[+] Init commit

.gitattributes

@@ -58,3 +58,12 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+task_1/topological_path_test.json filter=lfs diff=lfs merge=lfs -text
+task_2/topological_connectivity_test.json filter=lfs diff=lfs merge=lfs -text
+task_3/physical_test.json filter=lfs diff=lfs merge=lfs -text
+task_4/counterfactual_test.json filter=lfs diff=lfs merge=lfs -text
+task_6/opf_pglib_opf_case118_ieee_test.json filter=lfs diff=lfs merge=lfs -text
+task_6/opf_pglib_opf_case14_ieee_test.json filter=lfs diff=lfs merge=lfs -text
+task_6/opf_pglib_opf_case30_ieee_test.json filter=lfs diff=lfs merge=lfs -text
+task_7/ctu_test.json filter=lfs diff=lfs merge=lfs -text
+task_7/lcld_test.json filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,395 @@
+Attribution 4.0 International
+
+=======================================================================
+
+Creative Commons Corporation ("Creative Commons") is not a law firm and
+does not provide legal services or legal advice. Distribution of
+Creative Commons public licenses does not create a lawyer-client or
+other relationship. Creative Commons makes its licenses and related
+information available on an "as-is" basis. Creative Commons gives no
+warranties regarding its licenses, any material licensed under their
+terms and conditions, or any related information. Creative Commons
+disclaims all liability for damages resulting from their use to the
+fullest extent possible.
+
+Using Creative Commons Public Licenses
+
+Creative Commons public licenses provide a standard set of terms and
+conditions that creators and other rights holders may use to share
+original works of authorship and other material subject to copyright
+and certain other rights specified in the public license below. The
+following considerations are for informational purposes only, are not
+exhaustive, and do not form part of our licenses.
+
+     Considerations for licensors: Our public licenses are
+     intended for use by those authorized to give the public
+     permission to use material in ways otherwise restricted by
+     copyright and certain other rights. Our licenses are
+     irrevocable. Licensors should read and understand the terms
+     and conditions of the license they choose before applying it.
+     Licensors should also secure all rights necessary before
+     applying our licenses so that the public can reuse the
+     material as expected. Licensors should clearly mark any
+     material not subject to the license. This includes other CC-
+     licensed material, or material used under an exception or
+     limitation to copyright. More considerations for licensors:
+	wiki.creativecommons.org/Considerations_for_licensors
+
+     Considerations for the public: By using one of our public
+     licenses, a licensor grants the public permission to use the
+     licensed material under specified terms and conditions. If
+     the licensor's permission is not necessary for any reason--for
+     example, because of any applicable exception or limitation to
+     copyright--then that use is not regulated by the license. Our
+     licenses grant only permissions under copyright and certain
+     other rights that a licensor has authority to grant. Use of
+     the licensed material may still be restricted for other
+     reasons, including because others have copyright or other
+     rights in the material. A licensor may make special requests,
+     such as asking that all changes be marked or described.
+     Although not required by our licenses, you are encouraged to
+     respect those requests where reasonable. More_considerations
+     for the public:
+	wiki.creativecommons.org/Considerations_for_licensees
+
+=======================================================================
+
+Creative Commons Attribution 4.0 International Public License
+
+By exercising the Licensed Rights (defined below), You accept and agree
+to be bound by the terms and conditions of this Creative Commons
+Attribution 4.0 International Public License ("Public License"). To the
+extent this Public License may be interpreted as a contract, You are
+granted the Licensed Rights in consideration of Your acceptance of
+these terms and conditions, and the Licensor grants You such rights in
+consideration of benefits the Licensor receives from making the
+Licensed Material available under these terms and conditions.
+
+
+Section 1 -- Definitions.
+
+  a. Adapted Material means material subject to Copyright and Similar
+     Rights that is derived from or based upon the Licensed Material
+     and in which the Licensed Material is translated, altered,
+     arranged, transformed, or otherwise modified in a manner requiring
+     permission under the Copyright and Similar Rights held by the
+     Licensor. For purposes of this Public License, where the Licensed
+     Material is a musical work, performance, or sound recording,
+     Adapted Material is always produced where the Licensed Material is
+     synched in timed relation with a moving image.
+
+  b. Adapter's License means the license You apply to Your Copyright
+     and Similar Rights in Your contributions to Adapted Material in
+     accordance with the terms and conditions of this Public License.
