INDEX

Explanations

mentions or references to being in the spotlight or receiving attention, especially in the media or public eye

oai_token-act-pair · gpt-3.5-turbo

references to the term 'spotlight.'

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

meal

-0.75

lements

-0.68

aves

-0.68

tein

-0.68

ihad

-0.63

forms

-0.63

reek

-0.62

remlin

-0.62

neg

-0.62

ippi

-0.61

POSITIVE LOGITS

 spotlight

1.41

elight

1.21

 shone

0.90

 Spotlight

0.82

 shines

0.81

 shine

0.80

wcs

0.79

 showc

0.74

lights

0.71

shine

0.70

Activations Density 0.007%

mentions or references to being in the spotlight or receiving attention, especially in the media or public eye

references to the term 'spotlight.'

No Comments

No Known Activations