datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
smart_contractsmart-contract-aggregators-educationalseli-smartcontract-audit-sft-backup
SELI smart-contract audit SFT — v7.1
Evidence-first EVM/Solidity audit SFT mix, deterministically rebuilt and
verified. Supersedes the v6.2-prepared mix (stage2 removed; the old state is
preserved on branch v6.2-prepared-backup and under legacy/v6.1).
Files
file
rows
sha256
train.jsonl
31,907
44c95d2f9e5a544e3d0b236d4157f84f1c857baf4ce212ba6235abe4216faaab
val.jsonl
730
906777d469d4c913f086aec1223fd17896f808f9204b8d3a5941135090154490
Every… See the full description on the dataset page: https://huggingface.co/datasets/0xtoshi/seli-smartcontract-audit-sft-backup.smart_contract_code_commentsSmart_Contract_HF_Tokenized
Dataset Card for "Smart_Contract_HF_Tokenized"
More Information needed
smart-contract-audit-findings
Smart Contract Audit Findings Dataset
A dataset of 49,611 smart contract security audit findings for fine-tuning LLMs on vulnerability detection.
Dataset Description
This dataset contains real security audit findings from 30 professional audit firms including Code4rena, OpenZeppelin, Sherlock, Cantina, and others. Formatted for training models to analyze smart contract code and identify vulnerabilities.
Splits
Split
Examples
Description
train
47,130… See the full description on the dataset page: https://huggingface.co/datasets/SkywardNomad92/smart-contract-audit-findings.smart-contracts-instructions
Smart Contracts Instructions
A dataset containing 6,003 GPT-generated human instruction and Solidity source code data pairs.
GPT models used to make this data are GPT-3.5 turbo, GPT-3.5 turbo 16k context, and GPT-4. Solidity source codes are used from mwritescode's Slither Audited Smart Contracts (https://huggingface.co/datasets/mwritescode/slither-audited-smart-contracts).
Distributions of the GPT models used to make this dataset:
GPT-3.5 Turbo: 5,276
GPT-3.5 Turbo 16k Context:… See the full description on the dataset page: https://huggingface.co/datasets/AlfredPros/smart-contracts-instructions.smart-contract-vuln-detectionSmart_Contract_HF_sc_succ
Dataset Card for "Smart_Contract_HF_sc_succ"
More Information needed
Smart_Contract_HF_bc_succ
Dataset Card for "Smart_Contract_HF_bc_succ"
More Information needed
smart-contract-vulnerabilitiesETH_SmartContractVulnerability_LLMsmart-contract-gensmart-contract-security-audit-v2smart_contract_vulnerability_kagglesmart-contracts-instructions-cleanedThe Cleaned Smart Contracts Instructions dataset is a refined version of the original dataset created by AlfredPros. This dataset contains structured and detailed instructions related to smart contracts, optimized for clarity and usability. It is intended for developers, researchers, and enthusiasts working on blockchain technology, smart contract development, and related fields.
blockchain-smartcontracts-1000
Blockchain y Contratos Inteligentes Dataset 1000
Dataset de 1000 instrucciones sobre los fundamentos de Blockchain, Contratos Inteligentes, DAOs, DeFi y programación en Solidity.
Uso
from datasets import load_dataset
dataset = load_dataset("miguelmejias0512/solidity_personal_dataset")
---
## Licencia
CC-BY-4.0 - Uso educativo
Smart-Contract-MultiTask-Dataset
Overview
This is a dataset designed for smart contract generation. It includes two subsets:
Requirement-FSM-Code subset: Contains user requirement descriptions, finite state machine (FSM) representations, and corresponding smart contract code.
Comment-Code subset: Includes functional comments and their corresponding implementation code.
Dataset Structure
Subset 1: Requirement-FSM-Code
Description: Contains natural language descriptions of user requirements… See the full description on the dataset page: https://huggingface.co/datasets/lohoz/Smart-Contract-MultiTask-Dataset.smart-contract-LLM-FT
Dataset Card for "smart-contract-LLM-FT"
More Information needed
smart_contracts_malicious
Dataset Card for "smart_contracts_malicious"
More Information needed
Smart_Contract_HF
Dataset Card for "Smart_Contract_HF"
More Information needed
smart-contracts-sourcecad_blockchain_smartcontract
Dataset Card for Dataset Name
Dataset Description
This dataset is collected from top token in etherscan which are verifed and opensource. It contains useful information within the token, like token address, holders, sourcecode and so on.
Besides, we pay much attention to analysis the source code of tokens’ contract, which refers to the tag column in this dataset. The value of tag contains LLM analysis of contract source code with the token, which is useful for solidity… See the full description on the dataset page: https://huggingface.co/datasets/tat1111/cad_blockchain_smartcontract.smart-contract-vuln-detectionSmart-Contract-Benchmark
Dataset Description
This dataset is used as a benchmark for evaluating the functional correctness of smart contract at both the multi-function level and the entire contract level.
Data Instances
{
'task_id': 1,
'user_requirement': 'This smart contract is an [ERC-721 NFT contract]. Please name the smart contract as [UniqueArt]. The main functionalities include [minting unique NFTs, transferring ownership, and managing approvals]. The main variables include......'… See the full description on the dataset page: https://huggingface.co/datasets/lohoz/Smart-Contract-Benchmark.smart-contract-kaggle-source-files
Smart Contract Kaggle Source Files
Non-CSV Kaggle source files. High-fanout source directories are stored as lossless CSV shards with content_base64.
Converted: {'bcccdatasets_bccc-vulscs-2023/Secure_SourceCodes/Secure_SourceCodes': 26914, 'bcccdatasets_bccc-vulscs-2023/Vulnerable_SourceCodes/Vulnerable_SourceCodes': 9756}
Raw non-CSV files: 4
Generated: 2026-06-24T16:03:35.197530+00:00
smart-contract-small
Dataset Card for "smart-contract-small"
More Information needed
smart-contract-w-Slither
Dataset Card for "smart-contract-w-Slither"
More Information needed
smart-contract-security-sft
Smart Contract Security SFT Dataset
Dataset for fine-tuning language models as Solidity smart contract security auditors.
Overview
327 training samples in ChatML conversational format
Sources: 291 attack vectors + 36 benchmark findings (DODO, Megapot, PoolTogether)
Format: Each sample has system prompt (auditor persona), user message (vulnerable code), assistant message (FINDING block + Foundry PoC)
Schema
Column
Type
Description
messages… See the full description on the dataset page: https://huggingface.co/datasets/oxdev/smart-contract-security-sft.ETH_SmartContractBugs_gpt