Advancing Formal Verification: Fine-Tuning LLMs for Translating Natural Language Requirements to CTL Specifications
Bibliographic record
Abstract
In the domain of formal verification, translating natural language (NL) requirements into Computation Tree Logic (CTL) specifications presents a notable challenge due to the disparity between human-readable documents and formal specifications. This paper introduces a novel approach that leverages Large Language Models (LLMs) to automate this translation process, thereby enhancing the accuracy and efficiency of formal verification practices. We fine-tune three state-of-the-art LLMs—LLAMA3, Mistral, and Qwen2—with a particular focus on optimizing the Mistral model due to its superior performance. Our methodology is supported by the Natural2CTL dataset, consisting of 2,095 NL requirements and their corresponding CTL specifications. We employ evaluation metrics such as validation loss, accuracy, semantic similarity, and Structural Operator Jaccard Similarity (SOJS) for a comprehensive assessment of model performance. Additionally, a comparative analysis with human translators, trained in CTL logic, underscores the LLMs’ potential to match or even surpass human accuracy in translating NL requirements into formal specifications. Our findings reveal that the fine-tuned Mistral model significantly outperforms the other LLMs and human participants, demonstrating superior accuracy in generating CTL specifications. This study advances the field of formal verification by proposing a scalable solution to the NL-to-CTL translation challenge, setting a new benchmark for the integration of AI tools in complex specification tasks.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.006 | 0.036 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".