galakenya/JapanRailwayGuideWatcher

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 21, 2026Architecture:Transformer Featherless Exclusive Cold

galakenya/JapanRailwaysimLLM is a 7.9 billion parameter specialized language model, fine-tuned from Google's Gemma-2-9b-it using QLoRA. It is specifically optimized for Japanese railway technical specifications, operational control rules, and safety standards. This model excels in Retrieval-Augmented Generation (RAG) tasks, leveraging PDF documents and precise JSONL data related to Japanese railway systems, making it ideal for applications requiring deep domain-specific knowledge in this area.

Loading preview...

Overview

JapanRailwaySimLLM is a highly specialized 7.9 billion parameter language model developed by galakenya, built upon Google's Gemma-2-9b-it base model. It has been fine-tuned using QLoRA to acquire deep expertise in Japanese railway technical specifications, operational control rules, and safety standards. This model is designed for high-precision Retrieval-Augmented Generation (RAG) applications, integrating with PDF documents and detailed JSONL data covering areas like security systems, vehicle design, and timetable management.

Key Capabilities

  • Domain-Specific Expertise: Optimized for Japanese railway technology, regulations, and safety.
  • RAG Integration: Designed to work seamlessly with RAG systems, leveraging technical PDF documents for enhanced accuracy.
  • Multi-Lingual Support: Supports multi-lingual RAG search, backed by official railway manuals.
  • GGUF Compatibility: Provided in GGUF format, compatible with inference engines like llama.cpp and koboldcpp.

Good For

  • Applications requiring detailed knowledge of Japanese railway systems.
  • Building RAG systems that query technical railway documentation.
  • Research and development in railway operations and safety.
  • Generating precise information based on specialized railway data.