Multimodal Image Generator
Follow the data, decisions, feedback, and validation boundaries before writing the full system.
How to read this diagram
Read left to right for the forward path: raw information becomes a representation, passes through the project’s main computational ideas, and produces an output that can be measured. Then follow the feedback path back toward the trainable or decision-making components.
VQ-VAE
A VQ-VAE maps continuous encoder outputs to the nearest learned codebook vectors. The discrete indices become image tokens; commitment and codebook updates prevent the encoder or dictionary from drifting apart.
Boundary check: document its accepted input, output shape, mutable state, failure modes, and the metric that proves this stage is correct before connecting it downstream.
Discrete tokens
A VQ-VAE maps continuous encoder outputs to the nearest learned codebook vectors. The discrete indices become image tokens; commitment and codebook updates prevent the encoder or dictionary from drifting apart.
Boundary check: document its accepted input, output shape, mutable state, failure modes, and the metric that proves this stage is correct before connecting it downstream.
Classifier-free guidance
Classifier-free guidance trains conditional and unconditional predictions in one model. Sampling extrapolates away from the unconditional prediction, increasing prompt alignment while too much guidance reduces diversity and can distort samples.
Boundary check: document its accepted input, output shape, mutable state, failure modes, and the metric that proves this stage is correct before connecting it downstream.
Architecture review checklist
- Every arrow has a documented shape, dtype, unit, or schema.
- Training and evaluation paths cannot leak information into each other.
- Randomness is seeded and captured in experiment metadata.
- Expensive stages expose timing, memory, throughput, and error metrics.
- Each feedback loop has a stop condition and a rollback strategy.
- Small reference implementations exist for numerical comparisons.