INDEX

Explanations

mentions of attacks and conflict-related events

oai_token-act-pair · gpt-3.5-turbo

references to casualties and violent incidents in conflict areas

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

 Alloy

-0.69

 Darrell

-0.68

 Lifetime

-0.67

 Morty

-0.66

Â¯

-0.66

 Genius

-0.66

 Cary

-0.65

 Nerd

-0.64

STEM

-0.64

ãĤ¦ãĤ¹

-0.64

POSITIVE LOGITS

 Nato

1.04

 ceasefire

1.01

PKK

0.96

 Afghan

0.95

 Islamist

0.94

 Syrian

0.94

 armoured

0.94

Syria

0.94

 Kurdish

0.93

Syrian

0.93

Activations Density 1.945%

mentions of attacks and conflict-related events

references to casualties and violent incidents in conflict areas

No Comments

No Known Activations