INDEX

Explanations

phrases and keywords related to political news and speeches, especially involving different countries and political figures

oai_token-act-pair · gpt-3.5-turbo

political events and statements related to elections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 metre

-0.72

 realised

-0.71

 analyse

-0.70

 endeavour

-0.67

 explorers

-0.66

 labou

-0.65

gements

-0.65

 realise

-0.64

CLR

-0.61

 Whilst

-0.61

POSITIVE LOGITS

 Republican

1.11

 bipartisan

1.07

Republicans

1.02

Republican

1.02

 Politico

1.02

GOP

1.01

partisan

1.00

 Hillary

0.99

Democrats

0.99

 Republicans

0.98

Activations Density 4.182%

phrases and keywords related to political news and speeches, especially involving different countries and political figures

political events and statements related to elections

No Comments

No Known Activations