K1mG0ng/AI-taste-business-finance-4B

TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:May 1, 2026License:mitArchitecture:Transformer Open Weights Cold

K1mG0ng/AI-taste-business-finance-4B is a 4 billion parameter Qwen3-based model fine-tuned for evaluating social science research articles, specifically in Business and Finance. With a 32768 token context length, it excels at tasks like research question quality assessment, article-level prompt scoring, and social science journal tier benchmarking. This model is optimized for classifying research-article prompts into discrete quality levels, achieving a Top-1 accuracy of 53.50% and a Top-1+2 accuracy of 82.50% on 200 samples.

Loading preview...

AI-Taste-Business-Finance-4B Overview

This model, developed by K1mG0ng, is a fine-tuned Qwen3 4B architecture specifically designed for evaluating social science research articles, with a primary focus on Business and Finance content. It leverages a 32768 token context length to process detailed article prompts.

Key Capabilities

  • Research Article Evaluation: Optimized for ranking or classifying research-article prompts into discrete quality levels.
  • Specialized Assessment: Capable of assessing research question quality, scoring article-level prompts, and benchmarking social science journal tiers.
  • High Accuracy: Achieves a Top-1 accuracy of 53.50% and a Top-1+2 accuracy of 82.50% on 200 samples, with a balanced accuracy and Macro F1 of 53.50% and 53.19% respectively.

Good for

  • Academic Research: Ideal for researchers and institutions conducting internal experiments on structured article-evaluation prompts in Business and Finance.
  • Content Quality Assessment: Useful for automated assessment of research article quality and relevance within specific social science domains.
  • Journal Benchmarking: Can be applied to benchmark the quality of articles across different social science journal tiers.

This model is intended for research and internal experimentation and is not a general-purpose factual assistant or a substitute for expert peer review.