INDEX

Explanations

references to individuals who have been killed in various incidents

oai_token-act-pair · gpt-3.5-turbo

instances of the word "in" related to events involving casualties or deaths

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

enium

-0.81

enn

-0.77

DragonMagazine

-0.77

boys

-0.73

uity

-0.71

ï¸ı

-0.71

Bey

-0.71

linger

-0.69

$$$$

-0.69

Favorite

-0.69

POSITIVE LOGITS

 childbirth

1.24

 infancy

1.06

 accidents

1.04

 clashes

1.04

 vain

1.01

 captivity

0.98

 action

0.97

 custody

0.96

 pursuit

0.96

 retaliation

0.94

Activations Density 0.104%

references to individuals who have been killed in various incidents

instances of the word "in" related to events involving casualties or deaths

No Comments

No Known Activations