Weyl09/mosim-humanoid-walk-tdmpc2-ar10-dt005

MoSim Humanoid Walk TD-MPC2 AR10, dt=0.005

Humanoid Walk state transition dataset collected using a TD-MPC2 policy. Rows are recorded every physics step at dt=0.005 seconds; the policy action is refreshed every 10 physics rows.

Format

Each .npz contains:

• —data: float32 array with shape (num_trajectories, 1000, state_dim + action_dim + state_dim).
• —metadata: object with state_dim, action_dim, and DMC qpos/qvel sizes when available.

Each transition row is [s_t, a_t, s_(t+1)].

For Humanoid Walk:

• —state_dim = 55
• —action_dim = 21
• —row dimension = 55 + 21 + 55 = 131
• —dataset row timestep dt = 0.005 seconds

Files

• —train_95pct/humanoid_walk_state_tdmpc2_humanoidwalk1_ar10_dt_train95pct_mosim.npz: 1.95 GiB
• —test_5pct/humanoid_walk_state_tdmpc2_humanoidwalk1_ar10_dt_test5pct_mosim.npz: 0.10 GiB

Dataset Details

{
  "checkpoint": "humanoid-walk-1.pt",
  "compile": false,
  "dataset_row_timestep": 0.005,
  "dt": 0.005,
  "effective_physics_action_repeat": 10,
  "eval_mode": false,
  "gpus": "redacted",
  "mpc": true,
  "n_sub_steps": 5,
  "policy_repeat_physics_steps": 10,
  "record_mode": "physics",
  "seed": 20260513,
  "tdmpc_control_repeat": 2,
  "test_npz": "humanoid_walk_state_tdmpc2_humanoidwalk1_ar10_dt_test5pct_mosim.npz",
  "test_trajectories": 500,
  "train_npz": "humanoid_walk_state_tdmpc2_humanoidwalk1_ar10_dt_train95pct_mosim.npz",
  "train_trajectories": 9500,
  "transition_steps": 1000
}