INDEX

Explanations

phrases related to security and security features

oai_token-act-pair · gpt-3.5-turbo

terms related to security vulnerabilities and measures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

oric

-0.89

kson

-0.82

hes

-0.76

zos

-0.65

slow

-0.63

eric

-0.62

shire

-0.62

INK

-0.62

ting

-0.61

kins

-0.61

POSITIVE LOGITS

ilege

0.80

 guards

0.79

 vulnerabilities

0.78

ailability

0.78

ulence

0.75

 clearance

0.75

 lapse

0.74

 assurance

0.74

eties

0.73

policy

0.73

Activations Density 0.024%

phrases related to security and security features

terms related to security vulnerabilities and measures

No Comments

No Known Activations