Neural-network building blocks · MLD-1173
Stop-gradient
Preventing derivatives from flowing through a selected computation path.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Treat a target-network output as fixed.
Original caveat
It changes learning behavior without necessarily changing forward values.
Context and aliases
Level: Intermediate. Stages: Training · Inference · Theory.
Tags: DL · SLM · LLM · Vision · NLP.
Aliases: Detach.
Related concepts
Original references
- Deep Learning: online textbook ↗ · Consulted
- PyTorch neural-network layers ↗ · Reference link
- PyTorch documentation ↗ · Reference link
- Original reference URL ↗