INDEX

Explanations

various forms of the word "safety" and related terms that emphasize safety concerns and regulations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 23-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L23R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.23.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

 safe

-0.21

safe

-0.20

 safely

-0.18

 safer

-0.17

Safe

-0.16

 Safe

-0.16

 safest

-0.15

-safe

-0.15

à¹Ĩ

-0.14

imon

-0.14

POSITIVE LOGITS

/security

0.23

-net

0.22

 measures

0.22

net

0.20

-conscious

0.20

Net

0.19

 measure

0.18

-FIRST

0.17

 margins

0.17

NET

0.17

Activations Density 0.016%

various forms of the word "safety" and related terms that emphasize safety concerns and regulations

No Comments

No Known Activations