INDEX

Explanations

the word "neutral" or variations of it

oai_token-act-pair · gpt-3.5-turbo

references to neutrality and neutral positions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 millenn

-0.76

 challeng

-0.70

Mill

-0.68

omething

-0.66

Hop

-0.66

teenth

-0.65

heres

-0.64

RET

-0.64

PER

-0.63

 toget

-0.63

POSITIVE LOGITS

izing

1.39

ization

1.27

izers

1.19

ized

1.19

ize

1.18

izes

1.16

izer

1.16

ity

1.14

izable

1.06

ising

1.05

Activations Density 0.019%

the word "neutral" or variations of it

references to neutrality and neutral positions

No Comments

No Known Activations