code generation
Code generation in AI involves automatically producing source code based on requirements, specifications, or higher-level abstractions, leveraging natural language processing or machine learning techniques for intelligent synthesis.
- Can LLMs Correct Themselves? A Benchmark of Self-Correction in LLMs
- CoRe: Benchmarking LLMs’ Code Reasoning Capabilities through Static Analysis Tasks
- EffiBench-X: A Multi-Language Benchmark for Measuring Efficiency of LLM-Generated Code
- GraLoRA: Granular Low-Rank Adaptation for Parameter-Efficient Fine-Tuning
- ICPC-Eval: Probing the Frontiers of LLM Reasoning with Competitive Programming Contests
- InstructFlow: Adaptive Symbolic Constraint-Guided Code Generation for Long-Horizon Planning
- Learning to Solve Complex Problems via Dataset Decomposition
- Let's Revise Step-by-Step: A Unified Local Search Framework for Code Generation with LLMs
- Lookahead Routing for Large Language Models
- MLE-STAR: Machine Learning Engineering Agent via Search and Targeted Refinement
- MaintainCoder: Maintainable Code Generation Under Dynamic Requirements
- Mitigating Overthinking in Large Reasoning Models via Manifold Steering
- Mixture of Inputs: Text Generation Beyond Discrete Token Sampling
- More Than Just Functional: LLM-as-a-Critique for Efficient Code Generation
- On-Policy Optimization with Group Equivalent Preference for Multi-Programming Language Understanding
- Practical and Effective Code Watermarking for Large Language Models
- Preserving LLM Capabilities through Calibration Data Curation: From Analysis to Optimization
- Rendering-Aware Reinforcement Learning for Vector Graphics Generation
- RepoMaster: Autonomous Exploration and Understanding of GitHub Repositories for Complex Task Solving
- ResearchCodeBench: Benchmarking LLMs on Implementing Novel Machine Learning Research Code
- Rethinking Fine-Tuning when Scaling Test-Time Compute: Limiting Confidence Improves Mathematical Reasoning
- Rethinking Verification for LLM Code Generation: From Generation to Testing
- Training Language Models to Generate Quality Code with Program Analysis Feedback
- Win Fast or Lose Slow: Balancing Speed and Accuracy in Latency-Sensitive Decisions of LLMs