INDEX

Explanations

words related to conflict or disagreement

oai_token-act-pair · gpt-3.5-turbo

terms related to affirmation or support

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

externalActionCode

-0.84

DragonMagazine

-0.79

assetsadobe

-0.70

BOOK

-0.69

BuyableInstoreAndOnline

-0.69

SpaceEngineers

-0.68

OPLE

-0.66

ppo

-0.66

STD

-0.64

GER

-0.64

POSITIVE LOGITS

irming

1.15

luence

1.08

ront

1.08

liction

1.03

aff

1.00

irmed

0.97

semb

0.97

ording

0.97

irms

0.92

onso

0.90

Activations Density 0.013%

words related to conflict or disagreement

terms related to affirmation or support

No Comments

No Known Activations