ayushhh1662309/qwen2.5-3b-legal-merged
The ayushhh1662309/qwen2.5-3b-legal-merged model is a 3.1 billion parameter, Qwen2.5-3B-based causal language model fine-tuned by Ayush Ghatak. Optimized for legal domain text processing, it excels at contract deconstruction and generating plain-English summaries of complex legal documents. This model specializes in structured legal clause analysis, providing both simplified summaries and a step-by-step cognitive roadmap of legal texts.
Loading preview...
Overview
This model, developed by Ayush Ghatak, is a fine-tuned, merged version of the Qwen2.5-3B architecture, specifically optimized for legal domain text processing. It integrates custom LoRA adapters trained on a curated dataset of legal clauses, terms of service agreements, and privacy policies. The model's 3.1 billion parameters and 32768 token context length make it suitable for detailed legal analysis.
Key Capabilities
- Plain-English Summarization: Condenses complex legal boilerplate into highly scannable, high-quality bullet points.
- Meta-Cognitive Reasoning: Provides an algorithmic, step-by-step cognitive roadmap, identifying document types, legal parties, translating specific terms, and tracing logical breakdowns.
- Structured Legal Clause Analysis: Designed to perform structured analysis of legal clauses.
Usage and Differentiation
This model is distinct due to its specialized focus on legal texts, offering capabilities beyond general-purpose LLMs. It is particularly effective for tasks requiring the deconstruction and simplification of legal documents. For optimal performance in legal clause deconstruction, the model is trained to respond to structural tags like <Reasoning>: and <Summary>:, allowing for clean UI-layer splitting and structured output. This enables users to obtain both a detailed logical trace and a concise summary from a single input.