Update 5.mdx

chapters/en/chapter1/5.mdx

@@ -68,7 +68,7 @@ Text generation involves creating coherent and contextually relevant text based
 
 1. GPT-2 uses [byte pair encoding (BPE)](https://huggingface.co/docs/transformers/tokenizer_summary#bytepair-encoding-bpe) to tokenize words and generate a token embedding. Positional encodings are added to the token embeddings to indicate the position of each token in the sequence. The input embeddings are passed through multiple decoder blocks to output some final hidden state. Within each decoder block, GPT-2 uses a *masked self-attention* layer which means GPT-2 can't attend to future tokens. It is only allowed to attend to tokens on the left. This is different from BERT's [`mask`] token because, in masked self-attention, an attention mask is used to set the score to `0` for future tokens.
 
-2. The output from the decoder is passed to a language modeling head, which performs a linear transformation to convert the hidden states into logits. The label is the next token in the sequence, which are created by shifting the logits to the right by one. The cross-entropy loss is calculated between the shifted logits and the labels to output the next most likely token.
+2. The output from the decoder is passed to a language modeling head, which performs a linear transformation to convert the hidden states into logits. The label is the next token in the sequence, which is created by shifting the logits to the right by one. The cross-entropy loss is calculated between the shifted logits and the labels to output the next most likely token.
 
 GPT-2's pretraining objective is based entirely on [causal language modeling](https://huggingface.co/docs/transformers/glossary#causal-language-modeling), predicting the next word in a sequence. This makes GPT-2 especially good at tasks that involve generating text.