INDEX

Explanations

religious or belief-related terms

oai_token-act-pair · gpt-3.5-turbo

words related to disbelief and skepticism

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Noon

-0.65

ĵĺ

-0.63

hire

-0.62

ĸļ

-0.60

EMS

-0.60

chnology

-0.59

 Detail

-0.58

avorite

-0.57

 eclipse

-0.56

ALE

-0.56

POSITIVE LOGITS

ievers

1.16

iever

1.14

ieving

1.12

ieve

1.10

ief

1.08

ichick

1.02

bel

0.95

inqu

0.91

iev

0.91

ayer

0.91

Activations Density 0.019%

religious or belief-related terms

words related to disbelief and skepticism

No Comments

No Known Activations