# Jam-sojm
---
license: bigscience-openrail-m
datasets:
- apcl/so13m
---

Jam-sojm is a GPT2-like model for research in fine-grained Java analysis. It is intended for fine-grained analysis of Java source code at the level of methods, statements, and variables, as a foundation for downstream tasks like code completion, comment generation, and automated bug repair. 


## Datasets: [jm52m dataset](https://huggingface.co/datasets/apcl/jm52m) and [so13m dataset](https://huggingface.co/datasets/apcl/so13m)
## Epochs: Two ( one with each dataset, with the the learning rate and decay reset in between)
## Iterations : ~600,000