Open klei22 opened 1 month ago
Dataset preparation:
cd data/aozorabunko_clean
bash get_dataset.sh
- obtains dataset (requires jq
command)bash process.sh
- takes a whilepython3 prepare.py -t input.txt --method char
Manual embedding table creation:
python3 mapping.py
maps from the table to an .npy file. train.py
and model.py
setup:
model.py
uses this file to set the word table and lm head if set up in the GPT class constructortrain.py
's --n_embd_main
needs to match the embedding dimension of the npy matrix
This is a draft for experimenting with new variations on embedding tables.