🚀 Qwen2.5-3B Fine-Tuned on BBH (Formal Fallacies) - Model Card 📌 Model Overview Model Name: Qwen2.5-3B Fine-Tuned on BBH (Formal Fallacies) Base Model: Qwen2.5-3B-Instruct Fine-Tuned Dataset: BBH (BigBench Hard) - Formal Fallacies Task: Logical Reasoning & Deductive Validity Classification Fine-Tuning Objective: Improve the model’s ability to classify logical arguments as valid or invalid based on deductive reasoning principles. 📌 Dataset Information This model was fine-tuned on the Formal Fallacies subset of the BigBench Hard (BBH) dataset.
Dataset characteristics:
Task Type: Deductive reasoning & formal logic classification Input Format: Logical argument statements presented in natural language Target Labels: "valid" or "invalid" Example:
Input: "Here comes a perfectly valid argument: First, being a cousin of Chris is sufficient for not being a son of Kermit. We may conclude that whoever is not a son of Kermit is a cousin of Chris."
Target: "invalid"
This dataset evaluates a model’s ability to identify logically valid vs. invalid arguments, which is crucial for AI-assisted legal analysis, debate systems, and automated theorem proving.
- Downloads last month
- 4