+
+  c. Copyright and Similar Rights means copyright and/or similar rights
+     closely related to copyright including, without limitation,
+     performance, broadcast, sound recording, and Sui Generis Database
+     Rights, without regard to how the rights are labeled or
+     categorized. For purposes of this Public License, the rights
+     specified in Section 2(b)(1)-(2) are not Copyright and Similar
+     Rights.
+
+  d. Effective Technological Measures means those measures that, in the
+     absence of proper authority, may not be circumvented under laws
+     fulfilling obligations under Article 11 of the WIPO Copyright
+     Treaty adopted on December 20, 1996, and/or similar international
+     agreements.
+
+  e. Exceptions and Limitations means fair use, fair dealing, and/or
+     any other exception or limitation to Copyright and Similar Rights
+     that applies to Your use of the Licensed Material.
+
+  f. Licensed Material means the artistic or literary work, database,
+     or other material to which the Licensor applied this Public
+     License.
+
+  g. Licensed Rights means the rights granted to You subject to the
+     terms and conditions of this Public License, which are limited to
+     all Copyright and Similar Rights that apply to Your use of the
+     Licensed Material and that the Licensor has authority to license.
+
+  h. Licensor means the individual(s) or entity(ies) granting rights
+     under this Public License.
+
+  i. Share means to provide material to the public by any means or
+     process that requires permission under the Licensed Rights, such
+     as reproduction, public display, public performance, distribution,
+     dissemination, communication, or importation, and to make material
+     available to the public including in ways that members of the
+     public may access the material from a place and at a time
+     individually chosen by them.
+
+  j. Sui Generis Database Rights means rights other than copyright
+     resulting from Directive 96/9/EC of the European Parliament and of
+     the Council of 11 March 1996 on the legal protection of databases,
+     as amended and/or succeeded, as well as other essentially
+     equivalent rights anywhere in the world.
+
+  k. You means the individual or entity exercising the Licensed Rights
+     under this Public License. Your has a corresponding meaning.
+
+
+Section 2 -- Scope.
+
+  a. License grant.
+
+       1. Subject to the terms and conditions of this Public License,
+          the Licensor hereby grants You a worldwide, royalty-free,
+          non-sublicensable, non-exclusive, irrevocable license to
+          exercise the Licensed Rights in the Licensed Material to:
+
+            a. reproduce and Share the Licensed Material, in whole or
+               in part; and
+
+            b. produce, reproduce, and Share Adapted Material.
+
+       2. Exceptions and Limitations. For the avoidance of doubt, where
+          Exceptions and Limitations apply to Your use, this Public
+          License does not apply, and You do not need to comply with
+          its terms and conditions.
+
+       3. Term. The term of this Public License is specified in Section
+          6(a).
+
+       4. Media and formats; technical modifications allowed. The
+          Licensor authorizes You to exercise the Licensed Rights in
+          all media and formats whether now known or hereafter created,
+          and to make technical modifications necessary to do so. The
+          Licensor waives and/or agrees not to assert any right or
+          authority to forbid You from making technical modifications
+          necessary to exercise the Licensed Rights, including
+          technical modifications necessary to circumvent Effective
+          Technological Measures. For purposes of this Public License,
+          simply making modifications authorized by this Section 2(a)
+          (4) never produces Adapted Material.
+
+       5. Downstream recipients.
+
+            a. Offer from the Licensor -- Licensed Material. Every
+               recipient of the Licensed Material automatically
+               receives an offer from the Licensor to exercise the
+               Licensed Rights under the terms and conditions of this
+               Public License.
+
+            b. No downstream restrictions. You may not offer or impose
+               any additional or different terms or conditions on, or
+               apply any Effective Technological Measures to, the
+               Licensed Material if doing so restricts exercise of the
+               Licensed Rights by any recipient of the Licensed
+               Material.
+
+       6. No endorsement. Nothing in this Public License constitutes or
+          may be construed as permission to assert or imply that You
+          are, or that Your use of the Licensed Material is, connected
+          with, or sponsored, endorsed, or granted official status by,
+          the Licensor or others designated to receive attribution as
+          provided in Section 3(a)(1)(A)(i).
+
+  b. Other rights.
+
+       1. Moral rights, such as the right of integrity, are not
+          licensed under this Public License, nor are publicity,
+          privacy, and/or other similar personality rights; however, to
+          the extent possible, the Licensor waives and/or agrees not to
+          assert any such rights held by the Licensor to the limited
+          extent necessary to allow You to exercise the Licensed
+          Rights, but not otherwise.
+
+       2. Patent and trademark rights are not licensed under this
+          Public License.
