INDEX

Explanations

the word "one" followed by a number, particularly with a high activation value indicating strong match

oai_token-act-pair · gpt-3.5-turbo

instances of the word "one."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ooks

-0.87

folk

-0.76

æĸ¹

-0.71

ãĥ¼

-0.70

lov

-0.69

Hy

-0.67

ÃĥÃĤ

-0.66

Ec

-0.66

lations

-0.66

Lie

-0.65

POSITIVE LOGITS

 hundred

1.22

 dimensional

1.01

 Hundred

1.01

 thousand

0.99

 sided

0.98

 million

0.89

 dozen

0.88

handedly

0.85

 month

0.84

 billion

0.81

Activations Density 0.105%

the word "one" followed by a number, particularly with a high activation value indicating strong match

instances of the word "one."

No Comments

No Known Activations