SeongryongJung/qwen3-8b-chemistry-rlsd-ema005

TEXT GENERATIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 2, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

SeongryongJung/qwen3-8b-chemistry-rlsd-ema005 is an 8 billion parameter Qwen3-based causal language model fine-tuned by SeongryongJung with RLSD (EMA 0.05) specifically on a chemistry dataset. This model, with a 32768 token context length, is optimized for chemistry-related tasks, achieving a validation performance of 74.38% on the sciknoweval reward metric. Its primary strength lies in specialized scientific reasoning within the chemistry domain.

Loading preview...

Model Overview

This model, qwen3-8b-chemistry-rlsd-ema005, is an 8 billion parameter language model derived from Qwen/Qwen3-8B. It has been fine-tuned by SeongryongJung using the RLSD (EMA 0.05) method, specifically targeting the chemistry split of a dataset. The model is designed to excel in tasks related to chemistry, leveraging its specialized training.

Key Capabilities

  • Chemistry-Specific Performance: Achieves a val-aux/sciknoweval/reward/mean@16 score of 74.38% after 100 training steps, indicating strong performance in chemistry-related evaluations.
  • RLSD Fine-tuning: Utilizes Reinforcement Learning from Scientific Data (RLSD) with an Exponential Moving Average (EMA) of 0.05 for enhanced domain adaptation.
  • Qwen3 Architecture: Built upon the robust Qwen3-8B base model, providing a solid foundation for its specialized capabilities.

Good For

  • Chemistry Research: Ideal for applications requiring deep understanding and generation of chemistry-specific content.
  • Scientific Text Processing: Suitable for tasks involving analysis, summarization, or question-answering within the chemistry domain.
  • Specialized AI Development: Useful for developers building AI tools and services focused on scientific disciplines, particularly chemistry.