INDEX

Explanations

terms related to passwords or security measures

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "password."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

66666666

-0.77

 Sheen

-0.73

 Taiwanese

-0.71

 Parables

-0.70

worldly

-0.70

obe

-0.67

aan

-0.65

gaard

-0.64

 Relief

-0.64

 Mostly

-0.63

POSITIVE LOGITS

 password

1.45

 passwords

1.40

password

1.24

 Password

1.22

Password

1.17

 login

1.06

 encryption

0.95

 cryptography

0.88

 authentication

0.86

 username

0.86

Activations Density 0.010%

terms related to passwords or security measures

occurrences of the word "password."

No Comments

No Known Activations