INDEX

Explanations

phrases instructing to enter information such as email addresses or to re-enter information

oai_token-act-pair · gpt-3.5-turbo

commands related to input actions

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

killer

-0.69

retty

-0.68

ussy

-0.65

auld

-0.62

 killer

-0.62

estic

-0.60

ecause

-0.60

orah

-0.59

emo

-0.59

fork

-0.58

POSITIVE LOGITS

prise

1.38

prises

1.30

tainment

0.93

prising

0.80

lde

0.73

jection

0.71

taining

0.66

LECT

0.66

ipher

0.65

å£

0.65

Activations Density 0.009%

phrases instructing to enter information such as email addresses or to re-enter information

commands related to input actions

No Comments

No Known Activations