kmseong/qwen2_5_32b_instruct-gsm8k-safeinstr-lr5e-5-ratio0.1

TEXT GENERATIONConcurrent Unit Cost:2Model Size:32.8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 29, 2026License:llama3.1Architecture:Transformer Featherless Exclusive Cold

The kmseong/qwen2_5_32b_instruct-gsm8k-safeinstr-lr5e-5-ratio0.1 model is a 32.8 billion parameter instruction-tuned causal language model based on the Qwen2.5 architecture. It is specifically fine-tuned for enhanced safety and improved performance on mathematical reasoning tasks, leveraging a unique safety alignment process. This model is optimized for use cases requiring a balance between robust safety mechanisms and strong utility in problem-solving, particularly in areas like GSM8K. Its 32768 token context length supports processing longer and more complex prompts.

Loading preview...

Model Overview

This model, kmseong/qwen2_5_32b_instruct-gsm8k-safeinstr-lr5e-5-ratio0.1, is a 32.8 billion parameter instruction-tuned language model built upon the Qwen2.5 architecture. It has been specifically fine-tuned to achieve a strong balance between safety and utility, particularly in mathematical reasoning tasks. The model incorporates a unique safety alignment process, making it suitable for applications where both robust safety and problem-solving capabilities are critical.

Key Capabilities

  • Enhanced Safety: Integrates advanced safety mechanisms to maintain refusal capabilities for harmful requests.
  • Improved Reasoning: Demonstrates strong performance on mathematical reasoning benchmarks like GSM8K.
  • Balanced Performance: Achieves a careful balance between safety and utility through a specialized fine-tuning approach.
  • Long Context: Supports a context length of 32768 tokens, enabling the processing of extensive inputs.

Good For

  • Applications requiring a large language model with strong mathematical reasoning abilities.
  • Use cases where safety and the ability to handle sensitive queries are paramount.
  • Scenarios demanding a model that can process and understand long, complex instructions or documents.