INDEX

Explanations

mentions of smoking-related terms

oai_token-act-pair · gpt-3.5-turbo

instances of the word "smoke" or its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

DragonMagazine

-0.76

 Cerberus

-0.71

RTX

-0.71

ITAL

-0.70

 Highlander

-0.67

heit

-0.66

wright

-0.64

EPS

-0.62

 Vanguard

-0.61

TAIN

-0.61

POSITIVE LOGITS

ooth

1.26

ugg

1.20

oky

1.20

oke

1.10

udge

1.08

itten

1.07

iley

1.07

okers

1.06

oking

1.06

oked

1.03

Activations Density 0.006%

mentions of smoking-related terms

instances of the word "smoke" or its variations

No Comments

No Known Activations