INDEX

Explanations

the word "bur" at various intensities

oai_token-act-pair · gpt-3.5-turbo

references to burial or interment

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

uncture

-0.81

 SYSTEM

-0.72

 Spac

-0.67

 partName

-0.67

require

-0.64

CFR

-0.63

 Physicians

-0.62

 Prelude

-0.61

 Mandatory

-0.61

 CODE

-0.61

POSITIVE LOGITS

bur

1.30

querque

1.21

idays

1.02

ger

0.99

rito

0.97

undy

0.87

thur

0.84

riers

0.81

rowing

0.81

ertodd

0.81

Activations Density 0.004%

the word "bur" at various intensities

references to burial or interment

No Comments

No Known Activations