On the Limitations of Embedding Based Methods for Measuring Functional Correctness for Code Generation

Open in new window