INDEX

Explanations

words related to politics and government figures, especially the phrase "President-elect" and the name "Donald Trump"

oai_token-act-pair · gpt-3.5-turbo

references to the term "president-elect."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Dickinson

-0.75

 marrow

-0.66

 skim

-0.65

 REAL

-0.64

Cir

-0.64

 Fallon

-0.63

 interval

-0.63

bri

-0.63

 intervals

-0.63

 strips

-0.62

POSITIVE LOGITS

elect

1.43

turned

1.26

design

1.15

Elect

1.12

appointed

1.11

administ

1.04

CEO

1.03

level

1.01

employ

1.00

cum

1.00

Activations Density 0.062%

words related to politics and government figures, especially the phrase "President-elect" and the name "Donald Trump"

references to the term "president-elect."

No Comments

No Known Activations