Instructions to use tsilva/Level1-2_stable-baselines3-ppo_91df110f with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- stable-baselines3
How to use tsilva/Level1-2_stable-baselines3-ppo_91df110f with stable-baselines3:
from huggingface_sb3 import load_from_hub checkpoint = load_from_hub( repo_id="tsilva/Level1-2_stable-baselines3-ppo_91df110f", filename="{MODEL FILENAME}.zip", ) - Notebooks
- Google Colab
- Kaggle
Release Level1-2 PPO checkpoint
Browse files- .gitattributes +1 -0
- README.md +82 -0
- model.zip +3 -0
- model_metadata.json +328 -0
- release_manifest.json +0 -0
- replay.mp4 +3 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
replay.mp4 filter=lfs diff=lfs merge=lfs -text
|
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: stable-baselines3
|
| 3 |
+
pipeline_tag: reinforcement-learning
|
| 4 |
+
tags:
|
| 5 |
+
- reinforcement-learning
|
| 6 |
+
- stable-baselines3
|
| 7 |
+
- ppo
|
| 8 |
+
- stable-retro
|
| 9 |
+
- rlab
|
| 10 |
+
- super-mario-bros
|
| 11 |
+
- nes
|
| 12 |
+
- SuperMarioBros-Nes-v0
|
| 13 |
+
metrics:
|
| 14 |
+
- completion-rate
|
| 15 |
+
---
|
| 16 |
+
|
| 17 |
+
# SuperMarioBros-NES Level1-2 PPO
|
| 18 |
+
|
| 19 |
+
PPO policy checkpoint for completing `SuperMarioBros-Nes-v0` `Level1-2` with `supermariobrosnes-turbo`, trained with [`rlab`](https://github.com/tsilva/rlab).
|
| 20 |
+
|
| 21 |
+
## Quick Start
|
| 22 |
+
|
| 23 |
+
Install `rlab` once, import the ROM, then play or evaluate this checkpoint directly from Hugging Face:
|
| 24 |
+
|
| 25 |
+
```bash
|
| 26 |
+
uv tool install --from git+https://github.com/tsilva/rlab rlab
|
| 27 |
+
rlab import-roms ~/roms --game SuperMarioBros-Nes-v0
|
| 28 |
+
rlab play hf://tsilva/SuperMarioBros-Nes-v0_Level1-2
|
| 29 |
+
rlab eval hf://tsilva/SuperMarioBros-Nes-v0_Level1-2
|
| 30 |
+
```
|
| 31 |
+
|
| 32 |
+
## Evaluation Results
|
| 33 |
+
|
| 34 |
+
| `eval_profile` | `episodes` | `seed_start` | `completion_rate` | `max_x_max` | `reward_mean` | `checkpoint_step` |
|
| 35 |
+
|---|---:|---:|---:|---:|---:|---:|
|
| 36 |
+
| `local release eval`, deterministic policy | 100 | 10000 | `100/100` episodes | 3129 | 3148.550 | 7500000 |
|
| 37 |
+
|
| 38 |
+
This release uses the `step-7500000` checkpoint from the W&B run below. The training run reached the Level1-2 early-stop gate at `7,506,208` timesteps; this staged release eval was run locally because the run's post-training checkpoint eval crashed after its first checkpoint during W&B summary bookkeeping.
