INDEX

Explanations

mentions of siblings, specifically sisters

oai_token-act-pair · gpt-3.5-turbo

references to familial relationships, specifically sisters

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

protected

-0.72

-0.71

ered

-0.70

veyard

-0.70

ustomed

-0.67

atility

-0.67

urbed

-0.65

intensity

-0.64

ech

-0.64

oS

-0.63

POSITIVE LOGITS

 sister

1.16

hood

0.99

 brother

0.94

 sisters

0.87

hesis

0.85

 aunt

0.85

heses

0.84

 cousin

0.81

wife

0.81

 daughter

0.81

Activations Density 0.004%

mentions of siblings, specifically sisters

references to familial relationships, specifically sisters

No Comments

No Known Activations