INDEX

Explanations

names or titles of military personnel

oai_token-act-pair · gpt-3.5-turbo

references to military ranks, specifically Lieutenants

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

BOOK

-0.82

stakes

-0.77

fulness

-0.68

Offline

-0.68

bags

-0.67

ebook

-0.67

FORE

-0.66

worms

-0.63

Interstitial

-0.62

icles

-0.62

POSITIVE LOGITS

 colonel

1.31

utenant

1.22

 Colonel

1.10

 Commander

0.97

geant

0.92

chief

0.86

seless

0.85

Lt

0.77

 lieutenant

0.76

 commander

0.76

Activations Density 0.019%

names or titles of military personnel

references to military ranks, specifically Lieutenants

No Comments

No Known Activations