INDEX

Explanations

words related to violent or harmful actions

oai_token-act-pair · gpt-3.5-turbo

phrases related to violent acts and crime sprees

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

 slightest

-0.74

fixed

-0.66

medium

-0.65

 Patriarch

-0.65

Medium

-0.62

pires

-0.62

acons

-0.61

rouch

-0.59

yang

-0.59

Nat

-0.59

POSITIVE LOGITS

 spree

1.21

 rampage

1.06

tir

0.88

atical

0.81

ï¸

0.80

onte

0.78

acre

0.73

 Shooter

0.72

steen

0.71

 Mania

0.71

Activations Density 0.017%

words related to violent or harmful actions

phrases related to violent acts and crime sprees

No Comments

No Known Activations