+
+       3. To the extent possible, the Licensor waives any right to
+          collect royalties from You for the exercise of the Licensed
+          Rights, whether directly or through a collecting society
+          under any voluntary or waivable statutory or compulsory
+          licensing scheme. In all other cases the Licensor expressly
+          reserves any right to collect such royalties.
+
+
+Section 3 -- License Conditions.
+
+Your exercise of the Licensed Rights is expressly made subject to the
+following conditions.
+
+  a. Attribution.
+
+       1. If You Share the Licensed Material (including in modified
+          form), You must:
+
+            a. retain the following if it is supplied by the Licensor
+               with the Licensed Material:
+
+                 i. identification of the creator(s) of the Licensed
+                    Material and any others designated to receive
+                    attribution, in any reasonable manner requested by
+                    the Licensor (including by pseudonym if
+                    designated);
+
+                ii. a copyright notice;
+
+               iii. a notice that refers to this Public License;
+
+                iv. a notice that refers to the disclaimer of
+                    warranties;
+
+                 v. a URI or hyperlink to the Licensed Material to the
+                    extent reasonably practicable;
+
+            b. indicate if You modified the Licensed Material and
+               retain an indication of any previous modifications; and
+
+            c. indicate the Licensed Material is licensed under this
+               Public License, and include the text of, or the URI or
+               hyperlink to, this Public License.
+
+       2. You may satisfy the conditions in Section 3(a)(1) in any
+          reasonable manner based on the medium, means, and context in
+          which You Share the Licensed Material. For example, it may be
+          reasonable to satisfy the conditions by providing a URI or
+          hyperlink to a resource that includes the required
+          information.
+
+       3. If requested by the Licensor, You must remove any of the
+          information required by Section 3(a)(1)(A) to the extent
+          reasonably practicable.
+
+       4. If You Share Adapted Material You produce, the Adapter's
+          License You apply must not prevent recipients of the Adapted
+          Material from complying with this Public License.
+
+
+Section 4 -- Sui Generis Database Rights.
+
+Where the Licensed Rights include Sui Generis Database Rights that
+apply to Your use of the Licensed Material:
+
+  a. for the avoidance of doubt, Section 2(a)(1) grants You the right
+     to extract, reuse, reproduce, and Share all or a substantial
+     portion of the contents of the database;
+
+  b. if You include all or a substantial portion of the database
+     contents in a database in which You have Sui Generis Database
+     Rights, then the database in which You have Sui Generis Database
+     Rights (but not its individual contents) is Adapted Material; and
+
+  c. You must comply with the conditions in Section 3(a) if You Share
+     all or a substantial portion of the contents of the database.
+
+For the avoidance of doubt, this Section 4 supplements and does not
+replace Your obligations under this Public License where the Licensed
+Rights include other Copyright and Similar Rights.
+
+
+Section 5 -- Disclaimer of Warranties and Limitation of Liability.
+
+  a. UNLESS OTHERWISE SEPARATELY UNDERTAKEN BY THE LICENSOR, TO THE
+     EXTENT POSSIBLE, THE LICENSOR OFFERS THE LICENSED MATERIAL AS-IS
+     AND AS-AVAILABLE, AND MAKES NO REPRESENTATIONS OR WARRANTIES OF
+     ANY KIND CONCERNING THE LICENSED MATERIAL, WHETHER EXPRESS,
+     IMPLIED, STATUTORY, OR OTHER. THIS INCLUDES, WITHOUT LIMITATION,
+     WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR
+     PURPOSE, NON-INFRINGEMENT, ABSENCE OF LATENT OR OTHER DEFECTS,
+     ACCURACY, OR THE PRESENCE OR ABSENCE OF ERRORS, WHETHER OR NOT
+     KNOWN OR DISCOVERABLE. WHERE DISCLAIMERS OF WARRANTIES ARE NOT
+     ALLOWED IN FULL OR IN PART, THIS DISCLAIMER MAY NOT APPLY TO YOU.
+
+  b. TO THE EXTENT POSSIBLE, IN NO EVENT WILL THE LICENSOR BE LIABLE
+     TO YOU ON ANY LEGAL THEORY (INCLUDING, WITHOUT LIMITATION,
+     NEGLIGENCE) OR OTHERWISE FOR ANY DIRECT, SPECIAL, INDIRECT,
+     INCIDENTAL, CONSEQUENTIAL, PUNITIVE, EXEMPLARY, OR OTHER LOSSES,
+     COSTS, EXPENSES, OR DAMAGES ARISING OUT OF THIS PUBLIC LICENSE OR
+     USE OF THE LICENSED MATERIAL, EVEN IF THE LICENSOR HAS BEEN
+     ADVISED OF THE POSSIBILITY OF SUCH LOSSES, COSTS, EXPENSES, OR
+     DAMAGES. WHERE A LIMITATION OF LIABILITY IS NOT ALLOWED IN FULL OR
+     IN PART, THIS LIMITATION MAY NOT APPLY TO YOU.
