This model was made using the following parameters:
output_dir: data/activations/sorted
ds_repo: austindavis/lichess-uci-scored
ds_config: 202302-00000-00009
ds_split: train
model_checkpoint: austindavis/chessGPT2
ds_input_column: Transcript
ds_label_columns: ['Site', 'WhiteElo', 'BlackElo', 'Transcript', 'Scores']
max_shards_created: 480
batches_per_shard: 1
batch_size: 128
auto_find_batch_size: True
sort_ds_by_len: True
sort_ds_reversed: False
ds_shuffle_seed:… See the full description on the dataset page:
https://huggingface.co/datasets/austindavis/chessGPT2_hidden_states.
