INDEX

Explanations

geographical locations, specifically focusing on the city of London

oai_token-act-pair · gpt-3.5-turbo

mentions of the city of London

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ktop

-0.82

icio

-0.81

earance

-0.80

onies

-0.80

hemy

-0.80

asta

-0.74

utation

-0.73

mble

-0.71

vit

-0.70

merce

-0.68

POSITIVE LOGITS

 Underground

1.01

mouth

0.99

bridge

0.98

 Bridge

0.89

 Borough

0.88

 Harbour

0.86

 Spit

0.82

 London

0.80

 borough

0.79

 Beckham

0.79

Activations Density 0.013%

geographical locations, specifically focusing on the city of London

mentions of the city of London

No Comments

No Known Activations