+
+  c. The disclaimer of warranties and limitation of liability provided
+     above shall be interpreted in a manner that, to the extent
+     possible, most closely approximates an absolute disclaimer and
+     waiver of all liability.
+
+
+Section 6 -- Term and Termination.
+
+  a. This Public License applies for the term of the Copyright and
+     Similar Rights licensed here. However, if You fail to comply with
+     this Public License, then Your rights under this Public License
+     terminate automatically.
+
+  b. Where Your right to use the Licensed Material has terminated under
+     Section 6(a), it reinstates:
+
+       1. automatically as of the date the violation is cured, provided
+          it is cured within 30 days of Your discovery of the
+          violation; or
+
+       2. upon express reinstatement by the Licensor.
+
+     For the avoidance of doubt, this Section 6(b) does not affect any
+     right the Licensor may have to seek remedies for Your violations
+     of this Public License.
+
+  c. For the avoidance of doubt, the Licensor may also offer the
+     Licensed Material under separate terms or conditions or stop
+     distributing the Licensed Material at any time; however, doing so
+     will not terminate this Public License.
+
+  d. Sections 1, 5, 6, 7, and 8 survive termination of this Public
+     License.
+
+
+Section 7 -- Other Terms and Conditions.
+
+  a. The Licensor shall not be bound by any additional or different
+     terms or conditions communicated by You unless expressly agreed.
+
+  b. Any arrangements, understandings, or agreements regarding the
+     Licensed Material not stated herein are separate from and
+     independent of the terms and conditions of this Public License.
+
+
+Section 8 -- Interpretation.
+
+  a. For the avoidance of doubt, this Public License does not, and
+     shall not be interpreted to, reduce, limit, restrict, or impose
+     conditions on any use of the Licensed Material that could lawfully
+     be made without permission under this Public License.
+
+  b. To the extent possible, if any provision of this Public License is
+     deemed unenforceable, it shall be automatically reformed to the
+     minimum extent necessary to make it enforceable. If the provision
+     cannot be reformed, it shall be severed from this Public License
+     without affecting the enforceability of the remaining terms and
+     conditions.
+
+  c. No term or condition of this Public License will be waived and no
+     failure to comply consented to unless expressly agreed to by the
+     Licensor.
+
+  d. Nothing in this Public License constitutes or may be interpreted
+     as a limitation upon, or waiver of, any privileges and immunities
+     that apply to the Licensor or You, including from the legal
+     processes of any jurisdiction or authority.
+
+
+=======================================================================
+
+Creative Commons is not a party to its public
+licenses. Notwithstanding, Creative Commons may elect to apply one of
+its public licenses to material it publishes and in those instances
+will be considered the “Licensor.” The text of the Creative Commons
+public licenses is dedicated to the public domain under the CC0 Public
+Domain Dedication. Except for the limited purpose of indicating that
+material is shared under a Creative Commons public license or as
+otherwise permitted by the Creative Commons policies published at
+creativecommons.org/policies, Creative Commons does not authorize the
+use of the trademark "Creative Commons" or any other trademark or logo
+of Creative Commons without its prior written consent including,
+without limitation, in connection with any unauthorized modifications
+to any of its public licenses or any other arrangements,
+understandings, or agreements concerning use of licensed material. For
+the avoidance of doubt, this paragraph does not form part of the
+public licenses.
+
+Creative Commons may be contacted at creativecommons.org.

README.md

@@ -0,0 +1,186 @@
+# ORC-bench
+
+| Task 1: Topological Path Finding | Task 2: Topological Connectivity | Task 3: Linear Power Flow |
+|---|---|---|
+| ![Task 1](figures/task%201.svg) | ![Task 2](figures/task%202.svg) | ![Task 3](figures/task%203.svg) |
+
+| Task 4: Contingency Analysis | Task 5: Power Grid Control | Task 6: Power Flow Optimization |
+|---|---|---|
+| ![Task 4](figures/task%204.svg) | ![Task 5](figures/task%205.svg) | ![Task 6](figures/task%206.svg) |
+
+
+## Task 1: Topological Path Finding
+
+### Problem Formulation
+This task assesses the spatial reasoning ability of the model by asking it to determine the shortest path between two specific buses in a given power grid state. The grid state may include modifications, such as the temporary removal of transmission lines, requiring the model to dynamically reason over the provided text-based topology rather than relying on memorized standard grid structures.
