INDEX

Explanations

news about violent incidents, attacks, and arrests

oai_token-act-pair · gpt-3.5-turbo

references to violent events and their aftermath

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Cary

-0.73

 darn

-0.68

 Barney

-0.67

 Briggs

-0.66

Madison

-0.63

 dinosaurs

-0.63

DeL

-0.63

 Whitney

-0.62

Kansas

-0.62

 Madison

-0.62

POSITIVE LOGITS

 ceasefire

1.11

zbollah

1.04

 Hezbollah

0.96

 Nato

0.96

PKK

0.93

 UNHCR

0.90

 Daesh

0.89

 militants

0.89

 Kurdish

0.88

 separatist

0.87

Activations Density 1.828%

news about violent incidents, attacks, and arrests

references to violent events and their aftermath

No Comments

No Known Activations