Skip to contents

A formula marker interpreted by llm_probe(). Calling it directly is an error.

Usage

activations(layer, component = "residual")

Arguments

layer

Distinct positive 1-based layer indices.

component

One of "residual", "attn_out" or "mlp_out".

Value

No standalone value; identifies activation slices inside a probe formula.

Examples

label ~ activations(layer = 1:3, component = "residual")
#> label ~ activations(layer = 1:3, component = "residual")
#> <environment: 0x556b62dfa200>