INDEX

Explanations

proper nouns, specifically names

oai_token-act-pair · gpt-3.5-turbo

instances of the letter "J" in various contexts

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

 galleries

-0.68

ãĤ¤ãĥĪ

-0.66

 pony

-0.65

 Starfleet

-0.63

 MacArthur

-0.63

 beware

-0.62

holders

-0.62

ciation

-0.62

houses

-0.60

 dictators

-0.59

POSITIVE LOGITS

upiter

1.38

ournals

1.29

ealous

1.26

unction

1.23

acket

1.22

umbo

1.21

igsaw

1.21

ourney

1.21

umping

1.20

AMES

1.19

Activations Density 0.030%

proper nouns, specifically names

instances of the letter "J" in various contexts

No Comments

No Known Activations