INDEX

Explanations

words related to enclosures or spaces that provide privacy or protection

oai_token-act-pair · gpt-3.5-turbo

words associated with closeness and proximity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Barnett

-0.88

Ø©

-0.78

 Uzbek

-0.78

 Columbia

-0.74

 Curve

-0.73

 Commonwealth

-0.71

 Dome

-0.70

 Wonderland

-0.70

 Settlement

-0.67

 Falcon

-0.66

POSITIVE LOGITS

eness

1.23

ety

1.12

uers

1.03

 clos

1.02

itude

1.01

itiz

0.99

elines

0.97

ings

0.95

estones

0.94

semb

0.92

Activations Density 0.016%

words related to enclosures or spaces that provide privacy or protection

words associated with closeness and proximity

No Comments

No Known Activations