INDEX

Explanations

names containing the pattern 'ardy' or 'andowski'

oai_token-act-pair · gpt-3.5-turbo

mentions of proper nouns, particularly names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

alach

-0.76

ropolitan

-0.76

akespe

-0.70

aeda

-0.70

uyomi

-0.68

orers

-0.67

srf

-0.66

onyms

-0.65

undai

-0.65

iv

-0.65

POSITIVE LOGITS

ardy

1.12

bage

0.83

gaard

0.74

backs

0.74

ener

0.74

keeping

0.74

kees

0.73

cipl

0.73

ce

0.72

thal

0.70

Activations Density 0.007%

names containing the pattern 'ardy' or 'andowski'

mentions of proper nouns, particularly names

No Comments

No Known Activations