INDEX

Explanations

references to psychological topics and abuses of power

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.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

 treason

-0.18

icina

-0.15

ptrdiff

-0.15

onso

-0.15

Sel

-0.15

 tslib

-0.14

 Ð²Ð»Ð°Ñģ

-0.14

tright

-0.14

interop

-0.14

atÃ¼rk

-0.14

POSITIVE LOGITS

 abuse

0.27

 sexual

0.25

 priest

0.24

 Abuse

0.23

 molest

0.23

 mole

0.23

 SNAP

0.22

 settlements

0.22

 priests

0.22

 predator

0.21

Activations Density 0.010%

references to psychological topics and abuses of power

No Comments

No Known Activations