INDEX

Explanations

phrases related to importance or focus

oai_token-act-pair · gpt-3.5-turbo

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

Not in Any Lists

No Comments

Negative Logits

UE

-0.81

 Tsukuyomi

-0.66

rosso

-0.66

FIX

-0.64

rahim

-0.62

outh

-0.62

ËĪ

-0.61

Flickr

-0.61

chnology

-0.61

@@

-0.61

POSITIVE LOGITS

ised

1.29

ity

1.18

izing

1.12

ization

1.10

isation

1.06

ising

1.05

izers

1.03

ized

1.00

 nervous

0.98

izes

0.97

Activations Density 0.023%

phrases related to importance or focus

No Comments

No Known Activations