|
| 39 |
+
|
| 40 |
+
## Environment Details
|
| 41 |
+
|
| 42 |
+
| Setting | Value |
|
| 43 |
+
|---|---|
|
| 44 |
+
| `env_provider` | `supermariobrosnes-turbo` |
|
| 45 |
+
| `env_id` | `supermariobrosnes-turbo:SuperMarioBros-Nes-v0` |
|
| 46 |
+
| `game` | `SuperMarioBros-Nes-v0` |
|
| 47 |
+
| `state` | `Level1-2` |
|
| 48 |
+
| `preprocessing` | crop top `32` px, grayscale, resize to `84 x 84` |
|
| 49 |
+
| `frame_stack` | `4` |
|
| 50 |
+
| `frame_skip` | `4` |
|
| 51 |
+
| `max_pool_frames` | disabled |
|
| 52 |
+
| `policy_observation_layout` | channel-first `(4, 84, 84)` |
|
| 53 |
+
| `action_set` | `simple` |
|
| 54 |
+
| `reward_mode` | `score` |
|
| 55 |
+
| `reward_shaping` | score-progress reward, `terminal_reward=50`, `death_penalty=25`, `reward_scale=10` |
|
| 56 |
+
| `max_episode_steps` | `4500` |
|
| 57 |
+
| `train_done_on_events` | life loss, level change |
|
| 58 |
+
| `eval_done_on_events` | level change |
|
| 59 |
+
|
| 60 |
+
## Training Recipe
|
| 61 |
+
|
| 62 |
+
| Setting | Value |
|
| 63 |
+
|---|---:|
|
| 64 |
+
| `goal_id` | `Level1-2` |
|
| 65 |
+
| `recipe_id` | `base` |
|
| 66 |
+
| `checkpoint_step` | 7500000 |
|
| 67 |
+
| `frame_skip` | 4 |
|
| 68 |
+
| `max_episode_steps` | 4500 |
|
| 69 |
+
| `reward_mode` | `score` |
|
| 70 |
+
| `reward_shaping` | score-progress reward, `terminal_reward=50`, `death_penalty=25`, `reward_scale=10` |
|
| 71 |
+
| `done_on_events` | life loss, level change |
|
| 72 |
+
|
| 73 |
+
## Provenance
|
| 74 |
+
|
| 75 |
+
| Item | Value |
|
| 76 |
+
|---|---|
|
| 77 |
+
| `source_project` | [`rlab`](https://github.com/tsilva/rlab) |
|
| 78 |
+
| `goal_id` | `Level1-2` |
|
| 79 |
+
| `goal_title` | `Sample-efficient SuperMarioBros-NES Level1-2 completion` |
|
| 80 |
+
| `wandb_run` | [`Level1-2_base_s1_20260704T072752Z`](https://wandb.ai/tsilva/SuperMarioBros-Nes-v0/runs/dtpvaqq2) |
|
| 81 |
+
| `wandb_artifact` | `tsilva/SuperMarioBros-Nes-v0/Level1-2_base_s1_20260704T072752Z-checkpoint:step-7500000` |
|
| 82 |
+
| `eval_source` | `local release eval` |
|
model.zip
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:031cad88ae2a0a12064e1459601bfa51fbe5e7aa69e0c39aefbd1ba0f4cfd483
|
| 3 |
+
size 21054227
|
model_metadata.json
ADDED
|
@@ -0,0 +1,328 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"artifact_storage_uri": "s3://wandb/rlab/SuperMarioBros-Nes-v0/Level1-2_base_s1_20260704T072752Z/checkpoint/ppo_supermariobros-nes-v0_7500000_steps.zip",
|
| 3 |
+
"checkpoint_step": 7500000,
|
| 4 |
+
"env_config": {
|
| 5 |
+
"action_set": "simple",
|
| 6 |
+
"clip_rewards": false,
|
| 7 |
+
"completion_reward": 0.0,
|
| 8 |
+
"death_penalty": 25.0,
|
| 9 |
+
"done_on_events": [
|
| 10 |
+
