INDEX

Explanations

keywords related to items that are discarded, unwanted, or of low value

oai_token-act-pair · gpt-3.5-turbo

references to "junk" in various contexts

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

Not in Any Lists

No Comments

Negative Logits

oward

-0.79

tymology

-0.75

 Contrast

-0.73

ÅŁ

-0.71

Âº

-0.70

 Laws

-0.69

ĸļ

-0.68

ales

-0.66

 Reign

-0.63

 Mata

-0.63

POSITIVE LOGITS

 junk

4.00

 Junk

2.22

 crap

1.44

 rubbish

1.34

 garbage

1.34

 clutter

1.30

 trash

1.28

 worthless

1.15

 bullshit

1.12

 scrap

1.09

Activations Density 0.017%

keywords related to items that are discarded, unwanted, or of low value

references to "junk" in various contexts

No Comments

No Known Activations