INDEX

Explanations

phrases related to power, authority, and discrimination

oai_token-act-pair · gpt-3.5-turbo

references to power dynamics and authority

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 recipes

-0.85

scripts

-0.85

 jackets

-0.84

 jams

-0.81

packages

-0.81

 helmets

-0.80

ptions

-0.79

 Stories

-0.78

tracks

-0.78

 beds

-0.78

POSITIVE LOGITS

 entity

1.52

 person

1.31

 institution

1.25

 perpetrator

1.23

 adversary

1.22

 corporation

1.21

 politician

1.19

 sentient

1.16

 foreigner

1.15

 cabal

1.14

Activations Density 0.334%

phrases related to power, authority, and discrimination

references to power dynamics and authority

No Comments

No Known Activations