"life_loss",
|
| 11 |
+
"level_change"
|
| 12 |
+
],
|
| 13 |
+
"env_provider": "supermariobrosnes-turbo",
|
| 14 |
+
"env_threads": 4,
|
| 15 |
+
"env_wrappers": [
|
| 16 |
+
{
|
| 17 |
+
"id": "SuperMarioBrosNesProgressInfoWrapper",
|
| 18 |
+
"kwargs": {}
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"id": "SuperMarioBrosNesRewardEnvWrapper",
|
| 22 |
+
"kwargs": {
|
| 23 |
+
"completion_reward": 0.0,
|
| 24 |
+
"death_penalty": 25.0,
|
| 25 |
+
"progress_reward_scale": 1.0,
|
| 26 |
+
"reward_mode": "score",
|
| 27 |
+
"reward_scale": 10.0,
|
| 28 |
+
"score_progress_clipped": false,
|
| 29 |
+
"terminal_reward": 50.0
|
| 30 |
+
}
|
| 31 |
+
}
|
| 32 |
+
],
|
| 33 |
+
"frame_skip": 4,
|
| 34 |
+
"game": "SuperMarioBros-Nes-v0",
|
| 35 |
+
"hud_crop_top": -1,
|
| 36 |
+
"info_events": {},
|
| 37 |
+
"max_episode_steps": 4500,
|
| 38 |
+
"max_pool_frames": false,
|
| 39 |
+
"no_progress_min_delta": 0,
|
| 40 |
+
"no_progress_timeout_steps": 0,
|
| 41 |
+
"obs_crop": [
|
| 42 |
+
32,
|
| 43 |
+
0,
|
| 44 |
+
0,
|
| 45 |
+
0
|
| 46 |
+
],
|
| 47 |
+
"obs_resize_algorithm": "area",
|
| 48 |
+
"observation_size": 84,
|
| 49 |
+
"progress_reward_cap": 30.0,
|
| 50 |
+
"progress_reward_scale": 1.0,
|
| 51 |
+
"reward_mode": "score",
|
| 52 |
+
"reward_scale": 10.0,
|
| 53 |
+
"score_progress_clipped": false,
|
| 54 |
+
"state": "Level1-2",
|
| 55 |
+
"state_distribution": [],
|
| 56 |
+
"state_probs": [],
|
| 57 |
+
"state_sampling_mode": "single",
|
| 58 |
+
"states": [],
|
| 59 |
+
"sticky_action_prob": 0.0,
|
| 60 |
+
"task_conditioning": false,
|
| 61 |
+
"task_conditioning_info_values": [],
|
| 62 |
+
"task_conditioning_info_vars": [],
|
| 63 |
+
"terminal_reward": 50.0,
|
| 64 |
+
"time_penalty": 0.0,
|
| 65 |
+
"use_retro_reward": false
|
| 66 |
+
},
|
| 67 |
+
"environment": {
|
| 68 |
+
"action": {
|
| 69 |
+
"action_set": "simple"
|
| 70 |
+
},
|
| 71 |
+
"env_id": "supermariobrosnes-turbo:SuperMarioBros-Nes-v0",
|
| 72 |
+
"preprocessing": {
|
| 73 |
+
"frame_skip": 4,
|
| 74 |
+
"frame_stack": 4,
|
| 75 |
+
"max_pool_frames": false,
|
| 76 |
+
"maxpool_last_two": false,
|
| 77 |
+
"obs_copy": "safe_view",
|
| 78 |
+
"obs_crop": [
|
| 79 |
+
32,
|
| 80 |
+
0,
|
| 81 |
+
0,
|
| 82 |
+
0
|
| 83 |
+
],
|
| 84 |
+
"obs_grayscale": true,
|
| 85 |
+
"obs_resize": [
|
| 86 |
+
84,
|
| 87 |
+
84
|
| 88 |
+
],
|
| 89 |
+
"obs_resize_algorithm": "area",
|
| 90 |
+
"pipeline": "supermariobrosnes_turbo_native_vec_env",
|
| 91 |
+
"policy_observation_layout": "channel_first",
|
| 92 |
+
"sticky_action_prob": 0.0
|
| 93 |
+
},
|
| 94 |
+
"reward": {
|
| 95 |
+
"clip_rewards": false,
|
| 96 |
+
"completion_reward": 0.0,
|
| 97 |
+
"death_penalty": 25.0,
|
| 98 |
+
