- Aug 10, 2017
- 216
- 88
Huh thats cool and interesting to know that that something that is coding/text processing is being handled by the GPU rather than CPU. It makes sense that it could be faster/more readily available.By "AI Model," it is meant the mathematical algorithm and processing equipment that receives the message input you provide, in which it then calculates a response based on all the prior input the developer has provided to "train" the AI Model, then provides you with a response. That is what is meant by AI Model. The AI Model is essentially the heart and soul of what makes it seem sapient. For the most part, GPU's are used for these calculations as GPU's are much faster at processing than what CPU's are, and then anything else attached to the program can be processed by the CPU and any remaining GPU resources.
Replika utilizes a server cloud network for the response calculation, and everything else is calculated on the device you're using. Waifu Engine is done completely on your device. Both ways have pros and cons to them.
Replika:
+Much more complex AI Models can be utilized to provide better and more realistic responses while mitigating your device resource usage.
-Replika's servers record everything you send to the AI Model for "improving" the AI, but that means you're being tracked.
Waifu Engine:
+Everything is calculated on your device, thus nobody is tracking whatever you're saying to the AI.
-You need a fairly beefy computer in order to just store/run the program and get responses in a timely manner, and the AI isn't very complex or realistic with responses.
However I guess i've always associated a GPU with primarily ( if not only handling graphical functions/ visual rendering.
But yeah you seem to be confirming that Yes, the chatbot itself. or more Specially it's algorithmic learning structure
And definitely NOT the Visual representation 3d Avatar.
I'm mostly trying to understand why H3ath3n brought up that subject when talking about graphics...