INDEX

Explanations

references to the word "gum."

oai_token-act-pair · gpt-3.5-turbo

mentions of "gum" and related forms or phrases

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ourge

-0.80

 Serbia

-0.74

essen

-0.73

 breast

-0.67

asters

-0.63

 Sard

-0.63

stanbul

-0.63

æ©Ł

-0.63

isd

-0.62

KN

-0.62

POSITIVE LOGITS

ammed

2.34

gum

1.87

lly

1.79

Gum

1.73

Nou

1.01

gow

1.01

 Boyd

0.99

oru

0.97

hol

0.96

oho

0.95

Activations Density 0.050%

references to the word "gum."

mentions of "gum" and related forms or phrases

No Comments

No Known Activations