INDEX

Explanations

phrases related to specific inner attributes or components

oai_token-act-pair · gpt-3.5-turbo

references to different types of "inner" and "outer" regions or contexts

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

NPR

-0.64

KD

-0.63

KR

-0.62

 compromised

-0.62

Veh

-0.59

 unusual

-0.59

Notable

-0.59

KO

-0.59

Lt

-0.58

NRS

-0.58

POSITIVE LOGITS

inner

1.34

ners

1.16

izont

1.09

cair

0.99

glers

0.94

getic

0.93

worm

0.91

rect

0.89

anguage

0.87

worms

0.87

Activations Density 0.003%

phrases related to specific inner attributes or components

references to different types of "inner" and "outer" regions or contexts

No Comments

No Known Activations