INDEX

Explanations

words related to negative biases or unfair treatment based on preconceived notions or opinions

oai_token-act-pair · gpt-3.5-turbo

references to prejudice and bias, particularly in relation to social issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Interstitial

-0.84

VID

-0.78

zech

-0.76

adra

-0.73

ascus

-0.72

sis

-0.72

avez

-0.71

ODE

-0.69

sol

-0.69

ctic

-0.69

POSITIVE LOGITS

 prejudice

1.38

 prejud

1.24

 prejudices

1.08

icial

0.82

 bigot

0.82

 intolerance

0.80

 hatred

0.78

 bigotry

0.78

eering

0.78

ophobic

0.78

Activations Density 0.009%

words related to negative biases or unfair treatment based on preconceived notions or opinions

references to prejudice and bias, particularly in relation to social issues

No Comments

No Known Activations