INDEX

Explanations

references to GIFs, particularly those rated highly or emphasized in the text

oai_token-act-pair · gpt-3.5-turbo

references to GIFs

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

armac

-0.72

 Protestant

-0.66

quartered

-0.65

ells

-0.64

abb

-0.64

herry

-0.63

monds

-0.62

iane

-0.61

ken

-0.60

dn

-0.59

POSITIVE LOGITS

GIF

4.33

gif

2.58

gif

1.80

PNG

1.52

 JPEG

1.42

 emoji

1.34

 memes

1.13

SVG

1.10

 VIDEOS

1.07

 selfies

1.07

Activations Density 0.029%

references to GIFs, particularly those rated highly or emphasized in the text

references to GIFs

No Comments

No Known Activations