INDEX

Explanations

the word "loving" or words associated with love and affection

oai_token-act-pair · gpt-3.5-turbo

references to the word "love."

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

IFrame

Link

•layer 0 - love

No Comments

Negative Logits

 Cardinal

-0.70

 bailout

-0.69

 testament

-0.66

 standard

-0.65

opsy

-0.65

GC

-0.63

 Racial

-0.63

 Billboard

-0.63

 Penny

-0.63

 Crystal

-0.62

POSITIVE LOGITS

lov

5.20

liv

1.65

lav

1.39

hov

1.18

lain

1.17

akov

1.17

lik

1.13

Ð´

1.11

lr

0.98

ov

0.98

Activations Density 0.027%

the word "loving" or words associated with love and affection

references to the word "love."

No Comments

No Known Activations