Inference
The process in machine learning of deploying a trained model to a production environment to perform forward computation and output predictions, focused on latency, throughput and cost efficiency, and a key dimension for the commercial viability of large models.
Timeline (last 90 days)
The name Kaffee Inferenz is a pun on the machine learning term "Inference," combined with the German word "Kaffee" (coffee).
Inference typically involves only one-way forward computation, without backpropagation or gradient updates, and focuses more on latency, throughput, and cost efficiency.
All Facts (2)
The name Kaffee Inferenz is a pun on the machine learning term "Inference," combined with the German word "Kaffee" (coffee).
50%UnverifiedInference typically involves only one-way forward computation, without backpropagation or gradient updates, and focuses more on latency, throughput, and cost efficiency.
50%