INDEX

Explanations

profanity and strong language

oai_token-act-pair · gpt-3.5-turbo

expressions of frustration or strong negative emotions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

APD

-0.84

Population

-0.75

Ô

-0.75

Dri

-0.72

Recomm

-0.71

Econom

-0.70

Vert

-0.69

Supported

-0.69

WN

-0.68

Decre

-0.68

POSITIVE LOGITS

 hell

1.02

ishly

0.96

fuck

0.93

enic

0.93

bender

0.89

fire

0.88

hole

0.87

ibur

0.86

spawn

0.82

oise

0.79

Activations Density 0.007%

profanity and strong language

expressions of frustration or strong negative emotions

No Comments

No Known Activations