INDEX

Explanations

phrases related to logging into accounts or services

oai_token-act-pair · gpt-3.5-turbo

phrases related to logging into online accounts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

catentry

-0.83

alon

-0.75

ranged

-0.73

 focal

-0.68

 bruised

-0.68

 siblings

-0.68

roo

-0.66

elman

-0.66

STD

-0.64

OIL

-0.63

POSITIVE LOGITS

 Register

0.79

 Username

0.71

 Proto

0.68

 Info

0.67

 Elections

0.66

gres

0.66

 Subscribe

0.66

 login

0.66

 inbox

0.64

register

0.63

Activations Density 0.060%

phrases related to logging into accounts or services

phrases related to logging into online accounts

No Comments

No Known Activations