INDEX

Explanations

the word "any" followed by an adjective or a verb

oai_token-act-pair · gpt-3.5-turbo

phrases related to universally accepted judgments or assessments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

endas

-0.94

redients

-0.72

ributes

-0.71

appings

-0.71

combe

-0.70

oute

-0.70

rex

-0.69

icz

-0.66

rea

-0.66

ros

-0.66

POSITIVE LOGITS

THING

1.23

 conceivable

1.10

 given

1.03

 imaginable

1.01

 semblance

0.92

 sane

0.90

body

0.88

how

0.87

 kind

0.86

ONE

0.86

Activations Density 0.065%

the word "any" followed by an adjective or a verb

phrases related to universally accepted judgments or assessments

No Comments

No Known Activations