INDEX

Explanations

references to the word 'guard'

oai_token-act-pair · gpt-3.5-turbo

references to guardians or protective figures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ciplinary

-0.67

pling

-0.65

GY

-0.62

 Coch

-0.62

BLE

-0.60

pled

-0.60

 unified

-0.59

Gou

-0.57

nec

-0.57

usha

-0.56

POSITIVE LOGITS

guard

1.36

Guard

1.24

ians

1.20

guards

1.12

ian

1.05

iola

1.00

 Guard

0.86

maid

0.83

aspers

0.82

arians

0.81

Activations Density 0.004%

references to the word 'guard'

references to guardians or protective figures

No Comments

No Known Activations