upload via upload_folder 2025-08-12T19:20:16.313075+00:00

README.md

@@ -0,0 +1,50 @@
+---
+env_name: LunarLander-v3
+tags:
+- LunarLander-v3
+- rainbow-dqn (with uniform sampling)
+- reinforcement-learning
+- custom-implementation
+- deep-q-learning
+- pytorch
+- rainbow
+- dqn
+model-index:
+- name: Rainbow-1d-LunarLander-v3-NoPer
+  results:
+  - task:
+      type: reinforcement-learning
+      name: reinforcement-learning
+    dataset:
+      name: LunarLander-v3
+      type: LunarLander-v3
+    metrics:
+    - type: mean_reward
+      value: 282.11 +/- 19.77
+      name: mean_reward
+      verified: false
+---
+
+# **Rainbow-DQN (with uniform sampling)** Agent playing **LunarLander-v3**
+This is a trained model of a **Rainbow-DQN (with uniform sampling)** agent playing **LunarLander-v3**.
+
+## Usage
+### create the conda env in https://github.com/GeneHit/drl_practice
+```bash
+conda create -n drl python=3.12
+conda activate drl
+python -m pip install -r requirements.txt
+```
+
+### play with full model
+```python
+# load the full model
+model = load_from_hub(repo_id="winkin119/Rainbow-1d-LunarLander-v3-NoPer", filename="full_model.pt")
+
+# Create the environment. 
+env = gym.make("LunarLander-v3")
+state, _ = env.reset()
+action = model.action(state)
+...
+```
+There is also a state dict version of the model, you can check the corresponding definition in the repo.

eval_result.json

@@ -0,0 +1,6 @@
+{
+    "mean_reward": 282.1136285308617,
+    "std_reward": 19.771440248266718,
+    "datetime": "2025-08-10T20:31:50.282987+00:00",
+    "train_duration_min": "6.54"
+}

full_model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46f828bd08bf9f5fe0b8143c8f1f65ca2138a0c592b6e97097648818a8422a78
+size 807925

params.json

@@ -0,0 +1,49 @@
+{
+    "env_config": {
+        "env_id": "LunarLander-v3",
+        "env_kwargs": {},
+        "max_steps": null,
+        "normalize_obs": false,
+        "use_image": false,
+        "vector_env_num": 6,
+        "use_multi_processing": true,
+        "image_shape": null,
+        "frame_stack": 1,
+        "frame_skip": 1,
+        "training_render_mode": null
+    },
+    "device": "cpu",
+    "learning_rate": 0.0003,
+    "gamma": 0.99,
+    "checkpoint_pathname": "",
+    "max_grad_norm": 10.0,
+    "log_interval": 100,
+    "eval_episodes": 100,
+    "eval_random_seed": 42,
+    "eval_video_num": 10,
+    "timesteps": 125000,
+    "epsilon_schedule": {
+        "_type": "ConstantSchedule",
+        "_module": "practice.utils_for_coding.scheduler_utils",
+        "value": 0.0
+    },
+    "replay_buffer_capacity": 0,
+    "batch_size": 64,
+    "train_interval": 1,
+    "target_update_interval": 250,
+    "update_start_step": 2000,
+    "dqn_algorithm": "rainbow",
+    "noisy_std": 0.5,
+    "per_buffer_config": {
+        "capacity": 150000,
+        "n_step": 3,
+        "gamma": 0.99,
+        "use_uniform_sampling": true,
+        "alpha": 0.6,
+        "beta": 0.4,
+        "beta_increment": 3e-06
+    },
+    "v_min": -300.0,
+    "v_max": 300.0,
+    "num_atoms": 51
+}

state_dict.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42b574f556a32b3316dde32946689497381e63c8fdfb85685dac0bc92ac65acc
+size 806005

tensorboard/events.out.tfevents.1754857508.winkindeMacBook-Air.local.34042.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92b91b010a452b2e77f1a42407f0a846697e5787d3811145778cb08159a01e77
+size 568536