INDEX

Explanations

words related to things that are prohibited or restricted

oai_token-act-pair · gpt-3.5-turbo

references to social taboos and restrictions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

teness

-0.99

atche

-0.81

owing

-0.80

 Kear

-0.79

kinson

-0.77

annis

-0.77

HCR

-0.74

ä»£

-0.74

yrim

-0.73

Downloadha

-0.72

POSITIVE LOGITS

 taboo

1.26

 fetish

0.81

 weap

0.68

reth

0.65

Myth

0.64

 disse

0.61

 scrimmage

0.61

 decree

0.60

 genital

0.60

 Belt

0.60

Activations Density 0.008%

words related to things that are prohibited or restricted

references to social taboos and restrictions

No Comments

No Known Activations