INDEX

Explanations

the word "idol" or its variations

oai_token-act-pair · gpt-3.5-turbo

terms related to idols and idolization

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

-0.81

abouts

-0.74

dain

-0.72

escription

-0.68

arnaev

-0.67

lla

-0.67

xit

-0.67

cham

-0.66

llor

-0.66

cloth

-0.66

POSITIVE LOGITS

rities

0.94

utions

0.85

ized

0.84

 idol

0.83

atable

0.82

ots

0.80

izable

0.79

ovy

0.77

aters

0.77

ove

0.76

Activations Density 0.083%

the word "idol" or its variations

terms related to idols and idolization

No Comments

No Known Activations