INDEX

Explanations

references to specific rooms and structural features in a house or property

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

IFrame

Link

Negative Logits

 foyer

-0.16

ãĥ³ãĥĸ

-0.16

zÃŃ

-0.15

åħ¬åľĴ

-0.15

 bracket

-0.15

.cp

-0.14

æľµ

-0.14

OUCH

-0.14

levard

-0.14

è¡Ĺéģĵ

-0.14

POSITIVE LOGITS

 basement

0.51

 Basement

0.43

bas

0.42

 underground

0.39

 cellar

0.37

åľ°ä¸ĭ

0.35

Bas

0.34

Bas

0.33

bas

0.32

 Underground

0.31

Activations Density 0.116%

references to specific rooms and structural features in a house or property

No Known Activations