+
+### Dataset Generation
+The dataset is generated using standard Institute of Electrical and Electronics Engineers (IEEE) 14, 30, and 118 bus cases. To ensure diversity and require active reasoning, up to 2 unique lines (for IEEE 14) or 5 unique lines (for larger cases) are randomly removed. A NetworkX MultiGraph is constructed using the remaining operational lines (buses can be connected by more than 1 line in IEEE 118). Two random buses are selected as the source and target. The `all_shortest_paths` algorithm is used to compute the valid paths. If a path exists, the ground truth is formatted as a sequence of nodes (e.g., `Bus A -> Bus B -> Bus C`); if none exists, the answer is `No path exists.`
+
+### LLM Evaluation
+The LLM is provided with the textual representation of the grid topology (including the base system and any offline lines) and the query asking for the shortest path.
+
+**System Prompt:**  
+> *You are a power systems expert. Provide the shortest path between the requested buses in the following format including the starting and ending buses: Bus A -> Bus B -> Bus C. If no path exists, respond with 'No path exists.' Do not provide any other text.*
+
+### Performance Metrics
+Performance is evaluated using an exact path-matching algorithm. To prevent false positives from partial matches (e.g., predicting “Bus 22” when the ground truth is “Bus 2”), the evaluation script enforces strict alphanumeric word boundaries when searching for the ground truth path within the model's output. The primary metric is **Accuracy**, computed both overall and per IEEE case.
+
+---
+
+## Task 2: Topological Connectivity
+
+### Problem Formulation
+This task evaluates the model's ability to deduce graph connectivity. The model must determine whether a path exists between two specific buses in a degraded power grid, effectively answering a binary “Yes” or “No” question based on the textual grid description.
+
+### Dataset Generation
+Similar to Task 1a, the dataset is generated using **Python** and **NetworkX**, built upon IEEE 14, 30, and 118 bus networks. Up to 5 edges are randomly removed from the base graph to create disconnected subgraphs or alternate routes. Two random buses are sampled, and the `networkx.has_path` function determines the ground truth connectivity state (“Yes” or “No”). The textual prompt lists the offline lines.
+
+### LLM Evaluation
+The model is prompted with the grid state and asked a direct question: “Is [Start Bus] connected to [End Bus]?” The exact system prompt used is:
+
+> *You are a power systems expert. Determine if two buses are connected. Respond only with 'Yes' or 'No'. Do not provide any other text.*
+
+### Performance Metrics
+The metric for this task is binary **Accuracy**. The model's predicted response is compared against the ground truth using a case-insensitive exact match. Accuracy is reported across the entire dataset as well as per grid case.
+
+---
+
+## Task 3: Linear Power Flow
+
+### Problem Formulation
+This task transitions from topological to physical reasoning by asking the model to calculate the active power flow (in Megawatts \[MW]) on a specific transmission line. This tests the LLM's capability to understand electrical parameters and perform implicit physical calculations or heuristics based on grid state descriptions.
+
+### Dataset Generation
+The dataset is generated using **Python** and the **PyPSA** (Python for Power System Analysis) library. It uses IEEE 14, 30, and 118 bus systems. To create unique scenarios, the active power load (`p_set`) and generator capacities (`p_nom`) are randomized uniformly between 80% and 120% of their base values. A Direct Current (DC) linear power flow (LPF) is solved using PyPSA's `network.lpf()` solver to obtain the steady-state active power flows. A target line is randomly chosen, and its simulated flow (rounded to two decimal places) is recorded as the ground truth.
+
+### LLM Evaluation
+The model is provided with the complete, randomized grid state (including node injections) and asked: “What is the active power flow (MW) on [Target Line]?” The exact system prompt used is:
+
+> *You are a power systems expert. Calculate the requested physical value (e.g., Power in MW). Respond only with the numerical value. Do not provide units or any other text.*
+
+### Performance Metrics
+Because the output is continuous, performance is measured using **Mean Absolute Error (MAE)** and **Root Mean Square Error (RMSE)**. A regular expression extracts the first floating-point number from the model's response to compare against the ground truth flow.
