INDEX

Explanations

references to various brands and brand-related activities

oai_token-act-pair · gpt-3.5-turbo

mentions of various consumer brands

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Bay

-0.78

pmwiki

-0.76

 Corps

-0.75

veyard

-0.69

Lar

-0.69

 Maid

-0.67

 Rays

-0.67

 Poverty

-0.67

 Houth

-0.66

 Torrent

-0.66

POSITIVE LOGITS

 brands

1.35

isine

1.01

wcs

0.89

 brand

0.88

Brand

0.87

cham

0.86

awei

0.86

liga

0.86

 advertisers

0.84

 Brands

0.83

Activations Density 0.005%

references to various brands and brand-related activities

mentions of various consumer brands

No Comments

No Known Activations