capreolus.reranker.TFBERTMaxP
¶
Module Contents¶
Classes¶
TFElectraRelevanceHead |
BERT-style ClassificationHead (i.e., out_proj only – no dense). See transformers.TFElectraClassificationHead |
TFBERTMaxP_Class |
|
TFBERTMaxP |
Base class for Reranker modules. The purpose of a Reranker is to predict relevance scores for input documents. Rerankers are generally supervised methods implemented in PyTorch or TensorFlow. |
-
class
capreolus.reranker.TFBERTMaxP.
TFElectraRelevanceHead
(dropout, out_proj, *args, **kwargs)[source]¶ Bases:
tensorflow.keras.layers.Layer
BERT-style ClassificationHead (i.e., out_proj only – no dense). See transformers.TFElectraClassificationHead
-
class
capreolus.reranker.TFBERTMaxP.
TFBERTMaxP_Class
(extractor, config, *args, **kwargs)[source]¶ Bases:
tensorflow.keras.layers.Layer
-
class
capreolus.reranker.TFBERTMaxP.
TFBERTMaxP
(config=None, provide=None, share_dependency_objects=False, build=True)[source]¶ Bases:
capreolus.reranker.Reranker
Base class for Reranker modules. The purpose of a Reranker is to predict relevance scores for input documents. Rerankers are generally supervised methods implemented in PyTorch or TensorFlow.
- Modules should provide:
- a
build_model
method that initializes the model used - a
score
and atest
method that take a representation created by anExtractor
module as input and return document scores - a
load_weights
and asave_weights
method, if the base class’ PyTorch methods cannot be used
- a