mazafard/esco-gemma4

Overview

esco-gemma4 is a specialized 7.9 billion parameter language model, fine-tuned from Google's Gemma 4 architecture by mazafard. Its core function is to accurately classify professional skills, job descriptions, and qualifications against the ESCO v1.2.1 (European Skills, Competences, Qualifications and Occupations) taxonomies and ISCO-08 unit codes. This model is particularly useful for applications requiring standardized occupational mapping within the European context.

Key Capabilities

• Precise Occupational Mapping: Maps diverse professional inputs to specific ESCO occupation titles and ISCO-08 codes.
• Standardized Output: Designed to produce deterministic and structured classification results, including the ESCO Occupation Title and ISCO-08 Code.
• Alpaca Prompt Template: Utilizes the standardized Alpaca prompt format for optimal input processing and classification.
• Multi-Format Availability: Provided in 16-bit Safetensors for vLLM and transformers, GGUF for Ollama and llama.cpp, and MLX for native Apple Silicon performance.

Training Details

The model was fine-tuned using a 4-bit QLoRA approach with the Unsloth engine. The training dataset was the official European Commission ESCO Taxonomy v1.2.1, ensuring its specialization in European labor market standards.

Good For

• HR Tech: Automating the classification of resumes, job applications, and job postings.
• Labor Market Analysis: Standardizing skill and occupation data for research and policy development.
• Career Guidance: Assisting individuals in understanding how their skills align with recognized occupations.
• Educational Platforms: Mapping course content and learning outcomes to industry-standard qualifications.