INDEX

Explanations

words related to specific levels or areas within buildings

oai_token-act-pair · gpt-3.5-turbo

references to different building floors and levels

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Canaver

-0.88

wana

-0.81

anson

-0.79

cific

-0.79

ubric

-0.78

REDACTED

-0.76

igel

-0.75

=-=-=-=-=-=-=-=-

-0.73

ciating

-0.71

 Nieto

-0.70

POSITIVE LOGITS

 balcony

1.33

 apartment

1.33

 apartments

1.22

 residential

1.07

 buildings

1.06

 balcon

1.02

 rooftop

1.02

 skysc

0.96

 tower

0.96

 basement

0.95

Activations Density 0.129%

words related to specific levels or areas within buildings

references to different building floors and levels

No Comments

No Known Activations