Datasets:

YuhaoZhan
/

PACE-Bench

C_01 dict	C_02 dict	C_03 dict	C_04 dict	C_05 dict	C_06 dict	D_01 dict	D_02 dict	D_03 dict	D_04 dict	D_05 dict	D_06 dict	E_01 dict	E_02 dict	E_03 dict	E_04 dict	E_05 dict	E_06 dict	F_01 dict	F_02 dict	F_03 dict	F_04 dict	F_05 dict	S_01 dict	S_02 dict	S_03 dict	S_04 dict	S_05 dict	S_06 dict	F_06 dict	K_01 dict	K_02 dict	K_03 dict	K_04 dict	K_05 dict	K_06 dict
{ "APPLY_CART_FORCE": "### Apply Cart Force\n```python\nsandbox.apply_cart_force(f: float)\n```\n- `f`: Horizontal force in Newtons (N). Positive is right, negative is left.\n- Actuator Limits: Clamped to [-450, 450] N each simulation step (no additional rate limit or actuation delay in the environment).\n- **Fun...	{ "APPLY_THRUST": "### Apply Thrust\n```python\nsandbox.apply_thrust(f1, f2)\n```\n- `f1`: Main engine thrust magnitude along the craft's body-up axis (world +y when the hull is upright).\n- `f2`: Steering torque.\n- Function: Primary propulsion and orientation control for the lander.", "GET_GROUND_Y_TOP": ...	{ "APPLY_SEEKER_FORCE": "### Apply Seeker Force\n```python\nsandbox.apply_seeker_force(fx, fy)\n```\n- `fx, fy`: Force vector components (Newtons).\n- Function: Primary control input for maneuvering the seeker.", "GET_CORRIDOR_BOUNDS": "### Get Corridor Bounds\n```python\nx_min, x_max = sandbox.get_corridor_bou...	{ "APPLY_AGENT_FORCE": "### Apply Agent Force\n```python\nsandbox.apply_agent_force(fx, fy)\n```\n- `fx, fy`: Force components (Newtons). No explicit clamp in this task; use reasonable magnitudes to respect the structural impulse limit.\n- Function: Standard movement control for the bot.", "GET_AGENT_BODY": "##...	{ "APPLY_AGENT_FORCE": "### Apply Agent Force\n```python\nsandbox.apply_agent_force(fx, fy)\n```\n- `fx, fy`: Force components in Newtons (N)—the same unit used for in-zone force limits and repulsion scales in the task description.\n- Function: Standard movement control for the bot.", "GET_AGENT_BODY": "###...	{ "APPLY_MOTOR_TORQUE": "### Apply Motor Torque\n```python\nsandbox.apply_motor_torque(torque)\n```\n- `torque`: Requested torque in N·m; magnitude is clamped to the actuator limit for that step.\n- Function: Applies the requested torque to the wheel for that step; close the loop using `get_target_speed()` and th...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: each dimension is clamped to [0.1...	{ "CONTROLLABLE_ENTITY_NOTE": "### Designated Controllable Entity (D-02)\n- The jumper is this task's designated controllable entity. The only sanctioned way to control its launch is `sandbox.set_jumper_velocity(vx, vy)` in `agent_action`. You may read its state via `sandbox.get_jumper()`. The prohibition on modi...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: each dimension is clamped to [0.0...	{ "APPLY_FORCE_TO_SEAT": "### Apply Force to Seat\n```python\nsandbox.apply_force_to_seat(fx, fy)\n```\n- `fx, fy`: Force components (Newtons). Clamped per step to the task's pump force limit (e.g. 42 N); use within that range for horizontal and/or vertical pumping.\n- Function: Applies force to the swing seat ce...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: each dimension is clamped to [0.1...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: each dimension is clamped to [0.1...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam.\n- `angle`: The initial rotation of the beam in radians (clock...	{ "APPLY_THRUST": "### Apply Thrust\n```python\nsandbox.apply_thrust(f1, f2)\n```\n- `f1, f2`: Control inputs.\n- Usage:\n - In Lunar Lander: `f1` is main engine thrust (vertical), `f2` is steering torque.\n - In Exotic Racer: `f1, f2` are horizontal and vertical thrust components.\n- Function: Task-specifi...	{ "APPLY_THRUST": "### Apply Thrust\n```python\nsandbox.apply_thrust(f1, f2)\n```\n- `f1, f2`: Control inputs.\n- Usage:\n - In Lunar Lander: `f1` is main engine thrust (vertical), `f2` is steering torque.\n - In Exotic Racer: `f1, f2` are horizontal and vertical thrust components.\n- Function: Task-specifi...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: each dimension is clamped to [0.1...	{ "APPLY_THRUST": "### Apply Thrust\n```python\nsandbox.apply_thrust(f1, f2)\n```\n- `f1, f2`: Control inputs.