How google-deepmind/ai-foundations works
AI Research Foundations is a repository of practical labs and utilities for an eight-course language-modeling curriculum. A learner’s broad workflow is to work through labs that build from tokenization and language modeling toward neural networks, transformers, fine-tuning, and model analysis; the README documents these course stages but not a single executable end-to-end application flow. The graph includes the sampled runtime components and the documented course-feedback branch. Several course subsystems are represented only by their tree paths because their implementations were not sampled.
Main entry points: gemma.py Gemma Generation · model.py Training Model · modules.py Gemma Attention · feedback Course Feedback Labs
Components 20
Language Modeling
- BPE Tokenizer
ai_foundations/tokenization/bpe_tokenizer.py - N-gram Model
ai_foundations/ngram/model.py - Gemma Generation
ai_foundations/generation/gemma.py - Model Loading
ai_foundations/generation/loaders.py
Neural Models
- MLP Utilities
ai_foundations/machine_learning/mlp.py - Transformer Layers
ai_foundations/transformers/layers.py - Gemma Attention
ai_foundations/attention/modules.py - Attention Operations
ai_foundations/attention/attention.py
Training and Feedback
- Training Model
ai_foundations/training/model.py - Training Losses
ai_foundations/training/losses.py - Training Callbacks
ai_foundations/training/callbacks.py - Course Feedback Labs
ai_foundations/feedback - Feedback Renderer
ai_foundations/feedback/utils.py - QA Formatting
ai_foundations/formatting/format_qa.py
Analysis Tools
- Embedding Tools
ai_foundations/embeddings - Plots and Visualizations
ai_foundations/visualizations - Attention Visualization
ai_foundations/visualizations/attention_visualization.py - Plotting Utilities
ai_foundations/visualizations/plots.py - Formatting Utilities
ai_foundations/utils/formatting.py
Other components
- Learner
Connections 14
- Learner → BPE Tokenizer: works through labs
README.md - Learner → N-gram Model: works through labs
README.md - Learner → MLP Utilities: works through labs
README.md - Learner → Transformer Layers: works through labs
README.md - Learner → Course Feedback Labs: uses course labs
README.md - Model Loading → Gemma Generation: loads model
ai_foundations/generation/loaders.py - Gemma Generation → Gemma Attention: uses attention model
ai_foundations/generation/gemma.py - Gemma Attention → Attention Operations: delegates attention
ai_foundations/attention/modules.py - Training Model → Training Losses: uses losses
ai_foundations/training/model.py - Training Model → Transformer Layers: uses layers
ai_foundations/training/model.py - Training Callbacks → Gemma Generation: uses generation
ai_foundations/training/callbacks.py - Course Feedback Labs → Feedback Renderer: renders feedback
ai_foundations/feedback/course_2/embeddings/embedding_tests.py - Embedding Tools → Plots and Visualizations: analyzes embeddings
- Attention Visualization → Plotting Utilities: uses plots