INDEX

Explanations

references to a particular name or entity "Oto"

oai_token-act-pair · gpt-3.5-turbo

mentions of 'oto', likely referring to some specific product, media, or brand associated with visuals or photography

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

 invaders

-0.67

nesses

-0.64

 prosecution

-0.63

 imposing

-0.63

 receptive

-0.62

drawn

-0.62

 boun

-0.61

 fair

-0.60

 incompet

-0.60

acious

-0.60

POSITIVE LOGITS

oto

1.09

zzi

1.07

oga

0.95

ypes

0.94

ÄŁ

0.92

amera

0.88

eca

0.86

veland

0.84

ython

0.83

irtual

0.82

Activations Density 0.006%

references to a particular name or entity "Oto"

mentions of 'oto', likely referring to some specific product, media, or brand associated with visuals or photography

No Comments

No Known Activations