transformer_lens.components.bert_mlm_head

Hooked Transformer Bert MLM Head Component.

This module contains all the component BertMLMHead.

class transformer_lens.components.bert_mlm_head.BertMLMHead(cfg: Union[Dict, HookedTransformerConfig])

Bases: Module

Transforms BERT embeddings into logits. The purpose of this module is to predict masked tokens in a sentence.

forward(resid: Float[Tensor, 'batch pos d_model']) → Tensor

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.