sf_finetuning_forgetting_human_monk

A(n) APPO model trained on the challenge environment.

This model was trained using Sample-Factory 2.0: https://github.com/BartekCupial/sample-factory/tree/nethack. Documentation for how to use Sample-Factory can be found at https://www.samplefactory.dev/

Downloading the model

After installing Sample-Factory, download the model with:

python -m sample_factory.huggingface.load_from_hub -r LLParallax/sf_finetuning_forgetting_human_monk

Using the model

To run the model after download, use the enjoy script corresponding to this environment:

python -m sf_examples.nethack.enjoy_nethack \
  --env=challenge \
  --character=mon-hum-neu-mal \
  --train_dir=./train_dir \
  --experiment=sf_finetuning_forgetting_human_monk \
  --use_pretrained_checkpoint=False \
  --teacher_path=./train_dir/sf_finetuning_forgetting_human_monk

For performance evaluation, use the eval script:

python -m sf_examples.nethack.eval_nethack \
  --env=challenge \
  --character=mon-hum-neu-mal \
  --sample_env_episodes=128 \
  --num_workers=16 \
  --num_envs_per_worker=32 \
  --worker_num_splits=2 \
  --train_dir=./train_dir \
  --experiment=sf_finetuning_forgetting_human_monk \
  --use_pretrained_checkpoint=False \
  --teacher_path=./train_dir/sf_finetuning_forgetting_human_monk

You can also upload models to the Hugging Face Hub using the same script with the --push_to_hub flag. See https://www.samplefactory.dev/10-huggingface/huggingface/ for more details