INDEX

Explanations

words related to enabling or enjoying activities

oai_token-act-pair · gpt-3.5-turbo

instances of the substring "en"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ĳå£«

-0.68

 distortion

-0.63

TPPStreamerBot

-0.62

 Pyth

-0.57

 Heights

-0.57

 autism

-0.56

 distortions

-0.56

dale

-0.55

 flats

-0.55

 killer

-0.55

POSITIVE LOGITS

emies

1.40

abling

1.33

vironments

1.29

chant

1.26

riched

1.22

viron

1.21

velop

1.21

forcing

1.20

rollment

1.15

joy

1.14

Activations Density 0.028%

words related to enabling or enjoying activities

instances of the substring "en"

No Comments

No Known Activations