+
+---
+
+## Task 4: Contingency Analysis
+
+### Problem Formulation
+This task tests N-1 (N minus one) contingency analysis and counterfactual reasoning. Given a base grid state and a scenario describing the sudden loss of a generator or transmission line, the model must identify the transmission line most likely to overload (experience the highest flow increase) and provide a logical explanation for its choice.
+
+### Dataset Generation
+The dataset is generated using **PyPSA**. Base flows are calculated for IEEE 14, 30, and 118 cases using the PyPSA Linear Power Flow (LPF) solver. A random N-1 contingency is applied programmatically: either a non-slack generator is tripped (output set to 0), or a transmission line is removed. The system is re-solved, and the absolute difference in line flows is calculated to find the line with the maximum spike. The dataset includes a Chain-of-Thought (CoT) answer that explains the lost element, the rerouting of power, and the identification of the overloaded line.
+
+### LLM Evaluation
+The model receives the grid state and the contingency scenario. It must predict the overloaded line and justify it. The exact system prompt used is:
+
+> *You are a power systems expert. Provide the line ID of the most likely overloaded transmission line in double brackets (e.g., \[\[L0\]\] for line 0) and then provide a step-by-step explanation.*
+
+### Performance Metrics
+The primary metric is **Accuracy** of identifying the correct overloaded line. Evaluation is performed by verifying if the ground truth target line string is explicitly mentioned within the predicted response.
+
+---
+
+## Task 5: Power Grid Control
+
+### Problem Formulation
+This task assesses the model's ability to act as a grid operator (agent). Given a set of system telemetry readings (frequency drop and voltage levels) and a sudden disturbance (e.g., wind generation drop), the model must choose the best corrective action from a multiple-choice list.
+
+### Dataset Generation
+The dataset is generated using **Python**'s standard libraries to simulate telemetry logic. For IEEE systems, 1,500 scenarios are generated representing four states: critical frequency drop, moderate frequency drop, low voltage, or stable system. Random telemetry parameters are generated within specific bounds for each state (e.g., critical frequency is between 49.0 and 49.49 Hz). Ground truth actions are deterministically mapped to the scenarios (e.g., a critical frequency drop requires spinning up a gas turbine).
+
+### LLM Evaluation
+The model is prompted with the grid state, telemetry data, and four possible actions (A, B, C, D). The exact system prompt used is:
+
+> *You are a power grid operator. Based on the telemetry provided, select the most appropriate control action from the given choices. Respond by providing the choice label in brackets, e.g., \[B\].*
+
+### Performance Metrics
+The metric is **Accuracy**, calculated by checking whether the correct textual explanation or the correct action bracket (e.g., `[B]`) is present within the model's lowercased prediction string.
+
+---
+
+## Task 6: Optimal Power Flow Optimization
+
+### Problem Formulation
+This task represents the most complex numerical reasoning challenge, assessing the model's ability to act as an end-to-end mathematical solver for the Optimal Power Flow (OPF) problem. Given a serialized heterogeneous graph representation of a power grid—including buses, loads, generators, and transmission lines—the model must compute the optimal operating state. This requires balancing complex non-linear physical constraints (Kirchhoff's laws) and operational limits (voltage bounds, generator capacities) to determine active and reactive power injections and bus voltage magnitudes and angles.
+
+### Dataset Generation
+The dataset is procedurally generated using **PyTorch Geometric (PyG)** library. Specifically, we are curating and adapting the `OPFDataset` used for traditional GNN evaluation for a LLM setting. Complex grid topologies are parsed into a heterogeneous graph format, which is then serialized into structured JSON dictionaries mapping node features and edge indices. To facilitate in-context learning, few-shot prompt examples are constructed, pairing an “Example Input JSON” with its corresponding solved “Example Output JSON”.
+
+### LLM Evaluation
+The model is prompted with the system instructions, the few-shot examples, and a final “Query Input JSON” that it must solve. The exact system prompt used is:
+
+> *You are a powergrid operator running Optimal Power Flow simulation and you need to return a JSON formatted responses based on the provided input JSON. The input is the description of the components of the grid, including the buses, generators, loads, lines, and external grid. The output is the solution to the optimal power flow problem. You will get a few examples of Input and Output json. You need to return the correct Output for the last given Input*
+
+A robust regex-based extraction script parses the model's text output to isolate the final predicted JSON object.
