INDEX

Explanations

names of specific individuals with high activation values, particularly "Ernest"

oai_token-act-pair · gpt-3.5-turbo

names and terms associated with individuals and locations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cius

-0.78

 Psychiatry

-0.75

isky

-0.75

=~=~

-0.74

lette

-0.71

verage

-0.71

NB

-0.66

FF

-0.65

vin

-0.64

utions

-0.63

POSITIVE LOGITS

prises

0.88

ally

0.87

sburg

0.87

ript

0.85

eday

0.79

prise

0.78

eers

0.77

eer

0.73

ypes

0.72

eded

0.72

Activations Density 0.019%

names of specific individuals with high activation values, particularly "Ernest"

names and terms associated with individuals and locations

No Comments

No Known Activations