INDEX

Explanations

references to a specific word "Angel" followed by a number (ranging from 8 to 10)

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Angel" or related variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãģ¦

-0.83

llah

-0.83

yip

-0.80

 elig

-0.77

olicy

-0.77

ĵĺ

-0.74

merce

-0.73

jri

-0.73

ongyang

-0.72

lease

-0.70

POSITIVE LOGITS

enos

1.15

eno

1.01

Angel

0.94

angel

0.94

 Angel

0.89

 Angels

0.84

inian

0.84

ique

0.83

 Gabriel

0.83

icals

0.82

Activations Density 0.025%

references to a specific word "Angel" followed by a number (ranging from 8 to 10)

instances of the word "Angel" or related variations

No Comments

No Known Activations