INDEX

Explanations

logos

oai_token-act-pair · gpt-3.5-turbo

references to logos

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

MENT

-0.76

uth

-0.72

mental

-0.70

 Lank

-0.70

ouses

-0.69

ousing

-0.69

astern

-0.68

tein

-0.65

Mand

-0.64

artz

-0.64

POSITIVE LOGITS

 logo

1.33

 logos

1.22

 Logo

1.14

 Doodle

0.94

 slogan

0.93

 mascot

0.89

 emblem

0.81

 flare

0.76

 flair

0.76

 colours

0.75

Activations Density 0.012%

logos

references to logos

No Comments

No Known Activations