Saxo/Linkbricks-Horizon-AI-Japanese-Base-70B

Hugging Face
TEXT GENERATIONPricing:Input $2.88 / Output $2.88Concurrent Unit Cost:4Model Size:70BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Dec 9, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Warm

Saxo/Linkbricks-Horizon-AI-Japanese-Base-70B is a 70 billion parameter language model developed by Saxo, a data scientist at Linkbricks. It is a Japanese-enhanced model based on Hermes-3-Llama-3.1-70B, fine-tuned with Japanese continued pre-training, supervised fine-tuning, and DPO. The model excels in high-dimensional analysis of customer reviews and social posts, coding, writing, mathematics, and logical reasoning, with cross-language reinforcement for Japanese, Korean, Chinese, and English.

Loading preview...

Saxo/Linkbricks-Horizon-AI-Japanese-Base-70B Overview

Saxo/Linkbricks-Horizon-AI-Japanese-Base-70B is a 70 billion parameter language model developed by Saxo, a data scientist at Linkbricks. This model is a Japanese-enhanced version of the Hermes-3-Llama-3.1-70B base model, fine-tuned using approximately 10% of its parameters through Continued Pre-training (CPT), Supervised Fine-Tuning (SFT), and Direct Preference Optimization (DPO) on 8 H100-80G GPUs. It was trained on 10 million Japanese news and Wiki corpora, along with cross-learning data for Japanese, Korean, Chinese, and English, and data for mathematical and logical reasoning.

Key Capabilities

  • Japanese Language Enhancement: Significantly improved performance in Japanese through extensive fine-tuning.
  • Multilingual Cross-Reinforcement: Trained with cross-learning data for Japanese, Korean, Chinese, and English, enabling robust performance across these languages.
  • Advanced Analysis: Enhanced for high-dimensional analysis of customer reviews and social media posts.
  • Coding and Writing: Strong capabilities in code generation and creative writing tasks.
  • Mathematical and Logical Reasoning: Designed to handle complex mathematical problems and logical judgments.
  • Function Calling: Supports function calling and tool calling for enhanced integration and automation.
  • Extended Context Window: Features a 128k-context window, allowing for processing longer inputs.

Good for

  • Applications requiring strong Japanese language understanding and generation.
  • Cross-lingual tasks involving Japanese, Korean, Chinese, and English.
  • Analyzing customer feedback and social media data.
  • Developing coding assistants or automated writing tools.
  • Solving complex mathematical and logical problems.

Finetunes
1 models