+
+### Performance Metrics
+- **Constraints:** We evaluate each predicted OPF solution using three feasibility criteria. For every bus (i), we compute the active and reactive power-balance residuals  
+  \(\Delta P_i = P_i^{\mathrm{spec}} - P_i^{\mathrm{imp}}\) and \(\Delta Q_i = Q_i^{\mathrm{spec}} - Q_i^{\mathrm{imp}}\),  
+  where the specified injections are given by the predicted generation minus demand, and the implied injections are those induced by the predicted bus voltages under the instance topology. A bus is deemed to satisfy the power-flow equations if:
+
+  \[
+  \sqrt{\Delta P_i^2 + \Delta Qi^2} \le \varepsilon{\mathrm{pf}}.
+  \]
+
+  Voltage feasibility is enforced via the bounds \(v_i^{\min} \le \hat v_i \le v_i^{\max}\), and generator feasibility via the active-power limits \(P_g^{\min} \le \hat P_g \le Pg^{\max}\).
+
+  For each sample, we compute the fraction of buses (or generators) that satisfy each condition, and report the dataset-level mean of these per-sample fractions. We additionally report an overall constraint satisfaction score obtained by micro-averaging all individual bus- and generator-level checks over the entire evaluation set, so that each constraint instance is weighted equally. Specifically, let (s) index evaluation samples, and let \(n{\mathrm{pf}}^{(s)}\), \(n{\mathrm{v}}^{(s)}\), and \(n{\mathrm{g}}^{(s)}\) denote the numbers of evaluated buses and generators for the power-flow, voltage, and generator constraints, with \(n{\mathrm{pf,sat}}^{(s)}\), \(n{\mathrm{v,sat}}^{(s)}\), and \(n_{\mathrm{g,sat}}^{(s)}\) the corresponding numbers satisfied.
+
+- **MSE:** To evaluate the numerical accuracy of the predicted OPF solutions, we compute the Mean Squared Error (MSE) between the model outputs and the ground-truth solver results in the dataset. We report the total of two components: (i) a generator MSE, calculated over each generator's active and reactive power outputs, and (ii) a bus MSE, calculated over the voltage magnitude and voltage angle at each bus. In addition to constraint satisfaction, the MSE quantifies how closely the predictions match the optimal dispatch numerically.
+
+- **Valid rate:** Tracks the model's ability to produce structurally correct JSON (can be parsed without raising error).
+
+---
+
+## Task 7-C: Credit Scoring
+
+### Problem Formulation
+This task assesses the model's ability to act as a financial expert to analyze loan application data and predict missing financial details. Given a structured tabular representation of an applicant's characteristics and loan parameters (in JSON format), the model must predict missing values such as installment, open accounts, bankruptcies, term, and various financial ratios. This evaluates the LLM's capability to reason over financial constraints and tabular data structures.
+
+### Dataset Generation
+The dataset is derived from the Lending Club Loan Data (LCLD) dataset. The objective is to predict the value of a dropped column that is involved in constrained relationship with other columns. For the few-shot in-context learning evaluation, multiple context examples are selected, where each example presents an “Example Input JSON” (containing applicant and loan features) and an “Example Output JSON” (containing the target financial columns). The target variables include continuous and discrete fields.
+
+### LLM Evaluation
+The model is prompted with the few-shot examples and asked to predict the target columns for a “Query Input JSON”. The exact system prompt used is:
+
+> *You are a financial expert. Your task is to analyze the provided loan application data and predict the missing financial details. You need to return a JSON formatted response based on the provided input JSON. The input contains various applicant and loan characteristics. The output should contain the predicted missing values for specific fields such as installment, open accounts, bankruptcies, term, and various financial ratios. You will receive an Input JSON and you must return the correct Output JSON.*
+
+### Performance Metrics
+Evaluation is conducted using a constraints checker from the `TabularBench` framework. The primary quantitative metric is **Mean Squared Error (MSE)** between the predicted numerical values and the ground truth. Additionally, structural and logical validity is measured via the **Constraint Satisfaction (CTR)** metric, which checks if the predicted fields satisfy intrinsic tabular constraints and mathematical invariants defined by the dataset. The proportion of **Invalid JSON** responses is also tracked.
+
+---
+
+## Task 7-U: Phishing Detection
+
+### Problem Formulation
+This task tests the model's capability in cyber-security, specifically acting as an expert in phishing detection. The model is provided with characteristics extracted from a URL (e.g., in JSON format) and must predict missing security details, such as the length of the URL, number of “.com” domains, presence of HTTP in the path, ratio of digits in the URL and host, and number of redirections.
+
+### Dataset Generation
+The dataset is generated from a Phishing URL tabular dataset. A structured prompt is built containing a specific number of few-shot examples. Each example splits the tabular row into an input JSON containing URL characteristics and an output JSON holding the target security fields.
