Small Language Models (SML) for the Win

Tim Spann
Cloudera
Published in
2 min readMay 2, 2024

--

I don’t need a model that knows a little bit about a lot of things up to last year. I need a model that knows everything about Apache NiFi or Python programming or how Bitcoin works. Not only are these trained on just the problem space, but they can run faster and on smaller hardware. We can usually run on smaller, cheaper machines with simple or no GPUs with just CPU.

We can build our own or look out in the open like in HuggingFace for models trained to our use case.

This model was trained on Stock News.

This model is trained on diseases.

This one is trained on IMDB movie reviews for sentiment.

This one was trained on Reddit (oh no), to find NSFW text.

RESOURCES

--

--

Tim Spann
Cloudera

Principal Developer Advocate, Zilliz. Milvus, Attu, Towhee, GenAI, Big Data, IoT, Deep Learning, Streaming, Machine Learning. https://www.datainmotion.dev/