\n- Usage:\n - In Lunar Lander: `f1` is main engine thrust (vertical), `f2` is steering torque.\n - In Exotic Racer: `f1, f2` are horizontal and vertical thrust components.\n- Function: Task-specifi...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam.\n- `angle`: The initial rotation of the beam in radians (clock...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=500.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: clamped to [MIN_BEAM_SIZE, MAX_...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: each dimension is clamped to [0.1...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=300.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam.\n- `angle`: The initial rotation of the beam in radians (clo...	{ "ADD_BEAM": "### Add Beam (Dynamic)\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=200.0)\n```\n- `x, y`: Center coordinates (meters).\n- `width, height`: Dimensions (meters). Enforced by environment: each dimension is clamped to [0.08, 1.0] m.\n- `angle`: Rotation in radians.\n- `den...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=150.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam. Enforced by environment: the center must lie inside the build zone from the task description (numeric bounds may differ in variants)...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0, friction=0.5)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Each dimension is clamped by the environment to ...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Enforced by environment: each dimension is clamped to [0.1...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Each dimension is clamped to [0.1, 15.0] m.\n- `angle`: The in...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0, friction=None)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Clamped by the environment: width to [0.1, ...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam.\n- `width, height`: The full width and height (meters) of the rectangular beam. Task constraints: each dimension in [0.1, 10.0] m.\n- `angle`:...	{ "ADD_BLOCK": "### Add Block\n```python\nsandbox.add_block(x, y, width, height, angle=0, density=None)\n```\n- `x, y`: The coordinates (meters) for the center of the block. The center must lie within the permitted build access zone (task-dependent; see spawn zone in the task description).\n- `width, height`: The dim...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=250.0)\n```\n- `x, y`: The coordinates (meters) for the center of the beam. Must lie within the build zone (see task description).\n- `width, height`: The full width and height (meters) of the rectangular beam. **Con...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: Center coordinates (meters).\n- `width, height`: Dimensions (meters). Constraints (enforced at build time): 0.05 ≤ width, height ≤ 5.0 m.\n- `angle`: Initial rotation in radians.\n- `density`...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: Center coordinates (meters) for the beam. Must lie within the build zone (see BUILD_ZONE).\n- `width, height`: Full width and height (meters). Constraints (enforced by environment): 0.05 ≤ wi...	{ "GET_ANCHOR_FOR_GRIPPER": "### Get Anchor for Gripper\n```python\ngantry = sandbox.get_anchor_for_gripper()\n```\n- Returns: The static gantry body (or None if absent). Use as body_a in add_joint with type='rigid' to attach your gripper base so it does not fall; then build slider and fingers from that base.", ...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x, y`: Center position in meters (m).\n- `width, height`: Full dimensions in m. Constraint: 0.05 <= width, height <= 3.0.\n- `angle`: Initial rotation in radians (default 0).\n- `density`: Material ...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x`, `y`: Center position in meters (m). Must lie within the build zone x=[0, 8], y=[1, 12].\n- `width`, `height`: Full width and height in meters. Constraints (enforced by environment): 0.05 <= widt...	{ "ADD_BEAM": "### Add Beam\n```python\nbeam = sandbox.add_beam(x, y, width, height, angle=0, density=1.0)\n```\n- `x`, `y`: Center position in meters. Must lie within build zone x=[0, 12], y=[2, 10].\n- `width`, `height`: Size in meters. Constraints (enforced): 0.05 <= width, height <= 2.0; values outside are cl...