+
+### LLM Evaluation
+The model is evaluated on its ability to infer the missing URL characteristics from the given inputs. The exact system prompt used is:
+
+> *You are a cyber-security expert in fishing detection. Your task is to analyze the provided URL data and predict the missing security details. You need to return a JSON formatted response based on the provided input JSON. The input contains various URL characteristics. The output should contain the predicted missing values for specific fields such as length_url, nb_com, http_in_path, ratio_digits_url, ratio_digits_host, and nb_redirection. You will receive an Input JSON and you must return the correct Output JSON.*
+
+### Performance Metrics
+The evaluation computes the **Mean Squared Error (MSE)** for continuous and discrete predicted URL properties. We also utilize the `tabularbench` constraints checker to report the **Constraint Satisfaction (CTR)** rate, ensuring that the model's output satisfies real-world data constraints (e.g., non-negative lengths and ratios). The **Invalid JSON** error rate is also measured to assess the structural formatting reliability of the LLM.
+
+---
+
+## Task 7-B: BOTNET detection
+
+### Problem Formulation
+This task assesses the model's proficiency in analyzing network traffic to identify botnet attacks. Acting as a cyber-security expert, the LLM analyzes a JSON representation of network traffic characteristics and must predict missing security details related to aggregated communication protocols (e.g., ICMP, UDP, TCP sums, byte flows, and durations).
+
+### Dataset Generation
+The dataset is processed from the CTU-13 NERIS tabular dataset using **Python** and **Pandas**. The data is partitioned, and a series of prompt contexts are generated. Each prompt includes multiple few-shot examples (“Example Input JSON” and “Example Output JSON”), isolating target columns like `icmp_sum_s_1`, `udp_sum_s_1`, `tcp_sum_s_1`, and byte/packet transfer sums from the rest of the network telemetry.
+
+### LLM Evaluation
+The LLM receives the prompt with few-shot network traffic examples and a target query. The exact system prompt used is:
+> *You are a cyber-security expert in botnet attack detection. Your task is to analyze the provided network traffic data and predict the missing security details. You need to return a JSON formatted response based on the provided input JSON. The input contains various network traffic characteristics. The output should contain the predicted missing values for specific fields such as icmp_sum_s_1, udp_sum_s_1, tcp_sum_s_1, bytes_in_sum_d_2, bytes_out_sum_d_2, bytes_out_sum_s_1, pkts_out_sum_s_1, bytes_out_sum_d_3, pkts_out_sum_d_3, duration_max_d_1, and duration_max_s_OTHER. You will receive an Input JSON and you must return the correct Output JSON.*
+
+### Performance Metrics
+Performance is validated using the `TabularBench` framework. The predictions are scored using **Mean Squared Error (MSE)** against the ground truth ports values. A constraint satisfaction check calculates the **Constraint Satisfaction (CTR)** metric, which ensures that the interdependent packet and byte counts adhere to logical network constraints, alongside tracking the **Invalid JSON** structure rate.

task_1/topological_path_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b07fef242644cfea8afdeaf8e9adecd71331dc0b85b625494f0e0a87c609090d
+size 17487470

task_2/topological_connectivity_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0484e46708b2b89cf1368d6e7f96e7d8f799511144453f4facf652761f688c01
+size 16840750

task_3/physical_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65ff3da0ca390100fa6ac24829ea1308bcb1a0994313b0a6788b839d88601ecd
+size 16655185

task_4/counterfactual_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cf0e58128477085586314441ca4f120ba3e92ca5a4c0e461e71673200fa47a7
+size 18578608

task_6/opf_pglib_opf_case118_ieee_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1347d1f4e371b31cdcbe9483331ec8d1842128e6c815c91d8ca53ac0c1fba424
+size 175450487

task_6/opf_pglib_opf_case14_ieee_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbee301de6ba1420553c56bcfc703fdb87d3fde48cb01cab407b0bf353dbe3db
+size 142810229

task_6/opf_pglib_opf_case30_ieee_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3171cdd830b6efae09552d446788bb834df475c53c388f2e69ef06f75887ea2
+size 133940685

task_7/ctu_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d35b186c3bb5874bd1ea7dadfc259c96cca5e36ac5fe37219a3ddee95bf6a62a
+size 407728077

task_7/lcld_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdecac54b51b5ac3e2b4971580bdd7b86a3ef3c2a143788b87608ec8e100bacc
+size 122011294