INDEX

Explanations

names of news organizations

oai_token-act-pair · gpt-3.5-turbo

references to a specific media outlet or journalist

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

egg

-0.76

meet

-0.75

ride

-0.73

makers

-0.73

tions

-0.72

swick

-0.71

aign

-0.71

tons

-0.71

maker

-0.70

stack

-0.69

POSITIVE LOGITS

Äĩ

1.19

udi

0.99

Ã¨re

0.97

zzo

0.95

ÃŁ

0.88

plom

0.87

qi

0.87

veyard

0.85

urnal

0.85

ennes

0.82

Activations Density 0.019%

names of news organizations

references to a specific media outlet or journalist

No Comments

No Known Activations