INDEX

Explanations

instances of the word "wrong" and related expressions indicating mistakes or moral failings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 31-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L31R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.31.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

.qual

-0.16

Ø§ÙĨÙĪ

-0.15

anki

-0.15

lesi

-0.15

arters

-0.15

lsa

-0.15

illet

-0.15

/cli

-0.15

apa

-0.14

cla

-0.14

POSITIVE LOGITS

headed

0.43

fully

0.40

-headed

0.37

/right

0.30

er

0.30

ed

0.30

 wrong

0.29

wrong

0.28

eous

0.27

est

0.27

Activations Density 0.066%

instances of the word "wrong" and related expressions indicating mistakes or moral failings

No Comments

No Known Activations