metadata
license: mit
base_model: openai-community/gpt2
Null-GPT2
Description
This is a GPT2 Model, but only with the architecture, no pre-trained weights, biases, attention, etc.
This is useful for researchers who want to play with training the model (not tuning).
Generated via the github repo Model Architecture Generator
Use
First go into the directory of the model,
git clone https://github.com/ivanhe123/Model-Architecture-Generator
python -m randomnize_params -in "./NullGPT2" -out path_model_out
path_model_out is just the output path of the newly randomnized model.