Python, JavaScript, Ruby)
- Data contamination: Uses recent problems (post-October 2023) to avoid memorization effects
- Complexity: Features algorithmically challenging problems requiring optimization techniques
Source Data
Data Collection and Processing
Problems are curated from competitive programming platforms. Each problem includes:
- Human expert solutions verified for correctness and efficiency
- 100 programmatically generated test cases
- Test runners and evaluators for automated assessment
- Cross-language validation to ensure consistency
Who are the source data producers?
- Problem creators: Competitive programming platforms and contest organizers
- Solution authors: Human expert programmers from competitive programming communities
- Dataset curators: EffiBench research team
Citation
Please cite our paper if you use this dataset:
@article{qing2025effibench,
title={EffiBench-X: A Multi-Language Benchmark for Measuring Efficiency of LLM-Generated Code},
author={Qing, Yuhao and Zhu, Boyu and Du, Mingzhe and Guo, Zhijiang and Zhuo, Terry Yue and Zhang, Qianru and Zhang, Jie M and Cui, Heming and Yiu, Siu-Ming and Huang, Dong and Ng, See-Kiong and Tuan, Luu Anh},
journal={Advances in neural information processing systems},
year={2025}
}
More Information
- Dataset Statistics: 623 problems, 100 test cases per problem, 6 languages
- Evaluation: Sandboxed execution environment for consistent performance measurements
- For detailed information and benchmark results, please refer to the paper and GitHub repository
Dataset Card Contact
For questions and feedback, please open an issue on our GitHub repository.
- Downloads last month
- 184