INDEX

Explanations

references to law enforcement and police-related contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

guards

-0.16

 military

-0.15

ships

-0.15

spy

-0.15

 Defense

-0.15

åĨĽ

-0.14

 army

-0.14

adin

-0.14

Army

-0.14

yms

-0.14

POSITIVE LOGITS

 police

0.23

 unit

0.22

ACP

0.22

 polic

0.22

 officers

0.21

pol

0.20

 Crime

0.20

 detectives

0.20

 Police

0.20

 cops

0.20

Activations Density 0.193%

references to law enforcement and police-related contexts

No Comments

No Known Activations