INDEX

Explanations

political terms and information related to party performance

oai_token-act-pair · gpt-3.5-turbo

phrases related to political polling and party support

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

DOE

-0.93

natureconservancy

-0.90

 honoring

-0.88

assetsadobe

-0.85

urized

-0.82

 honored

-0.80

 Smithsonian

-0.78

 endeavors

-0.78

 NCAA

-0.75

 subpoena

-0.73

POSITIVE LOGITS

 UKIP

1.61

 Ukip

1.59

 Labour

1.41

 favour

1.34

Ips

1.33

SNP

1.29

 Tories

1.28

 Britons

1.28

Labour

1.27

 Remain

1.25

Activations Density 0.382%

political terms and information related to party performance

phrases related to political polling and party support

No Comments

No Known Activations