Module owlyshield_ransom::predictions::prediction::input_tensors [−][src]
Expand description
Contains structures to transform features computed in crate::process::ProcessRecord into input tensors.
Structs
Record of the features used to feed models’ inputs tensors. Features are the results of aggregate functions (mainly sum, max and count) applied to:
A matrix with fixed_size to feed the model’s input tensors, because too long sequences (> 1000 steps) would deserve the predictions with RNN, unless tbtt is used.
Enums
Any error in VecvecCapped.
Type Definitions
Typedef used by VecvecCapped
Our models inputs take f32 tensors, but VecvecCapped uses generics.