INDEX

Explanations

start or finish

np_max-act · gemini-2.0-flash

This neuron fires on words that denote downward movement or the lowest/end position (e.g. “down,” “bottom,” “end”).

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

 pensions

-0.06

.hits

-0.06

 метал

-0.06

 territorial

-0.06

obj

-0.06

 readability

-0.06

 Magnum

-0.06

 concessions

-0.06

anything

-0.06

avadoc

-0.05

POSITIVE LOGITS

(true

0.09

ského

0.08

 '''
↵

0.07

MG

0.07

:list

0.07

 سو

0.07

IOUS

0.07

ILLED

0.07

CRE

0.07

_AR

0.07

Activations Density 0.024%

start or finish

This neuron fires on words that denote downward movement or the lowest/end position (e.g. “down,” “bottom,” “end”).

No Comments

No Known Activations