Upload with huggingface_hub

README.md

@@ -0,0 +1,32 @@
+---
+library_name: mbrl-lib
+tags:
+- mbrl-Hopper-v2
+- deep-reinforcement-learning
+- reinforcement-learning
+- mbrl-lib
+model-index:
+- name: OneDTransitionRewardModel w/ SACAgent
+  results:
+  - task:
+      type: reinforcement-learning
+      name: reinforcement-learning
+    dataset:
+      name: mbrl-Hopper-v2
+      type: mbrl-Hopper-v2
+    metrics:
+    - type: mean_reward
+      value: 87.70 +/- 1.10
+      name: mean_reward
+      verified: false
+---
+# **OneDTransitionRewardModel w/ SACAgent** Agent playing **mbrl-Hopper-v2**
+This is a trained model of a **OneDTransitionRewardModel w/ SACAgent** agent playing **mbrl-Hopper-v2**
+using [MBRL-Lib](https://github.com/facebookresearch/mbrl-lib).
+
+## Usage (with MBRL-Lib)
+TODO: Add your code
+```python
+from mbrl import ...
+...
+```

config.yaml

@@ -0,0 +1,106 @@
+action_optimizer:
+  _target_: mbrl.planning.CEMOptimizer
+  alpha: 0.1
+  clipped_normal: false
+  device: cpu
+  elite_ratio: 0.1
+  lower_bound: ???
+  num_iterations: 5
+  population_size: 350
+  return_mean_elites: true
+  upper_bound: ???
+algorithm:
+  agent:
+    _target_: mbrl.third_party.pytorch_sac_pranz24.sac.SAC
+    action_space:
+      _target_: gym.env.Box
+      high:
+      - 1.0
+      - 1.0
+      - 1.0
+      low:
+      - -1.0
+      - -1.0
+      - -1.0
+      shape:
+      - 3
+    args:
+      alpha: 0.2
+      automatic_entropy_tuning: false
+      device: cpu
+      gamma: 0.99
+      hidden_size: 512
+      lr: 0.0003
+      policy: Gaussian
+      target_entropy: 1
+      target_update_interval: 4
+      tau: 0.005
+    num_inputs: 11
+  freq_train_model: 250
+  initial_exploration_steps: 5000
+  learned_rewards: true
+  name: mbpo
+  normalize: true
+  normalize_double_precision: true
+  num_eval_episodes: 1
+  random_initial_explore: false
+  real_data_ratio: 0.0
+  sac_samples_action: true
+  target_is_delta: true
+debug_mode: false
+device: cpu
+dynamics_model:
+  _target_: mbrl.models.GaussianMLP
+  activation_fn_cfg:
+    _target_: torch.nn.SiLU
+  deterministic: false
+  device: cpu
+  ensemble_size: 7
+  hid_size: 200
+  in_size: 14
+  learn_logvar_bounds: false
+  num_layers: 4
+  out_size: 12
+  propagation_method: random_model
+experiment: default
+log_frequency_agent: 1000
+overrides:
+  cem_alpha: 0.1
+  cem_clipped_normal: false
+  cem_elite_ratio: 0.1
+  cem_num_iters: 5
+  cem_population_size: 350
+  effective_model_rollouts_per_step: 400
+  env: gym___Hopper-v2
+  epoch_length: 1000
+  freq_train_model: 250
+  model_batch_size: 256
+  model_lr: 0.001
+  model_wd: 1.0e-05
+  num_elites: 5
+  num_epochs_to_retain_sac_buffer: 1
+  num_sac_updates_per_step: 40
+  num_steps: 125000
+  patience: 5
+  planning_horizon: 15
+  rollout_schedule:
+  - 20
+  - 150
+  - 1
+  - 15
+  sac_alpha: 0.2
+  sac_automatic_entropy_tuning: false
+  sac_batch_size: 256
+  sac_gamma: 0.99
+  sac_hidden_size: 512
+  sac_lr: 0.0003
+  sac_policy: Gaussian
+  sac_target_entropy: 1
+  sac_target_update_interval: 4
+  sac_tau: 0.005
+  sac_updates_every_steps: 1
+  term_fn: hopper
+  validation_ratio: 0.2
+root_dir: ./logs
+save_video: false
+seed: 0

model/env_stats.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e21598388c4adb0e0782789953c23e9e74ca9e136f19578b67ef7413b699755
+size 422

model/model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfad6ff5aa75a76004692d8b49f1027d8dafb6cc43b248b6c7f48e3c6189925b
+size 3599717