INDEX

Explanations

phrases related to security or law enforcement patrols

oai_token-act-pair · gpt-3.5-turbo

references to patrol activities or operations

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 prem

-0.68

 Authors

-0.66

amaz

-0.65

Ben

-0.64

èª

-0.62

 Diss

-0.62

 Influ

-0.62

mad

-0.61

taining

-0.61

ben

-0.61

POSITIVE LOGITS

 patrol

3.75

 patrols

2.77

 patrolling

2.59

 Patrol

2.56

 troopers

1.34

 cruiser

1.34

 platoon

1.33

 escort

1.32

 reconnaissance

1.31

 trooper

1.30

Activations Density 0.014%

phrases related to security or law enforcement patrols

references to patrol activities or operations

No Comments

No Known Activations