Final Exam: Architecting LLMs for Your Technical Solutions

WHAT YOU WILL LEARN

• 

  Introduce the hugging face platform
  

  recall how tokenization works for transformer models
  

  work with the hugging face platform
  

  set up a colab notebook
  

  train a bpe tokenizer
  

  perform normalization and pre-tokenization with wordpiece
  

  recall the use of hugging face pipelines
  

  perform zero-shot classification
  

  perform sentiment analysis
  

  perform sentiment analysis with finbert
  

  perform ner with a bert model
  

  perform language translation with the t5 model
  

  summarize text using bart and t5s
• 

  evaluate text summaries using rouge scores
  

  compute text similarity with sentence transformers
  

  fine-tune a bert classifier
  

  generate predictions with a fine-tuned model
  

  load and clean text for ner
  

  align ner tags with subword tokens
  

  fine-tune bert for ner
  

  load and process data for causal language modeling (clm)
  

  fine-tune a distilgpt model for clm
  

  fine-tune the t5-small model for translation
  

  summarize text with a regular t5-small model
  

  fine-tune a t5-small model for summarization