config.json

{
  "act_dim": 2,
  "action_tanh": true,
  "activation_function": "relu",
  "adv_act_dim": 3,
  "architectures": [
    "TrainableDT"
  ],
  "attn_pdrop": 0.1,
  "bos_token_id": 50256,
  "context_size": 1,
  "discrete": true,
  "embd_pdrop": 0.1,
  "eos_token_id": 50256,
  "hidden_size": 128,
  "initializer_range": 0.02,
  "lambda1": 1.0,
  "lambda2": 1.0,
  "layer_norm_epsilon": 1e-05,
  "log_interval_steps": 100,
  "max_ep_len": 1,
  "max_ep_return": 60.0,
  "max_obs_len": 1,
  "max_obs_return": 30.0,
  "min_obs_return": 0.0,
  "model_type": "decision_transformer",
  "n_head": 1,
  "n_inner": null,
  "n_layer": 3,
  "n_positions": 1024,
  "pr_act_dim": 2,
  "reorder_and_upcast_attn": false,
  "resid_pdrop": 0.1,
  "returns_scale": 30,
  "scale_attn_by_inverse_layer_idx": false,
  "scale_attn_weights": true,
  "state_dim": 1,
  "state_mean": [
    1.0
  ],
  "state_std": [
    1e-08
  ],
  "torch_dtype": "float32",
  "transformers_version": "4.31.0",
  "use_cache": true,
  "vocab_size": 1,
  "warmup_steps": 10
}