AI Engineering Core Skills — Open-Source Models and HuggingFace

Led by Yann LeCun Simulacrum

Working with open-source models through the HuggingFace ecosystem — pipelines, tokenizers, transformer internals, quantization, and running inference on GPU.

1. Module 1

   HuggingFace Ecosystem and Pipelines

   Led by Yann LeCun Simulacrum

   The question

   The HuggingFace platform (models, datasets, spaces) · HuggingFace libraries (Transformers, Datasets, Hub) · Google Colab and cloud GPUs · running Stable Diffusion and FLUX on Colab · the Pipelines API for quick inference · sentiment analysis, named e...

   Outcome

   Demonstrates engineering competence in huggingface ecosystem and pipelines.

   Sub-units

   1. ○ 1.1 HuggingFace Ecosystem and Pipelines Open this unit → ✓ ↻
   Begin the course → Print certificate of completion
2. Module 2

   Tokenizers, Transformer Internals and Quantization

   Led by Yann LeCun Simulacrum

   The question

   How tokenizers convert text to numbers (encoding and decoding) · comparing tokenizers across models (LLaMA, Phi, DeepSeek, Qwen) · chat templates and special tokens · the Transformers low-level API · quantization (8-bit, 4-bit) and its effect on mode...

   Outcome

   Demonstrates engineering competence in tokenizers, transformer internals and quantization.

   Sub-units

   1. ○ 2.2 Tokenizers, Transformer Internals and Quantization Open this unit → ✓ ↻
   Open this module → Print certificate of completion