WALL-E mining a diamond in Mars. Step 1-2: the agent makes decision via MPC with the initial unaligned world model, resulting in a failed action for mining iron. Step 3: leveraging previous trajectories and world model predictions, WALL-E learns symbolic knowledge, including rules, knowledge graphs, and scene graphs. Step 4-5: the learned symbolic knowledge helps the world model make accurate predictions and correct the previous mistake. Step 6-7: the agent adjusts its decision accordingly and replaces stone pickaxe with iron pickaxe toward completing the task.

Table of Contents

• Abstract
• Main Results
• Rule Learning Demo
• ALFWorld
• Mars
• Citation

Abstract:

Can we build accurate world models out of large language models (LLMs)? How can world models benefit LLM agents? The gap between the prior knowledge of LLMs and the specified environment's dynamics usually bottlenecks LLMs' performance as world models. To bridge the gap, we propose a training-free "world alignment" that learns an environment's symbolic knowledge complementary to LLMs. The symbolic knowledge covers action rules, knowledge graphs, and scene graphs, which are extracted by LLMs from exploration trajectories and encoded into executable codes to regulate LLM agents' policies. We further propose an RL-free, model-based agent "WALL-E" through the model-predictive control (MPC) framework. Unlike classical MPC requiring costly optimization on the fly, we adopt an LLM agent as an efficient look-ahead optimizer of future steps' actions by interacting with the neurosymbolic world model. While the LLM agent's strong heuristics make it an efficient planner in MPC, the quality of its planned actions is also secured by the accurate predictions of the aligned world model. They together considerably improve learning efficiency in a new environment. On open-world challenges in Mars (Minecraft) and ALFWorld (embodied indoor environments), WALL-E significantly outperforms existing methods, e.g., surpassing baselines in Mars by 16.1%–51.6% of success rate and by at least 61.7% in score. In ALFWorld, it achieves a new record 98% success rate after only 4 iterations.

Overview of WALL-E (Left) and NeuroSymbolic Learning details (Right). The agent determines actions to take via MPC, where an LLM optimizes future steps’ actions by interacting with a neurosymbolic world model. WALL-E iteratively refines the symbolic knowledge with the agent’s actual trajectories in the environment and the world model predicted trajectories. The NeuroSymbolic learning takes 4 stages: (1) comparing predicted and actual trajectories; (2) learning new symbolic knowledge from real trajectories; (3) translating symbolic knowledge to code; and (4) Code rule set pruning via solving a maximum coverage problem.

Main Results

Mars

MOD. TYPE	React	Reflexion	Skill Library	IfR	WALL-E
DEFAULT	7.7 ± 1.6	6.0 ± 1.7	8.0 ± 2.1	9.0 ± 2.3	9.5 ± 2.1
TERRAIN	7.4 ± 2.7	6.4 ± 3.0	9.5 ± 2.9	8.0 ± 3.7	10.7 ± 2.6
SURVIVAL	6.4 ± 3.7	4.6 ± 3.9	7.9 ± 2.9	7.7 ± 3.7	13.8 ± 4.4
TASK. DEP.	5.0 ± 2.1	3.2 ± 1.6	1.5 ± 1.9	5.6 ± 2.9	6.4 ± 2.9
TERR. SURV.	6.7 ± 2.5	4.9 ± 2.5	3.0 ± 2.5	6.8 ± 1.9	5.5 ± 2.7
TERR. TASK.	4.8 ± 2.0	5.3 ± 2.5	5.5 ± 1.5	6.9 ± 1.8	5.8 ± 2.2
SURV. TASK.	1.5 ± 1.3	1.0 ± 1.6	2.3 ± 1.5	3.3 ± 1.4	3.2 ± 1.4
ALL THREE	0.7 ± 1.6	−0.4 ± 0.7	−0.5 ± 0.5	0.1 ± 0.5	1.3 ± 1.6
AVG.	4.6	3.6	4.2	5.5	6.7

• The table shows rewards and corresponding std.
• MOD. TYPE refers to World types, including Default (original Crafter setting with no modifications), individual modifications (Terrain, Survival, Task Dependency), and combinations of two or all three modifications (Terr. Surv., Terr. Task., Surv. Task., All Three).

ALFWorld

Method	Avg.	Pick	Clean	Heat	Cool	Examine	Picktwo
BUTLER	26	31	41	60	27	12	29
GPT-BUTLER	69	62	81	85	78	50	47
DEPS	76	93	50	80	100	100	0
AutoGen	77	92	74	78	86	83	41
ReAct	74	79	54	96	85	83	51
AdaPlanner	91	100	100	89	100	97	47
Reflexion	86	92	94	70	81	90	88
RAFA	95	100	97	91	95	100	82
WALL-E (ours)	98	100	100	96	100	100	94
Human Performance	91	-	-	-	-	-	-

• The value in each cell represents the Success Rate (%).
• The - symbol indicates no information available for those entries.

Rule Learning Demo

The rule learning demo allows users to perform neurosymbolic learning, learning rules in natural language. You can choose to either:

1. Use an example subset of trajectories we provide for rule learning.
2. Collect and use your own trajectories from your environment for rule learning.

Setup

Install the module dependencies into your environment:

pip install -r requirements.txt

Running Rule Learning

Option 1: Use Provided Trajectories

We provide an example subset of trajectories for rule learning. Simply specify the path to the provided JSON file:

python run_rulelearning.py \
    --model_name gpt-4 \
    --temperature 0.5 \
    --buffer buffer_fact/traj_demo.json \
    --rule_save_dir /path/to/save/rules.json

Option 2: Collect Your Own Trajectories

If you prefer to collect your own data, gather trajectories from your environment and store them in the appropriate JSON format:

{
    "action_type_1": [
        {
            "state": { /* state details */ },
            "action": { /* action details */ },
            "action_result": { /* result details */ }
        },
        {
            "state": { /* state details */ },
            "action": { /* action details */ },
            "action_result": { /* result details */ }
        }
        // More actions...
    ],
    "action_type_2": [
        // Similar structure for different action types
    ]
    // More action types...
}

Replace "/path/to/your_collected_trajectories.json" with the path to your own trajectory data:

python run_rulelearning.py \
    --model_name gpt-4 \
    --temperature 0.5 \
    --buffer /path/to/your_collected_trajectories.json \
    --rule_save_dir /path/to/save/rules.json

ALFWorld

Stage 1: NeuroSymbolic Learning (exploration and knowledge extraction)

bash run_s1.sh

Stage 2: MPC (model-predictive control using learned knowledge)

bash run_s2.sh

Mars

Stage 1: NeuroSymbolic Learning (exploration and knowledge extraction)

python baselines/walle/s1_run.py  --load_world final_world/default --episode 5

Stage 2: MPC (model-predictive control using learned knowledge)

python baselines/walle/s2_run.py  --load_world final_world/default --episode 5

Note: Stage 2 should be run after Stage 1 completes, as it relies on the symbolic knowledge learned in Stage 1.

The backbone API model (GPT-4-32k) employed in the Mars experiments is no longer available. In response, we are optimizing WALL-E’s knowledge-graph extraction pipeline using a program-learning approach for scene-graph induction. Updated code will be released in an upcoming version.

Citation

If you find our work useful, please cite:

@article{zhou2025wall,
  title={WALL-E 2.0: World Alignment by NeuroSymbolic Learning improves World Model-based LLM Agents},
  author={Zhou, Siyu and Zhou, Tianyi and Yang, Yijun and Long, Guodong and Ye, Deheng and Jiang, Jing and Zhang, Chengqi},
  journal={arXiv preprint arXiv:2504.15785},
  year={2025}
}