INDEX

Explanations

references to clowns, specifically those in a negative or controversial context

oai_token-act-pair · gpt-3.5-turbo

references to clowns and their comedic portrayal

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

galitarian

-0.69

WARD

-0.67

AAF

-0.66

RA

-0.65

 Ethiopian

-0.65

ECD

-0.65

equality

-0.64

ASED

-0.64

ç¥ŀ

-0.63

ãĥ´

-0.63

POSITIVE LOGITS

 clown

1.00

 Clown

1.00

fish

0.98

amn

0.79

icter

0.77

 mascot

0.77

obyl

0.77

opard

0.76

pool

0.75

oon

0.75

Activations Density 0.007%

references to clowns, specifically those in a negative or controversial context

references to clowns and their comedic portrayal

No Comments

No Known Activations