INDEX

Explanations

references to malware or malicious activity

oai_token-act-pair · gpt-3.5-turbo

words or phrases related to "mal" or characteristics associated with malice or wrongdoing

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ERC

-0.85

VERTISEMENT

-0.72

=-=-=-=-=-=-=-=-

-0.68

 SERV

-0.67

RSS

-0.67

 Pound

-0.63

HOME

-0.63

 Caption

-0.63

 Scion

-0.63

ICLE

-0.62

POSITIVE LOGITS

mal

1.17

icious

0.97

colm

0.96

ignant

0.93

arial

0.89

ware

0.88

akia

0.86

acia

0.78

 Nadu

0.78

adjust

0.77

Activations Density 0.009%

references to malware or malicious activity

words or phrases related to "mal" or characteristics associated with malice or wrongdoing

No Comments

No Known Activations