vector-institute/Qwen3-8B-UnBias-Plus-SFT-Instruct-Legacy

TEXT GENERATIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Mar 22, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Qwen3-8B-UnBias-Plus-SFT-Instruct-Legacy is an 8 billion parameter instruction-tuned causal language model developed by the Vector Institute, fine-tuned from Qwen/Qwen3-8B. It specializes in news media bias detection, classification of bias type and severity, and neutral rewriting of biased text. The model provides a fully rewritten unbiased article in a single structured JSON response, making it suitable for production inference due to its direct output and disabled chain-of-thought thinking.

Loading preview...

Model Overview

This model, Qwen3-8B-UnBias-Plus-SFT-Instruct-Legacy, is an 8 billion parameter instruction-tuned variant of the Qwen3-8B base model, developed by the Vector Institute. It is specifically fine-tuned for the task of news media bias detection and neutral rewriting. Unlike other variants, this "Instruct" version is trained without chain-of-thought thinking blocks (enable_thinking=False), ensuring it produces clean, structured JSON directly for faster and more reliable production inference.

Key Capabilities

  • Bias Detection: Identifies biased language segments within news articles.
  • Bias Classification: Classifies the type and severity of detected biases (e.g., loaded language, dehumanizing framing, sensationalism).
  • Neutral Rewriting: Provides neutral replacements for biased phrases and generates a fully rewritten, unbiased version of the article.
  • Structured Output: Delivers all analysis and rewritten content in a single, structured JSON response, ideal for programmatic use.
  • Production Ready: Optimized for deployment via vLLM or other OpenAI-compatible serving backends due to its direct JSON output.

Use Cases

  • Automated News Analysis: Analyzing large volumes of news content for potential bias.
  • Content Moderation: Assisting in identifying and mitigating biased language in published or user-generated content.
  • Journalism Tools: Providing journalists with a tool to review and refine their articles for neutrality.
  • Research: Studying patterns of bias in media and evaluating the effectiveness of bias mitigation strategies.

Limitations

  • Primarily trained on English-language news articles.
  • Political bias detection reflects patterns present in its training data.
  • Best performance is observed on articles under 5000 characters.
  • Human review of outputs is recommended for production use.

Finetunes
1 models