"env_wrappers": [
|
| 99 |
+
{
|
| 100 |
+
"id": "SuperMarioBrosNesProgressInfoWrapper",
|
| 101 |
+
"kwargs": {}
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"id": "SuperMarioBrosNesRewardEnvWrapper",
|
| 105 |
+
"kwargs": {
|
| 106 |
+
"completion_reward": 0.0,
|
| 107 |
+
"death_penalty": 25.0,
|
| 108 |
+
"progress_reward_scale": 1.0,
|
| 109 |
+
"reward_mode": "score",
|
| 110 |
+
"reward_scale": 10.0,
|
| 111 |
+
"score_progress_clipped": false,
|
| 112 |
+
"terminal_reward": 50.0
|
| 113 |
+
}
|
| 114 |
+
}
|
| 115 |
+
],
|
| 116 |
+
"progress_reward_cap": 30.0,
|
| 117 |
+
"progress_reward_scale": 1.0,
|
| 118 |
+
"reward_mode": "score",
|
| 119 |
+
"reward_scale": 10.0,
|
| 120 |
+
"score_progress_clipped": false,
|
| 121 |
+
"terminal_reward": 50.0,
|
| 122 |
+
"time_penalty": 0.0,
|
| 123 |
+
"use_retro_reward": false
|
| 124 |
+
},
|
| 125 |
+
"schema_version": 1,
|
| 126 |
+
"state": "Level1-2",
|
| 127 |
+
"state_probs": [],
|
| 128 |
+
"states": [],
|
| 129 |
+
"task_conditioning": {
|
| 130 |
+
"task_conditioning": false,
|
| 131 |
+
"task_conditioning_info_values": [],
|
| 132 |
+
"task_conditioning_info_vars": []
|
| 133 |
+
},
|
| 134 |
+
"termination": {
|
| 135 |
+
"done_on_events": [
|
| 136 |
+
"life_loss",
|
| 137 |
+
"level_change"
|
| 138 |
+
],
|
| 139 |
+
"info_events": {},
|
| 140 |
+
"max_episode_steps": 4500,
|
| 141 |
+
"no_progress_min_delta": 0,
|
| 142 |
+
"no_progress_timeout_steps": 0
|
| 143 |
+
}
|
| 144 |
+
},
|
| 145 |
+
"environment_hash": "sha256:24189df154a9b4e09521a674eb4147577412ef49d6652ec4683b523ed33a93a2",
|
| 146 |
+
"filename": "model.zip",
|
| 147 |
+
"goal_slug": "Level1-2",
|
| 148 |
+
"kind": "checkpoint",
|
| 149 |
+
"metadata_version": 2,
|
| 150 |
+
"queue_train_job_id": 4,
|
| 151 |
+
"recipe_path": "experiments/goals/SuperMarioBros-Nes-v0/Level1-2/recipes/base.yaml",
|
| 152 |
+
"recipe_slug": "base",
|
| 153 |
+
"run_description": "Seed 1 runs the Level1-2 transfer of the successful Level1-1 B55 low-KL late-decay recipe, created to test whether that policy recipe transfers without changing PPO or reward hyperparameters while overriding only Level1-2 identity and reporting metadata.",
|
| 154 |
+
"run_name": "Level1-2_base_s1_20260704T072752Z",
|
| 155 |
+
"training_metadata": {
|
| 156 |
+
"env_config": {
|
| 157 |
+
"action_set": "simple",
|
| 158 |
+
"clip_rewards": false,
|
| 159 |
+
"completion_reward": 0.0,
|
| 160 |
+
"death_penalty": 25.0,
|
| 161 |
+
"done_on_events": [
|
| 162 |
+
"life_loss",
|
| 163 |
+
"level_change"
|
| 164 |
+
],
|
| 165 |
+
"env_provider": "supermariobrosnes-turbo",
|
| 166 |
+
"env_threads": 4,
|
| 167 |
+
"env_wrappers": [
|
| 168 |
+
{
|
| 169 |
+
