INDEX

Explanations

mentions of intelligence-related terms and organizations

oai_token-act-pair · gpt-3.5-turbo

references to intelligence and intelligence agencies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

pain

-0.70

Benz

-0.69

TON

-0.64

 Daylight

-0.64

 Slayer

-0.62

zl

-0.62

 Candle

-0.62

steen

-0.61

uberty

-0.61

ergy

-0.61

POSITIVE LOGITS

 agencies

1.11

agency

1.06

 briefings

1.03

agents

0.99

 gathering

0.97

 agency

0.93

 operatives

0.88

 agents

0.87

 analysts

0.87

 Agency

0.87

Activations Density 0.050%

mentions of intelligence-related terms and organizations

references to intelligence and intelligence agencies

No Comments

No Known Activations