INDEX

Explanations

phrases related to unethical or morally questionable activities

oai_token-act-pair · gpt-3.5-turbo

references to "dirty" related concepts or topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

HCR

-1.04

*/(

-1.02

iculture

-0.86

ajor

-0.85

icultural

-0.84

ļéĨĴ

-0.80

ISC

-0.80

iphate

-0.77

ĸļ

-0.75

imus

-0.75

POSITIVE LOGITS

 tricks

1.08

 laundry

1.06

 linen

0.94

 dirty

0.87

 diapers

0.81

 diaper

0.79

 trick

0.78

 filthy

0.76

 little

0.74

 deeds

0.74

Activations Density 0.037%

phrases related to unethical or morally questionable activities

references to "dirty" related concepts or topics

No Comments

No Known Activations