INDEX

Explanations

the word "Perry" with varying activations

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Perry."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Carbuncle

-0.75

udic

-0.74

-0.69

ravings

-0.67

azeera

-0.67

hent

-0.67

################

-0.65

atorial

-0.64

########

-0.62

 Malays

-0.62

POSITIVE LOGITS

sburg

0.93

man

0.83

ball

0.81

ns

0.80

shire

0.75

burgh

0.74

mons

0.73

bard

0.73

balls

0.70

utical

0.70

Activations Density 0.012%

the word "Perry" with varying activations

mentions of the name "Perry."

No Comments

No Known Activations