Technology and operation · Question 54 of 60
A company wants to train a large model entirely on-premises. What are the consequences?
1 answer is correct. Hard
Show answer
Answer
- D It needs high-performance on-premises server clusters with costs for energy, cooling and hardware
Explanation
Developing locally keeps full control and keeps the data in-house. A normal PC is enough for small models such as decision trees, and medium ones need dedicated GPUs. Large models require powerful server clusters in your own data center, and then power, cooling and hardware add up considerably.