INDEX

Explanations

references to airplanes or aircraft

oai_token-act-pair · gpt-3.5-turbo

mentions of airplanes or planes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

laus

-0.96

PUT

-0.84

 Cumber

-0.79

FINE

-0.76

essee

-0.75

é»Ĵ

-0.73

UGC

-0.73

OFF

-0.69

é¾įå

-0.69

ä¼

-0.66

POSITIVE LOGITS

 airplanes

1.06

walker

1.05

walk

1.05

 planes

1.04

plane

1.04

 plane

1.03

 airliner

1.02

planes

1.01

 hangar

0.99

 prope

0.97

Activations Density 0.021%

references to airplanes or aircraft

mentions of airplanes or planes

No Comments

No Known Activations