INDEX

Explanations

phrases related to legal or official authorization

oai_token-act-pair · gpt-3.5-turbo

references to authorization-related concepts or terms

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hunt

-0.78

 Rove

-0.78

 Conway

-0.77

EMS

-0.72

bart

-0.72

spl

-0.71

 Gawker

-0.71

empl

-0.71

Boh

-0.71

ÃŁ

-0.71

POSITIVE LOGITS

 authorization

1.53

 authorizing

1.22

 authorize

1.01

 Authorization

0.96

etheless

0.89

 warr

0.87

 approvals

0.86

 permission

0.86

 expiration

0.85

 permissions

0.83

Activations Density 0.006%

phrases related to legal or official authorization

references to authorization-related concepts or terms

No Comments

No Known Activations