INDEX

Explanations

references to sexual misconduct and related crimes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

/current

-0.17

 Hack

-0.16

enser

-0.15

velope

-0.15

appa

-0.15

 HACK

-0.15

wl

-0.14

vron

-0.14

gether

-0.14

 Unsafe

-0.14

POSITIVE LOGITS

 charges

0.23

 cases

0.22

 ring

0.22

 attempt

0.21

 charge

0.20

 rings

0.20

 activity

0.19

 schemes

0.19

 attempts

0.18

 prevention

0.18

Activations Density 0.150%

references to sexual misconduct and related crimes

No Comments

No Known Activations