INDEX

Explanations

words related to manipulation or taking advantage of a situation or person

oai_token-act-pair · gpt-3.5-turbo

instances of the word "exploit" and its variations, focusing on contexts where exploitation occurs

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

arat

-0.85

cas

-0.71

ucket

-0.70

 mint

-0.70

shake

-0.67

thood

-0.67

 nodd

-0.67

think

-0.67

prints

-0.63

semble

-0.62

POSITIVE LOGITS

 loopholes

0.96

 weaknesses

0.92

 exploited

0.91

 exploiting

0.89

oise

0.86

 exploit

0.85

oit

0.82

ulner

0.80

 vulnerabilities

0.79

CVE

0.77

Activations Density 0.021%

words related to manipulation or taking advantage of a situation or person

instances of the word "exploit" and its variations, focusing on contexts where exploitation occurs

No Comments

No Known Activations