INDEX

Explanations

create

np_max-act · gemini-2.0-flash

The neuron strongly activates on subword tokens from the verbs “create” (in English) and “skape” (in Norwegian), i.e. terms that mean “to create” or “to build.”

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

resolved

-0.08

_md

-0.06

methodPointerType

-0.06

 Orthodox

-0.06

 underwear

-0.06

帮

-0.06

.SystemColors

-0.06

uta

-0.06

 didSet

-0.06

 helping

-0.06

POSITIVE LOGITS

 create

0.07

jax

0.07

 creating

0.07

 جست

0.07

 Wait

0.06

 storytelling

0.06

 creates

0.06

_STATS

0.06

ching

0.06

 created

0.06

Activations Density 0.033%

create

The neuron strongly activates on subword tokens from the verbs “create” (in English) and “skape” (in Norwegian), i.e. terms that mean “to create” or “to build.”

No Comments

No Known Activations