Empirical comparison to gpt-llm-trainer

[neubig]

There is another nice library for training models from prompts, gpt-llm-trainer that came out simultaneously with prompt2model. There are some differences in terms of features:

Aspect	prompt2model	gpt-llm-trainer
Data Retrieval	Yes	No
Model Retrieval	Yes	No
Data Generation	Yes	Yes
Fine Tuning	Yes	Yes
Evaluation	Yes	No
Supported Models	All Hugging Face Models	gpt-3.5-turbo and LLaMa-7b

One remaining question is how prompt2model compares to gpt-llm-trainer with respect to accuracy and resulting model size. It would be nice to do an empirical comparison between the two.