Saxo/Linkbricks-Horizon-AI-Korean-Advanced-8B-COT-boost

Hugging Face
TEXT GENERATIONPricing:Input $0.2 / Cached $0.028 / Output $0.32Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 27, 2024License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Warm

Saxo/Linkbricks-Horizon-AI-Korean-Advanced-8B-COT-boost is an 8 billion parameter Korean language model developed by Yunsung Ji (Saxo) at Linkbricks. Fine-tuned from Saxo/Linkbricks-Horizon-AI-Nous-Hermes-3-Llama3.1-Korean-cpt-8b, it features a 128k context window and is specifically enhanced for Chain of Thought (COT) reasoning, mathematics, and complex logic problems. This model excels in cross-lingual tasks involving Korean, Chinese, English, and Japanese, and supports Korean Function Call and Tool Calling.

Loading preview...

Model Overview

Saxo/Linkbricks-Horizon-AI-Korean-Advanced-8B-COT-boost is an 8 billion parameter Korean language model developed by Yunsung Ji (Saxo), a data scientist at Linkbricks. This model is a result of SFT (Supervised Fine-Tuning) and DPO (Direct Preference Optimization) training on a base model, utilizing 8 H100-80G GPUs.

Key Capabilities and Enhancements

  • Multilingual Proficiency: Trained with cross-lingual data for Korean, Chinese, English, and Japanese, enabling robust performance across these languages.
  • Advanced Reasoning: Significantly enhanced for Chain of Thought (COT) reasoning, making it adept at handling complex logical problems and mathematical tasks.
  • Extended Context Window: Features a substantial 128k context window, allowing for processing longer inputs and maintaining coherence over extended dialogues.
  • Function and Tool Calling: Provides native support for Korean Function Call and Tool Calling, facilitating integration with external tools and APIs.
  • Training Methodology: Utilizes advanced training techniques including Deepspeed Stage=3, rslora, and BAdam Layer Mode for efficient and effective fine-tuning.

Ideal Use Cases

This model is particularly well-suited for applications requiring:

  • Complex problem-solving in Korean, especially those involving mathematical reasoning or intricate logic.
  • Multilingual interactions and content generation across Korean, Chinese, English, and Japanese.
  • Applications that benefit from long context understanding and generation.
  • Integration with external systems via function and tool calling in a Korean context.

Finetunes
4 models