JunYun DeepSeek Solution

Juyun's core services include:

1. Inference service deployment Based on Amazon Web Services cloud inference service deployment, including quantized versions and non-quantized full-blood versions, hardware configuration support:
   • Large video memory deployment, high video memory requirements
   • Deploy small video memory and large general-purpose memory, and build inference services with fewer GPU resources through a balance between GPU memory and general-purpose memory.
2. Model distillation We provide a closed-loop model distillation service, that is, by constructing an evaluation system, digitally comparing model effects before and after distillation through the evaluation system, we provide high-quality distillation model delivery that meets application scenarios.
3. Model quantification Provide model quantification services and customize quantification standards according to hardware.
4. Upper-level development services According to the user scenario, help customers build upper level development services such as Q&A, number of questions, and intelligent business response, including engineering services for various prompts required in the process, and model effect evaluation services.
5. Other peripheral services Customers encounter various problems in building their own AI applications, including but not limited to: cross-language business, labeling, prompt engineering, structured information extraction, and vertical application services.