Task mirror only. This dataset repository contains the 36 executable task definitions. The CLI, shared runtime, evaluation engine, self-evolving methods, reporting code, and coding-agent sandbox live in the PACE-Bench GitHub repository.

Mirror provenance

Source: src/pace_bench/tasks/categories
Included: 36 base tasks and primitives_api.json
Excluded: shared runtime, evaluation methods, generated results, caches, and local artifacts

PACE-Bench

Self-evolving agents improve future behavior from interaction experience, yet existing evaluations typically keep execution conditions fixed. PACE-Bench tests whether an agent can adapt a previously successful code-driven design after an environment shift causes it to fail.

What is PACE-Bench?

PACE-Bench contains 144 source-to-target adaptation pairs across six physics domains. Each pair keeps the goal and interface fixed:

A code-driven design succeeds in the source environment.
The same design fails in a mutated target environment.
The agent uses diagnostic sandbox feedback to revise the design.
The adapted design must succeed under the target physics.

Benchmark scale	Count
Physics domains	6
Base tasks	36
Environments per task	5
Evaluation environments	180
Source-to-target pairs	144

What is included here?

Domain	Prefix	Tasks
Statics / Equilibrium	`S`	6
Kinematics / Linkages	`K`	6
Dynamics / Energy	`D`	6
Granular / Fluid Interaction	`F`	6
Cybernetics / Control	`C`	6
Exotic Physics	`E`	6

tasks/
├── Category1_Statics_Equilibrium/S_01 ... S_06/
├── Category2_Kinematics_Linkages/K_01 ... K_06/
├── Category3_Dynamics_Energy/D_01 ... D_06/
├── Category4_Granular_FluidInteraction/F_01 ... F_06/
├── Category5_Cybernetics_Control/C_01 ... C_06/
├── Category6_ExoticPhysics/E_01 ... E_06/
└── primitives_api.json

Each task package contains:

File	Role
`agent.py`	Source and four target reference solutions
`environment.py`	Box2D world, primitives, and mutable physics
`evaluator.py`	Success criteria, score, constraints, and raw metrics
`feedback.py`	Diagnostic feedback derived from measured metrics
`prompt.py`	Task description, exposed values, and primitive API
`renderer.py`	Evaluation-neutral visualization
`stages.py`	Four target mutations and prompt updates

These are executable benchmark definitions, not a conventional row-based dataset. The Dataset Viewer is therefore not the primary interface.

Download the task mirror

from huggingface_hub import snapshot_download

path = snapshot_download(
    repo_id="YuhaoZhan/PACE-Bench",
    repo_type="dataset",
    allow_patterns=["tasks/**"],
)
print(path)

Or clone it directly:

git clone https://huggingface.co/datasets/YuhaoZhan/PACE-Bench

Use this mirror when you need to inspect, archive, or distribute the task definitions without the full evaluation stack.

Run the benchmark

The Hugging Face mirror is not standalone. For evaluation, install the complete GitHub repository with uv:

git clone https://github.com/thunlp/PACE-Bench.git
cd PACE-Bench

uv venv .venv --python 3.10
source .venv/bin/activate  # Windows: .venv\Scripts\activate
uv pip install -r requirements.txt

pace-bench list --task S_01
pace-bench validate --task S_01

The GitHub checkout already contains the same task definitions. You do not need to download this mirror separately to run PACE-Bench.

Intended use

Evaluate adaptation after controlled physical environment changes
Study feedback-driven code evolution and self-evolving agents
Compare context-, memory-, search-, and parameter-based methods
Analyze physical reasoning, redesign, exploration, and convergence failures
Inspect or extend executable task definitions

Scope and limitations

Physics: 2D rigid-body systems in Box2D
Language: English prompts and diagnostic feedback
Not covered: 3D/deformable physics, full fluids, perception, navigation, and multi-agent coordination
Execution safety: run generated code on a dedicated evaluator host without unrelated credentials
Version note: these tasks include an additional difficulty-escalation pass beyond the paper version, so new scores may differ slightly while its conclusions remain unchanged

Issues and contributions

The Hugging Face repository is a distribution mirror. Please open task issues, fixes, and pull requests in the GitHub repository.

License

PACE-Bench is released under the MIT License.

Citation

The public preprint link will be added after release. Until then, please cite the project:

@misc{zhan2026pacebench,
  title        = {PACE-Bench: Benchmarking Physics Adaptation via Code Evolution in Dynamic Environments},
  author       = {Zhan, Yuhao and He, Bingxiang and Tang, Zecong and Xiao, Chaojun},
  year         = {2026},
  howpublished = {\url{https://github.com/thunlp/PACE-Bench}}
}

Downloads last month: 39

Number of rows:

Total file size:

3.01 MB