Vertical Domain LLM Fine-Tuning and Cust Diagram

Strengths

• More controllable costs and data remains within the network compared to general API solutions
• Significantly improves accuracy on vertical tasks after fine-tuning

Weaknesses

• Gap between client expectations for rapid results and actual model performance
• Heavy reliance on the open-source base ecosystem, where upstream model changes have a major impact

Opportunities

• Stricter data security regulations, increasing demand for domestic and international compliant outsourcing
• SME lack of dedicated ML teams, making them willing to outsource fine-tuning and operations

Threats

• Cloud vendors and model companies launching cheaper and easier-to-use fine-tuning tools, eroding market space
• Advancements in open-source frameworks continuously lowering the barrier for client self-fine-tuning