Decision Transformer model trained on expert trajectories sampled from the Gym Walker2d environment

This is a trained Decision Transformer model trained from scratch on expert trajectories sampled from the Gym Walker2d environment based on the modified version of the example training script provided by HuggingFace

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.0001
  • train_batch_size: 64
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 120

Training results

Framework versions

  • Transformers 4.22.2
  • Pytorch 1.12.1+cu113
  • Datasets 2.5.1
  • Tokenizers 0.12.1
Downloads last month
10
Video Preview
loading