INDEX

Explanations

names or proper nouns related to "junkie"

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 MacArthur

-0.70

 bombard

-0.68

 outgoing

-0.65

pen

-0.62

 constituent

-0.62

 Woody

-0.61

 Lauder

-0.60

IFIED

-0.60

 dispers

-0.59

IRA

-0.59

POSITIVE LOGITS

ournals

1.54

unction

1.39

ourney

1.34

ealous

1.31

oint

1.28

ordan

1.28

acket

1.26

ungle

1.24

eff

1.23

itsu

1.21

Activations Density 0.025%

names or proper nouns related to "junkie"

No Comments

No Known Activations