INDEX

Explanations

words related to positions or locations within physical objects

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

humans

-0.71

SHARE

-0.69

ebook

-0.67

atic

-0.65

OTA

-0.63

ãĤ¼

-0.62

lessly

-0.62

aughs

-0.61

ulia

-0.60

conom

-0.59

POSITIVE LOGITS

 bottom

1.60

top

1.33

end

1.29

 center

1.29

bottom

1.25

 sides

1.25

 ends

1.22

 upper

1.22

 centre

1.22

 underside

1.21

Activations Density 0.202%

words related to positions or locations within physical objects

No Comments

No Known Activations