INDEX

Explanations

phrases related to system vulnerabilities and security issues

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Verse

-0.56

pic

-0.56

>>

-0.56

 Frenchman

-0.54

Redditor

-0.52

eson

-0.52

 nutshell

-0.51

Sir

-0.50

 Statue

-0.49

 Nasa

-0.49

POSITIVE LOGITS

 prematurely

1.10

 unnecessarily

1.07

 sooner

0.99

 unexpectedly

0.97

 inappropriately

0.95

 unintentionally

0.93

 voluntarily

0.91

 excessively

0.89

 willingly

0.89

 because

0.89

Activations Density 0.632%

phrases related to system vulnerabilities and security issues

No Comments

No Known Activations