INDEX

Explanations

themes related to safety and protection

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

stub

-0.14

crud

-0.14

Ð°Ð½ÑĤÐ¸

-0.13

ÙĤÙħ

-0.13

alen

-0.13

CCI

-0.13

\uc

-0.13

icode

-0.12

Nay

-0.12

ä»Ļ

-0.12

POSITIVE LOGITS

 safe

0.91

 safety

0.81

safe

0.79

 Safe

0.77

-safe

0.74

 safer

0.71

Safe

0.69

 Safety

0.68

 safely

0.68

 SAFE

0.68

Activations Density 0.163%

themes related to safety and protection

No Comments

No Known Activations