Ranknet

Class for the ranknet implementation.

class src.models.ranknet.RankNet(num_features)[source]

Bases: torch.nn.modules.module.Module

A class to create pair wise RankNet models.

num_features

forward(input1, input2)[source]

predict_proba(input_)[source]

forward(input1, input2)[source]

.

Parameters

() (input2) – Document 1 features
() – Document 2 features

Returns

pairwise ranking

Return type

prob ()

predict_proba(input_)[source]

.

Parameters: () (input) –
Returns: pairwise ranking score
Return type: confidence ()

training: bool