michiyasunaga/DrRepair
[ICML 2020] DrRepair: Learning to Repair Programs from Error Messages
Leverages graph-based neural architectures with self-supervised learning on compiler error messages to repair C programs without requiring ground-truth fixes. The model encodes source code as abstract syntax graphs and uses diagnostic feedback as training signal, evaluated on DeepFix and SPoC datasets. Integrates with GCC/Clang for compilation, operates as a server endpoint for downstream repair tasks, and supports pretraining on large codebases like Codeforces.
197 stars. No commits in the last 6 months.
Stars
197
Forks
34
Language
Python
License
MIT
Category
Last pushed
May 24, 2021
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/michiyasunaga/DrRepair"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related frameworks
matlab-deep-learning/Quantized-Deep-Neural-Network-on-Jetson-AGX-Xavier
How to create, train and quantize network, then integrate it into pre/post image processing and...
bharathsudharsan/ML-Classifiers-on-MCUs
Supplementary material for IEEE Services Computing paper 'An SRAM Optimized Approach for...