INDEX

Explanations

the word "the" appearing before numbers

oai_token-act-pair · gpt-3.5-turbo

the definite article "the" and phrases indicating totals or numerical counts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

atics

-0.71

fax

-0.70

gat

-0.67

itia

-0.65

\-

-0.65

lund

-0.64

Supported

-0.63

wd

-0.63

ftime

-0.63

ãĥĺ

-0.62

POSITIVE LOGITS

 entire

1.15

 impression

1.02

 wearer

0.99

 slightest

0.98

 whole

0.96

 entirety

0.95

 remainder

0.91

lid

0.89

 difference

0.89

 disparate

0.86

Activations Density 0.271%

the word "the" appearing before numbers

the definite article "the" and phrases indicating totals or numerical counts

No Comments

No Known Activations