Pip compatible CodeBLEU metric implementation available for linux/macos/win
-
Updated
Mar 31, 2025 - Python
Pip compatible CodeBLEU metric implementation available for linux/macos/win
It uses Generative AI to create code from natural description to python code.
Multi-agent ReAct framework for CUDA code generation — extended CodeBLEU + unbiased Pass@1 across 25 GPU-kernel tasks from 13 real repos, 3 LLMs via NVIDIA NIM
Graph-Augmented Code Intelligence RAG engine pairing deterministic AST call graphs with dense vector search to boost direct-callee Recall@8 from 0.0% to 53.3% (+53.3 pp) under a 5.8ms p95 latency bound. Powered by a QLoRA fine-tuned Qwen2.5 adapter (70.02 CodeBLEU), quality verification harness, and production Model Context Protocol (MCP) server.
To associate your repository with the codebleu topic, visit your repo's landing page and select "manage topics."