INDEX

Explanations

references to acts of vandalism in various contexts

oai_token-act-pair · gpt-3.5-turbo

language related to vandalism and related acts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

dinand

-0.87

vana

-0.75

ramid

-0.75

nesota

-0.73

obar

-0.73

amacare

-0.71

croft

-0.69

otom

-0.68

urion

-0.68

hart

-0.67

POSITIVE LOGITS

 vandalism

1.09

 graffiti

1.01

 spree

0.92

 vandal

0.87

 leve

0.77

 dere

0.74

ruction

0.70

 detract

0.69

 Frie

0.66

 Alley

0.66

Activations Density 0.026%

references to acts of vandalism in various contexts

language related to vandalism and related acts

No Comments

No Known Activations