INDEX

Explanations

words related to criminal activities or law enforcement

oai_token-act-pair · gpt-3.5-turbo

references to media outlets and articles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

band

-0.74

 minds

-0.68

otin

-0.68

amic

-0.67

 recomm

-0.66

 nonviolent

-0.65

thro

-0.64

uga

-0.63

audi

-0.63

mitting

-0.62

POSITIVE LOGITS

 IMAGES

1.06

HAEL

0.95

PLIED

0.83

terday

0.82

 Images

0.77

Ù

0.76

camera

0.75

 Aerial

0.75

 srfAttach

0.74

 Whilst

0.74

Activations Density 0.034%

words related to criminal activities or law enforcement

references to media outlets and articles

No Comments

No Known Activations