INDEX

Explanations

the word "Carleton" with a very high activation level

oai_token-act-pair · gpt-3.5-turbo

references to the name "Carleton."

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

Su

-0.75

 word

-0.66

é¾įå¥ĳå£«

-0.59

 ratings

-0.58

 controlled

-0.57

 compensated

-0.57

 worldwide

-0.57

 term

-0.57

 warrants

-0.57

fre

-0.56

POSITIVE LOGITS

leton

4.55

letal

1.76

erton

1.21

legate

1.19

lington

1.16

alore

1.08

let

1.07

negie

1.07

erella

1.03

illac

1.02

Activations Density 0.020%

the word "Carleton" with a very high activation level

references to the name "Carleton."

No Comments

No Known Activations