"id": "SuperMarioBrosNesProgressInfoWrapper",
|
| 170 |
+
"kwargs": {}
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"id": "SuperMarioBrosNesRewardEnvWrapper",
|
| 174 |
+
"kwargs": {
|
| 175 |
+
"completion_reward": 0.0,
|
| 176 |
+
"death_penalty": 25.0,
|
| 177 |
+
"progress_reward_scale": 1.0,
|
| 178 |
+
"reward_mode": "score",
|
| 179 |
+
"reward_scale": 10.0,
|
| 180 |
+
"score_progress_clipped": false,
|
| 181 |
+
"terminal_reward": 50.0
|
| 182 |
+
}
|
| 183 |
+
}
|
| 184 |
+
],
|
| 185 |
+
"frame_skip": 4,
|
| 186 |
+
"game": "SuperMarioBros-Nes-v0",
|
| 187 |
+
"hud_crop_top": -1,
|
| 188 |
+
"info_events": {},
|
| 189 |
+
"max_episode_steps": 4500,
|
| 190 |
+
"max_pool_frames": false,
|
| 191 |
+
"no_progress_min_delta": 0,
|
| 192 |
+
"no_progress_timeout_steps": 0,
|
| 193 |
+
"obs_crop": [
|
| 194 |
+
32,
|
| 195 |
+
0,
|
| 196 |
+
0,
|
| 197 |
+
0
|
| 198 |
+
],
|
| 199 |
+
"obs_resize_algorithm": "area",
|
| 200 |
+
"observation_size": 84,
|
| 201 |
+
"progress_reward_cap": 30.0,
|
| 202 |
+
"progress_reward_scale": 1.0,
|
| 203 |
+
"reward_mode": "score",
|
| 204 |
+
"reward_scale": 10.0,
|
| 205 |
+
"score_progress_clipped": false,
|
| 206 |
+
"state": "Level1-2",
|
| 207 |
+
"state_distribution": [],
|
| 208 |
+
"state_probs": [],
|
| 209 |
+
"state_sampling_mode": "single",
|
| 210 |
+
"states": [],
|
| 211 |
+
"sticky_action_prob": 0.0,
|
| 212 |
+
"task_conditioning": false,
|
| 213 |
+
"task_conditioning_info_values": [],
|
| 214 |
+
"task_conditioning_info_vars": [],
|
| 215 |
+
"terminal_reward": 50.0,
|
| 216 |
+
"time_penalty": 0.0,
|
| 217 |
+
"use_retro_reward": false
|
| 218 |
+
},
|
| 219 |
+
"environment": {
|
| 220 |
+
"action": {
|
| 221 |
+
"action_set": "simple"
|
| 222 |
+
},
|
| 223 |
+
"env_id": "supermariobrosnes-turbo:SuperMarioBros-Nes-v0",
|
| 224 |
+
"preprocessing": {
|
| 225 |
+
"frame_skip": 4,
|
| 226 |
+
"frame_stack": 4,
|
| 227 |
+
"max_pool_frames": false,
|
| 228 |
+
"maxpool_last_two": false,
|
| 229 |
+
"obs_copy": "safe_view",
|
| 230 |
+
"obs_crop": [
|
| 231 |
+
32,
|
| 232 |
+
0,
|
| 233 |
+
0,
|
| 234 |
+
0
|
| 235 |
+
],
|
| 236 |
+
"obs_grayscale": true,
|
| 237 |
+
"obs_resize": [
|
| 238 |
+
84,
|
| 239 |
+
84
|
| 240 |
+
],
|
| 241 |
+
"obs_resize_algorithm": "area",
|
| 242 |
+
"pipeline": "supermariobrosnes_turbo_native_vec_env",
|
| 243 |
+
"policy_observation_layout": "channel_first",
|
| 244 |
+
"sticky_action_prob": 0.0
|
| 245 |
+
},
|
| 246 |
+
"reward": {
|
| 247 |
+
"clip_rewards": false,
|
| 248 |
+
"completion_reward": 0.0,
|
| 249 |
+
"death_penalty": 25.0,
|
| 250 |
+
"env_wrappers": [
|
| 251 |
+
{
|
| 252 |
+
"id": "SuperMarioBrosNesProgressInfoWrapper",
|
| 253 |
+
"kwargs": {}
