INDEX

Explanations

phrases related to law enforcement and government institutions

oai_token-act-pair · gpt-3.5-turbo

references to local government and police activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

GOP

-0.96

Washington

-0.94

Defense

-0.93

 Bundy

-0.93

caliber

-0.90

 Lamar

-0.90

heimer

-0.89

Texas

-0.88

 Whitman

-0.88

avored

-0.88

POSITIVE LOGITS

 realise

1.38

 centre

1.35

 Scotland

1.33

 Â£

1.33

 realised

1.31

 recognise

1.31

 enqu

1.29

 centres

1.29

 recognised

1.26

 organising

1.25

Activations Density 0.851%

phrases related to law enforcement and government institutions

references to local government and police activities

No Comments

No Known Activations