Why do we need large models? Based on my experience working with visual AI companies, here's why:
Simple scenarios with traditional algorithms can be very effective.
If the input is complex, traditional algorithms cannot handle both preceding and subsequent connections and interference.
So, a simple neural network was used as a filter at the front end, which proved to be quite effective.
Using large models for CW applications allows this system to infer potential information from local audio, which may be more aligned with its specialized function.
闽ICP备2021006864号-7
闽公网安备35020602002794号
互联网举报中心