INDEX

Explanations

news-related words or terms

oai_token-act-pair · gpt-3.5-turbo

words and phrases related to news and media content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 cooks

-0.63

 arsen

-0.62

 whisk

-0.61

 warp

-0.61

 runway

-0.60

 redund

-0.60

 certainty

-0.60

 defenses

-0.59

 ballpark

-0.59

 commencement

-0.59

POSITIVE LOGITS

oft

0.86

research

0.80

DragonMagazine

0.79

sand

0.78

milo

0.78

proxy

0.76

coins

0.76

ãĥķãĤ©

0.75

tenance

0.75

tree

0.74

Activations Density 0.213%

news-related words or terms

words and phrases related to news and media content

No Comments

No Known Activations