Ayansk11/qwen3-4b-financial-sentiment-grpo
The Ayansk11/qwen3-4b-financial-sentiment-grpo is a 4 billion parameter Qwen3-based language model specifically fine-tuned for financial sentiment analysis. It utilizes a Chain-of-Thought (CoT) approach, trained with SFT Warm-up and Group Relative Policy Optimization (GRPO) on 8,541 financial news samples. This model excels at providing explicit reasoning alongside sentiment classification for financial texts, offering a structured output format.
Loading preview...
Overview
Ayansk11/qwen3-4b-financial-sentiment-grpo is a specialized 4 billion parameter model built upon the Qwen3 architecture. Its primary function is to perform financial sentiment analysis, distinguishing itself by generating explicit reasoning (Chain-of-Thought) alongside the sentiment classification. This model was fine-tuned using a combination of Supervised Fine-Tuning (SFT) Warm-up and Group Relative Policy Optimization (GRPO) on a dataset comprising 8,541 financial news samples, each with CoT explanations.
Key Capabilities
- Financial Sentiment Analysis: Accurately classifies financial text into positive, negative, or neutral sentiment.
- Chain-of-Thought Reasoning: Provides detailed, structured reasoning for its sentiment predictions, including analysis of key financial indicators, tone, language, and market implications.
- Optimized for Inference: Achieves 40-60 tokens/sec on Mac M4 (Q5_K_M quantization) with approximately 4 GB memory usage.
- Structured Output: Delivers sentiment and reasoning in a consistent XML-like format for easy parsing.
Good For
- Automated Financial News Analysis: Processing large volumes of financial news to extract sentiment and underlying rationale.
- Market Monitoring: Gaining insights into market sentiment based on textual data.
- Research and Development: Applications requiring explainable AI in the financial domain.
- Edge Deployment: Suitable for deployment on devices like Mac M4 due to its optimized performance and memory footprint.