INDEX

Explanations

mentions of family members, particularly brothers

oai_token-act-pair · gpt-3.5-turbo

mentions of siblings, specifically brothers

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

Population

-0.73

argon

-0.73

ifact

-0.72

Effect

-0.70

protected

-0.69

pmwiki

-0.67

USE

-0.66

aceous

-0.62

mberg

-0.62

issions

-0.62

POSITIVE LOGITS

hood

1.32

 brother

1.22

brother

1.06

 brothers

1.05

heses

0.95

 nephew

0.94

friend

0.93

hesis

0.90

 uncle

0.88

 sibling

0.88

Activations Density 0.010%

mentions of family members, particularly brothers

mentions of siblings, specifically brothers

No Comments

No Known Activations