Learn Before
Different standards for tokenization
- Word tokenization: Penn Treebank tokenization; NLTK
- Character tokenization
- Subword tokenization: byte-pair encoding(BPE); wordpiece algorithm with MaxMatch decoding; SentencePiece
0
1
Contributors are:
Who are from:
Tags
Data Science
D2L
Dive into Deep Learning @ D2L
Related
Different standards for tokenization
Inference Process with a Fine-Tuned Model
Example of Tokenization into Words and Punctuation
Example of Word and Punctuation Tokenization
Methods of Tokenization
A language model is given the sentence: 'The researcher is studying neuroplasticity.' It processes the sentence using two different methods, resulting in two different sequences of tokens.
Method A:
['The', 'researcher', 'is', 'studying', 'neuroplasticity', '.']Method B:['The', 'researcher', 'is', 'study', 'ing', 'neuro', 'plasticity', '.']Assuming the model has never encountered the word 'neuroplasticity' during its training but has seen words like 'neuroscience' and 'plasticity' separately, which method is more advantageous for helping the model understand the new word, and why?
Tokenization Strategies
Evaluating Tokenization for a Specialized Chatbot