Overview

K1mG0ng/AI-taste-eco-4B is a specialized 4 billion parameter language model, fine-tuned from the Qwen/Qwen3-4B base model. It is specifically developed for AI Taste experiments within the ECO article-evaluation framework. The model utilizes the Qwen3ForCausalLM architecture and is distributed in Hugging Face Transformers format with Safetensors.

Key Capabilities

• Structured Article Evaluation: Optimized for processing prompts designed for article evaluation, including scoring, ranking, and label prediction.
• Research-Oriented: Intended for internal experimentation and research purposes in specific evaluation contexts.
• Efficient Deployment: The repository includes only essential files (weights, tokenizer, config) for loading and running the model, excluding training-state artifacts.

Intended Use Cases

This model is primarily designed for:

• AI Taste Experiments: Conducting research on article evaluation within the ECO setting.
• Structured Prompt Processing: Handling specific, structured prompts for tasks like assigning scores or categories to articles.

It is important to note that this model is not intended as a general-purpose factual assistant or a substitute for expert peer review. Its output quality is highly dependent on the prompt format and label schema used during inference.