INDEX

Explanations

proper nouns and locations, particularly those related to sports or media

oai_token-act-pair · gpt-3.5-turbo

references to images and photography

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 antit

-0.65

 comple

-0.63

 blinded

-0.62

 Arbit

-0.59

arde

-0.59

tot

-0.59

Zar

-0.59

 assurance

-0.58

 footing

-0.58

 propri

-0.58

POSITIVE LOGITS

ÂłÂłÂłÂł

0.85

BBC

0.85

ÂłÂłÂł

0.83

ccording

0.82

Politics

0.81

WASHINGTON

0.80

 SHARES

0.79

GREEN

0.78

ÂłÂłÂłÂłÂłÂłÂłÂł

0.78

Python

0.77

Activations Density 0.152%

proper nouns and locations, particularly those related to sports or media

references to images and photography

No Comments

No Known Activations