INDEX

Explanations

adjectives with a positive connotation indicating superiority or exceptionalism

oai_token-act-pair · gpt-3.5-turbo

the term "super" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

edIn

-0.77

 Seym

-0.70

 Lauder

-0.65

Reloaded

-0.64

 unborn

-0.64

 Frie

-0.63

 Greenwald

-0.59

 Anger

-0.59

 Franks

-0.58

 Transcript

-0.57

POSITIVE LOGITS

imposed

1.42

nova

1.32

visory

1.28

visor

1.23

powers

1.19

charged

1.17

cedes

1.16

charg

1.15

computer

1.08

cell

1.06

Activations Density 0.025%

adjectives with a positive connotation indicating superiority or exceptionalism

the term "super" in various contexts

No Comments

No Known Activations