INDEX

Explanations

mentions of laws, government agencies, and security-related terms

oai_token-act-pair · gpt-3.5-turbo

references to national security and intelligence-related topics

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

MLA

-0.82

 organise

-0.76

eva

-0.76

arre

-0.75

 Serie

-0.71

 Barcelona

-0.70

 Rowling

-0.70

Rober

-0.69

 analyse

-0.69

 Leicester

-0.68

POSITIVE LOGITS

 counterterrorism

1.02

DOD

1.01

 Pentagon

0.98

 Eisenhower

0.92

 neocons

0.90

 Reagan

0.90

 Mattis

0.89

 Cheney

0.88

 intelligence

0.83

 Clapper

0.82

Activations Density 0.318%

mentions of laws, government agencies, and security-related terms

references to national security and intelligence-related topics

No Comments

No Known Activations