INDEX

Explanations

phrases related to using tools or performing actions

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Career

-0.91

Politics

-0.88

 Events

-0.85

 Carib

-0.78

isSpecialOrderable

-0.78

 Communities

-0.77

ylum

-0.76

Sov

-0.74

 Ukrain

-0.74

 Equity

-0.74

POSITIVE LOGITS

 gently

1.33

 shove

1.30

 submer

1.27

 slit

1.22

 scrape

1.21

 chop

1.20

 peel

1.20

 spray

1.20

 remove

1.19

 sprinkle

1.18

Activations Density 0.419%

phrases related to using tools or performing actions

No Comments

No Known Activations