In the Efficient-Code-Generation-with-E-Code work, the authors use a fine-tuned pre-trained model to predict a range of codes to be generated. Due to the need for comparative experiments, three code generation models are available. The three code generation models are E-code 350M, GPT-Neo 125M, and No expert group E-code 350M. We use each of the three fine-tuned code generation models to generate codes. Below we have the code generated by the three fine-tuned code generation models.

We give the results of 3 times code generation in the E-code 350M model.

We give the case results of one code generation for the GPT-Neo 125M model.

We give the case results of one code generation for the no expert group E-code 350M model.