INDEX

Explanations

the name "Kurama"

oai_token-act-pair · gpt-3.5-turbo

the occurrence of the substring "ra" in words

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

 GOODMAN

-0.90

dayName

-0.84

ĳå£«

-0.83

lace

-0.78

OW

-0.73

ensional

-0.67

Pwr

-0.66

 EntityItem

-0.66

parts

-0.65

ONSORED

-0.62

POSITIVE LOGITS

ven

1.16

uth

1.01

Ã«

0.99

ussia

0.97

hea

0.94

fter

0.94

enei

0.93

fters

0.92

ppa

0.91

ppy

0.91

Activations Density 0.014%

the name "Kurama"

the occurrence of the substring "ra" in words

No Comments

No Known Activations