INDEX

Explanations

references to legal and political activities, such as seeking approval from Congress, creating statutes, and signing off on plans

oai_token-act-pair · gpt-3.5-turbo

references to legislative and governmental processes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

iour

-0.84

 flavours

-0.81

 Sask

-0.81

Ont

-0.76

 neighbours

-0.73

 toget

-0.73

 prolet

-0.72

 humour

-0.71

xual

-0.69

Picture

-0.69

POSITIVE LOGITS

 Congressional

1.35

 congressional

1.29

 Authorization

1.29

 appropriations

1.28

 declass

1.26

DOD

1.25

 authorization

1.23

 Pentagon

1.22

 FISA

1.21

 Appropriations

1.20

Activations Density 0.648%

references to legal and political activities, such as seeking approval from Congress, creating statutes, and signing off on plans

references to legislative and governmental processes

No Comments

No Known Activations