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"id": "SuperMarioBrosNesRewardEnvWrapper",
|
| 257 |
+
"kwargs": {
|
| 258 |
+
"completion_reward": 0.0,
|
| 259 |
+
"death_penalty": 25.0,
|
| 260 |
+
"progress_reward_scale": 1.0,
|
| 261 |
+
"reward_mode": "score",
|
| 262 |
+
"reward_scale": 10.0,
|
| 263 |
+
"score_progress_clipped": false,
|
| 264 |
+
"terminal_reward": 50.0
|
| 265 |
+
}
|
| 266 |
+
}
|
| 267 |
+
],
|
| 268 |
+
"progress_reward_cap": 30.0,
|
| 269 |
+
"progress_reward_scale": 1.0,
|
| 270 |
+
"reward_mode": "score",
|
| 271 |
+
"reward_scale": 10.0,
|
| 272 |
+
"score_progress_clipped": false,
|
| 273 |
+
"terminal_reward": 50.0,
|
| 274 |
+
"time_penalty": 0.0,
|
| 275 |
+
"use_retro_reward": false
|
| 276 |
+
},
|
| 277 |
+
"schema_version": 1,
|
| 278 |
+
"state": "Level1-2",
|
| 279 |
+
"state_probs": [],
|
| 280 |
+
"states": [],
|
| 281 |
+
"task_conditioning": {
|
| 282 |
+
"task_conditioning": false,
|
| 283 |
+
"task_conditioning_info_values": [],
|
| 284 |
+
"task_conditioning_info_vars": []
|
| 285 |
+
},
|
| 286 |
+
"termination": {
|
| 287 |
+
"done_on_events": [
|
| 288 |
+
"life_loss",
|
| 289 |
+
"level_change"
|
| 290 |
+
],
|
| 291 |
+
"info_events": {},
|
| 292 |
+
"max_episode_steps": 4500,
|
| 293 |
+
"no_progress_min_delta": 0,
|
| 294 |
+
"no_progress_timeout_steps": 0
|
| 295 |
+
}
|
| 296 |
+
},
|
| 297 |
+
"environment_hash": "sha256:24189df154a9b4e09521a674eb4147577412ef49d6652ec4683b523ed33a93a2",
|
| 298 |
+
"preprocessing": {
|
| 299 |
+
"frame_skip": 4,
|
| 300 |
+
"frame_stack": 4,
|
| 301 |
+
"maxpool_last_two": false,
|
| 302 |
+
"obs_copy": "safe_view",
|
| 303 |
+
"obs_crop": [
|
| 304 |
+
32,
|
| 305 |
+
0,
|
| 306 |
+
0,
|
| 307 |
+
0
|
| 308 |
+
],
|
| 309 |
+
"obs_grayscale": true,
|
| 310 |
+
"obs_resize": [
|
| 311 |
+
84,
|
| 312 |
+
84
|
| 313 |
+
],
|
| 314 |
+
"obs_resize_algorithm": "area",
|
| 315 |
+
"pipeline": "supermariobrosnes_turbo_native_vec_env",
|
| 316 |
+
"policy_observation_layout": "channel_first",
|
| 317 |
+
"sticky_action_prob": 0.0
|
| 318 |
+
},
|
| 319 |
+
"versions": {
|
| 320 |
+
"stable_baselines3": "2.8.0",
|
| 321 |
+
"stable_retro_turbo": "1.0.1.post4",
|
| 322 |
+
"supermariobrosnes_turbo": "0.2.4"
|
| 323 |
+
}
|
| 324 |
+
},
|
| 325 |
+
"training_metadata_hash": "5f907ca4343b943bca056c4cce6603023449a5b2dec1dbf4683fcd1591bdf526",
|
| 326 |
+
"wandb_run_id": "dtpvaqq2",
|
| 327 |
+
"wandb_run_path": "tsilva/SuperMarioBros-Nes-v0/dtpvaqq2"
|
| 328 |
+
}
|
release_manifest.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
replay.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:94b3184669e8ef9297ec298c5136d3340f6fda762664cc4cd1dcf8e7fba42d0b
|
| 3 |
+
size 628219
|