TACO: Topics in Algorithmic COde generation dataset

2023/12/22 by Rongni Li, Li, Rongao, Jie Fu +15 · 29 citations
Computer Science · #Advanced Malware Detection Techniques #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Software Engineering Research #Software Testing and Debugging Techniques

paper · pdf · doi:10.48550/arxiv.2312.14852

openalex publication_date 2023/12/22 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We introduce TACO, an open-source, large-scale code generation dataset, with a focus on the optics of algorithms, designed to provide a more challenging training dataset and evaluation benchmark in the field of code generation models. TACO includes competition-level programming questions that are more challenging, to enhance or evaluate problem understanding and reasoning abilities in real-world programming scenarios. There are 25433 and 1000 coding problems in training and test set, as well as up to 1.55 million diverse solution answers. Moreover, each TACO problem includes several fine-grained labels such as task topics, algorithms, programming skills, and difficulty levels, providing a more precise reference for the training and evaluation of code generation models. The dataset and evaluation scripts are available on Hugging Face Hub (https://huggingface.co/datasets/BAAI/TACO) and Github (https://github.com/FlagOpen/TACO).

Cited by

• GLM-5: from Vibe Coding to Agentic Engineering
• OmniOPD: Logit-Free On-Policy Distillation via Speculative Verification
• How LLM Task-Adaptation Reshapes Alignment: A Multi-dimensional Study of Behavioral and Representational Drift
• Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
• ATLAS: Automated Toolkit for Large-Scale Verified Code Synthesis
• The Peril of Preference: Why GRPO fails on Ordinal Rewards
• Verifiable Accuracy and Abstention Rewards in Curriculum RL to Alleviate Lost-in-Conversation
• CodeRL+: Improving Code Generation via Reinforcement with Execution Semantics Alignment
• Reasoning Distillation and Structural Alignment for Improved Code Generation
• Leave It to the Experts: Detecting Knowledge Distillation via MoE Expert Signatures
• The Valley of Code Reasoning: Scaling Knowledge Distillation of Large Language Models
• Learning on the Job: Test-Time Curricula for Targeted Reinforcement Learning
• TROLL: Trust Regions improve Reinforcement Learning for Large Language Models
• AP2O-Coder: Adaptively Progressive Preference Optimization for Reducing Compilation and Runtime Errors in LLM-Generated Code
• BroRL: Scaling Reinforcement Learning via Broadened Exploration
• Learning to Reason as Action Abstractions with Scalable Mid-Training RL
• AutoCode: LLMs as Problem Setters for Competitive Programming
• Variational Reasoning for Language Models
• Verification Limits Code LLM Training
• RL Grokking Recipe: How Does RL Unlock and Transfer New Algorithms in LLMs?
• Calibrated Reasoning: An Explanatory Verifier for Dynamic and Efficient Problem-Solving
• Can Multi-turn Self-refined Single Agent LMs with Retrieval Solve Hard Coding Problems?
• Learning to Generate Unit Test via Adversarial Reinforcement Learning
• Your Reward Function for RL is Your Best PRM for Search: Unifying RL and Search-Based TTS
• Accelerating Edge Inference for Distributed MoE Models with Latency-Optimized Expert Placement
• VisCodex: Unified Multimodal Code Generation via Merging Vision and Coding Models
• Klear-Reasoner: Advancing Reasoning Capability via Gradient-Preserving Clipping Policy Optimization
• Klear-CodeTest: Scalable Test Case Generation for Code Reinforcement Learning
• Cognitive Kernel-Pro: A Framework for Deep Research Agents and Agent Foundation Models Training

Related

• Knowledge-Aware Code Generation with Large Language Models
• GPT-Neo: Large Scale Autoregressive Language Modeling with Mesh-Tensorflow
• KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding
• CodeNet: A Large-Scale AI for Code Dataset for Learning a Diversity of Coding Tasks
• CodeT: Code Generation with Generated Tests