INDEX

Explanations

mentions of the word "Afghan"

oai_token-act-pair · gpt-3.5-turbo

references to Afghanistan

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

creen

-1.04

tower

-0.84

ayers

-0.81

ometimes

-0.79

vous

-0.74

yss

-0.74

hare

-0.73

ynt

-0.73

enment

-0.72

ection

-0.72

POSITIVE LOGITS

istan

1.26

istani

1.26

 Afghan

1.18

 Afghans

1.06

 Afgh

1.03

 Taliban

1.00

 Kabul

0.99

ghan

0.88

 Afghanistan

0.87

 civilians

0.81

Activations Density 0.017%

mentions of the word "Afghan"

references to Afghanistan

No Comments

No Known Activations