INDEX

Explanations

praiseworthy expressions related to accomplishment or performance

oai_token-act-pair · gpt-3.5-turbo

positive affirmations and expressions of encouragement

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

IFrame

Link

Negative Logits

minent

-0.77

abel

-0.75

ushima

-0.73

erous

-0.67

URA

-0.66

Ń·

-0.65

umat

-0.65

ierrez

-0.65

agen

-0.64

agin

-0.64

POSITIVE LOGITS

 luck

1.26

job

1.19

 thing

1.12

bye

1.11

 grief

1.02

 timing

0.98

luck

0.97

 lord

0.96

 idea

0.96

guy

0.92

Activations Density 0.093%

praiseworthy expressions related to accomplishment or performance

positive affirmations and expressions of encouragement

No Known Activations