INDEX

Explanations

names of individuals who may be involved in politics or incidents referenced in the documents

oai_token-act-pair · gpt-3.5-turbo

names of individuals and key figures mentioned in the text

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

alg

-1.02

Īè

-0.97

rums

-0.92

ichick

-0.89

algia

-0.86

axy

-0.86

ovo

-0.83

choes

-0.83

atable

-0.83

leader

-0.79

POSITIVE LOGITS

 Ender

0.78

 Dexter

0.75

Fet

0.68

 Feldman

0.66

 Madd

0.66

 Parenthood

0.64

Rod

0.62

 unfocusedRange

0.62

act

0.59

 Rudolph

0.59

Activations Density 0.048%

names of individuals who may be involved in politics or incidents referenced in the documents

names of individuals and key figures mentioned in the text

No Comments

No Known Activations