Buffer Boundary Verification Benchmark

#Boundary Safety Verification #Out-of-Bounds Access Detection #Code Reasoning #Reward Evaluation #Code Auditing #Memory Safety Research #Software Security Evaluation #Reinforcement Learning Training

3320
0.3GB
JSON
CATL
MOBIUSI INC

Updated:2026-11-05

AI Analysis & Value Prop

This benchmark consists of source-code verification samples involving arrays, buffers, string operations, and pointer arithmetic, covering boundary-safety issues such as index ranges, length calculations, capacity constraints, and memory accesses. Each sample includes memory-access records, boundary checks, an overall verdict, and an analysis rationale, along with a reward score from 0 to 1. It supports evaluation of code understanding and boundary verification, as well as verifiable-reward reinforcement learning and memory-safety research.

Dataset Insights

Sample Examples

{}

Technical Specifications

Field	Type	Description
file_name	string	File name
verdict	object	Summary judgment of the overall memory boundary safety of the source code execution paths.
source_code	string	Source code to verify, involving arrays, buffers, strings, or pointer arithmetic.
reward_score	number	Explicit reward score calculated from the boundary checks and overall verdict, ranging from 0 to 1.
analysis_reason	string	Explains how indices, length calculations, memory accesses, and control-flow conditions support the final verdict.
boundary_checks	array	Explicit checks on index ranges, length calculations, capacity constraints, and pointer positions.
memory_accesses	array	Extracted records of array indexing, buffer accesses, string operations, and pointer dereferences.
source_language	string	Programming language identified from the source code syntax.

Compliance Statement

Authorization Type	Proprietary - Commercial AI Training License (No Redistribution)
Commercial Use	Requires exclusive subscription or authorization contract (monthly or per-invocation charging)
Privacy and Anonymization	No PII, no real company names, simulated scenarios follow industry standards
Compliance System	Compliant with China's Data Security Law / EU GDPR / supports enterprise data access logs

Frequently Asked Questions

What kinds of code samples does the benchmark cover?: Samples involve source code using arrays, buffers, strings, and pointer arithmetic. They focus on whether index ranges, length calculations, capacity constraints, and memory accesses satisfy boundary conditions.

What verification information is provided for each sample?: The schema includes source code and its language, memory-access records, boundary checks, an overall verification verdict, and an analysis rationale. It also provides a verification reward score ranging from 0 to 1.

How can I obtain the data and confirm its format?: Check the dataset listing or contact the provider for available packages, download instructions, and file formats. No specific delivery format is stated here, so confirm the product details before use.

What are the dataset's licensing terms?: Specific licensing terms are not provided here. Before using, modifying, or redistributing the data, confirm the permitted uses with the provider and check the usage conditions for the source-code samples and related content.

Which tasks is this benchmark suitable for?: It is suitable for code understanding and verification, boundary-safety analysis, out-of-bounds access detection, memory-safety research, and model training or evaluation with verifiable rewards.

Can't find the data you need?

Post a request and let data providers reach out to you.

Get this Dataset

Verified for Enterprise Use

Cite this Work

@dataset{Mobiusi2026,
  title={Buffer Boundary Verification Benchmark},
  author={MOBIUSI INC},
  year={2026},
  url={https://www.mobiusi.com/datasets/8b2721424dd6d8f49b5125f41e8da827},
  urldate={2026-10-27},
  keywords={buffer boundary verification,memory safety benchmark,out-of-bounds access detection,source code verification,array bounds checking,pointer safety,code security evaluation,verifiable reward reinforcement learning},
  version={1.0}
}

Using this in research? Please cite us.

placeholder

Scene

Domain

Modality

Task