INDEX

Explanations

the prefix "im-" followed by a word

oai_token-act-pair · gpt-3.5-turbo

instances of the word "im" as a prefix or in various contexts

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

NetMessage

-0.83

 Whitman

-0.69

yards

-0.68

 Hitch

-0.67

 Bever

-0.67

 Sack

-0.66

OPLE

-0.66

 slash

-0.65

 Rudd

-0.64

 Colleges

-0.63

POSITIVE LOGITS

im

1.02

balanced

0.99

unity

0.98

medi

0.95

imi

0.95

itated

0.94

itates

0.94

ugi

0.93

mediate

0.90

manent

0.89

Activations Density 0.008%

the prefix "im-" followed by a word

instances of the word "im" as a prefix or in various contexts

No Comments

No Known Activations