How meet3456/llm_from_scratch works
This repository appears to be an educational collection of notebooks exploring language-model concepts, rather than an application with a demonstrated runtime workflow. The notebook names identify topics from tokenization and embeddings through attention and GPT-2 weight loading, but their contents were not provided, so their internal sequence and dependencies cannot be verified. The only sampled executable code is a standalone greeting in main.py; no links between it and the notebooks are evidenced.
Components 7
Model Foundations
- Tokenization
notebooks/tokenization.ipynb - Vector Embeddings
notebooks/vector_embeddings.ipynb - LLM Architecture
notebooks/llm_architecture.ipynb
Attention Concepts
- Attention Mechanism
notebooks/attention_mechanism_from_scratch.ipynb - Self Attention KQV
notebooks/self_attention_k_q_v.ipynb
Model Exploration
- GPT-2 Weights
notebooks/Loading_pretrained_gpt2_weights.ipynb
Other components
- Greeting Entrypoint
main.py