INDEX

Explanations

terminology related to deactivation or prevention of certain functionalities or features

oai_token-act-pair · gpt-3.5-turbo

terms related to disabling features or functions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Parish

-0.71

 Rising

-0.71

 Metropolitan

-0.70

Cav

-0.66

raised

-0.66

stru

-0.66

rose

-0.65

ourced

-0.64

Growing

-0.63

 Stud

-0.63

POSITIVE LOGITS

 disable

3.27

 disabling

2.78

 Disable

2.25

Disable

2.16

disable

2.16

enable

1.37

 blacklist

1.34

 uninstall

1.33

 disarm

1.23

 disabled

1.23

Activations Density 0.029%

terminology related to deactivation or prevention of certain functionalities or features

terms related to disabling features or functions

No Comments

No Known Activations