When fine-tuning large language models with Second, selecting the right base model is crucial for achieving optimal results. This guide will help you understand the available models and choose the one that best fits your specific use case.

​
Available Models

Second offers a variety of large language models for fine-tuning. Each model has its own strengths and characteristics:

Llama 3.1 (8B)

Open-source large language model by Meta AI, known for its versatility and strong performance across various tasks.

Mistral 7B

Powerful open-source model that excels in various tasks, including code and reasoning.

Gemma 2 (9B)

Lightweight, open-source model from Google. Versatile for various text generation tasks.

GPT-4o mini

Cost-efficient model optimized for speed and performance. Excels in various tasks while maintaining high accuracy.

Consider each model’s unique characteristics, performance metrics, and language support when selecting a base model for fine-tuning. Ensure the chosen model aligns with your specific requirements and target languages.

​
Expert Guidance

For clients planning fine-tuning projects with budgets exceeding $150, Second offers personalized expert guidance. Our team can provide:

  • In-depth model analysis tailored to your specific use case
  • Custom performance benchmarking
  • Optimization strategies for cost-effective fine-tuning
  • Recommendations for scaling and deployment

To access this specialized service, please contact our support team at support@usesecond.com with details about your project scope and requirements.

Leveraging expert advice can significantly enhance the efficiency and effectiveness of your fine-tuning efforts, especially for complex or high-impact projects.