INDEX

Explanations

numeric values relating to measurements, specifically focusing on "width" measurements

oai_token-act-pair · gpt-3.5-turbo

references to physical dimensions, particularly width

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ingo

-0.74

ischer

-0.69

alien

-0.66

SOS

-0.66

ongo

-0.65

iaries

-0.63

meet

-0.63

deb

-0.62

ictions

-0.62

atto

-0.62

POSITIVE LOGITS

 width

3.66

width

2.65

 Width

2.64

Width

2.42

 thickness

1.97

 diameter

1.86

 length

1.82

 breadth

1.77

 height

1.73

 size

1.60

Activations Density 0.022%

numeric values relating to measurements, specifically focusing on "width" measurements

references to physical dimensions, particularly width

No Comments

No Known Activations