Purpose & architecture
RetNet is local Generator Core registry configuration 84, created by NewRetNet in the transformer package. Its exact constructor and configuration settings are preserved below.
The shared implementation reshapes input into a sequence, adds sinusoidal positional embeddings, applies attention and feed-forward operations with residuals and layer normalization, then flattens the output. Train returns mean squared error against the target.
The registry labels cover architectures with substantially different published designs, while these constructors share one local implementation and configuration interface. Pretrained weights, published task behavior and architecture equivalence were not recovered or verified. Train does not update weights.
This source generation
This record follows the claude/improve-math-logic-ehIW1 branch of Janady13/complete-agi-core at commit 33aa84f2aa61cedbe010240ada7f4efbdfe795a7. The snapshot identifies the source inspected in this audit. It is not a statement about which revision is currently running on a device.
| Field | Source record |
|---|---|
| Revision | 33aa84f2aa61cedbe010240ada7f4efbdfe795a7 |
| Branch | claude/improve-math-logic-ehIW1 |
| Recorded generation | Generator Core · transformer |
Recovered implementation structure
These paths locate the components and entry points represented by this build. The source register contains their inspected revisions.
- generator_core/pkg/generators/transformer/transformer.go
Behavior & verification limits
The registry labels cover architectures with substantially different published designs, while these constructors share one local implementation and configuration interface. Pretrained weights, published task behavior and architecture equivalence were not recovered or verified. Train does not update weights.
Original project documentation
The following documentation is preserved as a dated source account. Claims in it retain the source’s wording and must be read alongside the implementation notes above.
func NewRetNet() *TransformerGenerator {
g := NewTransformerGenerator(84, "RetNet", 8, 12, 2048)
g.Config().Params["retention"] = true
g.Config().Params["chunkwise"] = true
return g
}Equations in this build
1 linked mathematical records. Each opens its notation, interpretation and source.