README.md

metadata

license: apache-2.0
base_model: Salesforce/codet5p-220m
pipeline_tag: text-generation
tags:
  - codet5
  - pytest
  - unit-testing
  - code-generation
  - nlp
  - edge-cases
  - generated_from_trainer
metrics:
  - eval_loss
model-index:
  - name: codet5p-220m-pytest-edge-generator
    results:
      - task:
          type: text-generation
          name: PyTest edge-case unit test generation
        dataset:
          name: Custom edge-case function–test pairs
          type: jsonl
        metrics:
          - type: eval_loss
            value: 1.7791306972503662
            name: Best eval loss

CodeT5p-220m PyTest Edge-Case Generator

This model is a fine-tuned version of Salesforce/codet5p-220m, starting from my normal-test fine-tuned checkpoint, for generating PyTest unit tests focused on edge cases and boundary conditions.

Overview

It prioritizes:

• empty inputs
• None handling
• extreme values
• invalid types
• exceptions (pytest.raises)
• boundary conditions

Training results

Eval loss by epoch:

Epoch	Eval Loss
1	1.8182
2	1.7890
3	1.7791
4	1.7804

• Best eval_loss: 1.7791306972503662

Citation

@misc{akinwale2025codet5pytestedge,
  author  = {Akinwale, Oluwaferanmi Philip},
  title   = {CodeT5p-220m PyTest Edge-Case Generator},
  year    = {2025},
  note    = {Hugging Face model repository},
  url     = {https://huggingface.co/Oluwaferanmiii/codet5p-220m-pytest-edge-generator}
}