INDEX

Explanations

be

np_max-act · gemini-2.0-flash

This neuron detects auxiliary‐verb constructions—especially forms of “be” (e.g. “be used,” “can be,” “is used”) indicating passive or modal usage.

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

Negative Logits

 =================================

-0.06

.Identity

-0.06

 divisions

-0.06

bpp

-0.06

 pooled

-0.06

apikey

-0.06

 wearer

-0.05

(_:

-0.05

/boot

-0.05

Players

-0.05

POSITIVE LOGITS

�

0.07

sphere

0.07

ेक

0.07

后

0.07

>")↵

0.07

[V

0.07

 chatte

0.07

 สามารถ

0.07

参

0.07

 torino

0.06

Activations Density 0.099%

be

This neuron detects auxiliary‐verb constructions—especially forms of “be” (e.g. “be used,” “can be,” “is used”) indicating passive or modal usage.

No Known Activations

be

This neuron detects auxiliary‐verb constructions—especially forms of “be” (e.g. “be used,” “can be,” “is used”) indicating passive or modal usage.

No Known Activations