INDEX

Explanations

mentions of government-related entities and their actions

oai_token-act-pair · gpt-3.5-turbo

references to the government

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

CONT

-0.77

 Rider

-0.73

ritch

-0.71

 Roses

-0.70

thy

-0.68

 Locations

-0.68

 Crew

-0.68

AUT

-0.67

CAST

-0.67

 Oval

-0.66

POSITIVE LOGITS

 officials

0.97

 decree

0.94

 shutdown

0.92

 administration

0.92

 agencies

0.90

 ministries

0.89

 bureaucracy

0.85

 government

0.85

 minister

0.84

 ministers

0.83

Activations Density 0.058%

mentions of government-related entities and their actions

references to the government

No Comments

No Known Activations