---
license: apache-2.0
datasets:
- huggingface-course/codeparrot-ds-train
- huggingface-course/codeparrot-ds-valid
language:
- en
metrics:
- code_eval
pipeline_tag: text-generation
tags:
- code
- gpt2
- pytorch
- causal-lm
---

# python-ds-accelerate (GPT-2 124M)

This model is a GPT-2 (124M parameter) causal language model trained from scratch specifically for **Python code completion** in Data Science contexts.

## Model Details

### Model Description

This model is an implementation of the GPT-2 architecture optimized for generating functional Python code snippets. It was trained using a custom training pipeline that incorporates a **keytoken weighted loss** function to prioritize important programming keywords (like `plt`, `pd`, `fit`, `predict`), making it more effective at suggesting Data Science-related code.

- **Developed by:** [Pranav Guhan R](https://github.com/PranavGuhanR)
- **Model type:** Transformer-based Causal Language Model
- **Language(s):** Python (English comments)
- **License:** Apache 2.0
- **Finetuned from model:** Trained from scratch

### Model Sources

- **Repository:** [GPT-2-124M-pretraining-for-code-completion](https://github.com/PranavGuhanR/GPT-2-124M-pretraining-for-code-completion)

## Uses

### Direct Use
The model is intended to be used for code completion tasks, specifically for completing Python scripts involving libraries like `pandas`, `matplotlib`, and `scikit-learn`.

### Out-of-Scope Use
The model is not suitable for general-purpose natural language conversation or generating code in languages other than Python.

## How to Get Started with the Model

You can use the model directly with a Hugging Face pipeline:

```python
from transformers import pipeline

pipe = pipeline("text-generation", model="PranavGuhan/python-ds-accelerate")

txt = """# create dataframe from x and y
df = pd.DataFrame({'x':x, 'y':y})
"""
print(pipe(txt, num_return_sequences=1)[0]["generated_text"])