INDEX

Explanations

names of specific individuals

oai_token-act-pair · gpt-3.5-turbo

mentions of specific individuals, particularly those related to political contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

anim

-0.86

 Animation

-0.84

cius

-0.79

gaard

-0.76

 Anim

-0.75

Japan

-0.72

Ky

-0.70

 animation

-0.69

Åį

-0.68

 Guards

-0.68

POSITIVE LOGITS

 Donna

2.30

 Debbie

1.75

 Podesta

1.74

 Braz

1.64

DNC

1.56

 Wasserman

1.54

Ðº

1.43

 Chevy

1.33

 Imran

1.26

 Betty

1.23

Activations Density 0.034%

names of specific individuals

mentions of specific individuals, particularly those related to political contexts

No Comments

No Known Activations