Hugging Face Transformers topics
12 short topics in Data Science and Machine Learning. Each one is a single focused exercise with a clear piece of evidence at the end.
12 topics, one exercise each.
Work through them in order, or jump to the one you need.
These are draft topic outlines (draft; instructor review pending). Full lessons follow after review; ask Ajay if you need a topic sooner.
- 01
Model architecture selection
Investigate model architecture selection using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 02
Pretrained weight loading
Investigate pretrained weight loading using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 03
Input preprocessing
Investigate input preprocessing using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 04
Tokenizer or feature representation
Investigate tokenizer or feature representation using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 05
Batch inference
Investigate batch inference using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 06
Device and precision configuration
Investigate device and precision configuration using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 07
Fine-tuning data organization
Investigate fine-tuning data organization using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 08
Model evaluation
Investigate model evaluation using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 09
Checkpoint save and load
Investigate checkpoint save and load using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 10
Memory and inference optimization
Investigate memory and inference optimization using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 11
Model serialization and deployment
Investigate model serialization and deployment using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course → - 12
Model licence and provenance review
Investigate model licence and provenance review using a repeatable Hugging Face Transformers experiment; compare results with a baseline and record limitations. Keep the exercise bounded and record the behavior supported by the chosen version.
Open micro-course →
More in LLM agent and retrieval frameworks.
You have the curiosity.
Let’s turn it into capability.
Start with a practical workshop or bring a specific question to Ajay.