INDEX

Explanations

then

np_max-act · gemini-2.0-flash

The neuron flags common instructional cue words—especially step‐sequencing and imperative verbs like “then,” “use,” “need,” “be,” and “to” in how-to or procedural text.

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

ビ

-0.07

Writer

-0.07

 BIOS

-0.07

growth

-0.06

CSV

-0.06

flu

-0.06

Ald

-0.06

ocabulary

-0.06

loub

-0.06

ΜΠ

-0.06

POSITIVE LOGITS

 remain

0.09

.failed

0.07

 невозможно

0.07

 afin

0.06

，便

0.06

entic

0.06

 notch

0.06

 เข

0.06

 فقط

0.06

.<

0.06

Activations Density 0.027%

then

The neuron flags common instructional cue words—especially step‐sequencing and imperative verbs like “then,” “use,” “need,” “be,” and “to” in how-to or procedural text.

No Comments

No Known Activations

then

The neuron flags common instructional cue words—especially step‐sequencing and imperative verbs like “then,” “use,” “need,” “be,” and “to” in how-to or procedural text.

No Comments

No Known Activations