🚀 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.
