Enjoying the LLM journey so far?
I've been introducing LLM concepts so that you can understand what some terms mean and how they appear as code outputs. This is a reflective post about concepts covered in the previous chapters, namely tokenization, encoding, and decoding. Don't remember what these words mean? Have a look at the posts below:
As promised in the last post, this time we'll be looking at how to actually USE T5 models to do the following tasks: translation, summarization, and paraphrasing.
For future reference, generating answers using LLM models is called "inference."
Have a look at the Python code below!
You'll be able to see how one can:
- Import the model
- Execute a query
- See the output the model generates
- See which tasks the model can perform best.
I'm in the mood for trying something new for a change, so here's a quiz based on the code above. Read the following questions and comment down below!
Q1: What was the name of the model that I used?
Q2: What were the tasks that I had the model perform?
Q3: Which task did the model perform best at?
Ending
Hi everyone!
Are you enjoying the series so far? I've actually found all of this to be fun to present. ChatGPT can be useful and entertaining but I started to enjoy them even more once I started using and training my own models. It doesn't always work in the way that I want it to... but I guess that's part of what makes them interesting to work with. 🤣
Since it's now halfway through 2024, I'm thinking of sharing a post about neurodiversity next month. Still considering what to write about but I'll let you know when I'm comfortable sharing it. 👊
This is a great way to slow things down and help readers connect the theory to actual usage. Framing tokenization, encoding, and decoding as concepts you can recognize in real code outputs makes the learning curve much less intimidating, especially for people who are new to working directly with models.
ReplyDeleteI also like the shift into inference with a concrete T5 example. Seeing how to load a model, run a query, and evaluate which tasks it handles best is exactly the kind of hands on context that helps concepts stick. The quiz is a nice touch too. It encourages readers to actively engage with the material instead of passively skimming through code.
As more teams start experimenting with LLMs in real projects, the conversation naturally expands beyond model performance into validation, traceability, and confidence in outputs. That is where having structured processes and software test management tools becomes increasingly relevant, especially when models are used for summarization or paraphrasing in production workflows.