INDEX

Explanations

references to "late" in various contexts

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

lick

-0.16

 dá»¥ng

-0.15

 maxi

-0.15

rup

-0.14

Ø¨ÛĮÙĨ

-0.14

inz

-0.14

aar

-0.14

ths

-0.14

agedList

-0.14

nbsp

-0.14

POSITIVE LOGITS

-stage

0.29

-night

0.27

com

0.26

blo

0.20

 stages

0.20

ef

0.19

-breaking

0.19

Rooms

0.19

-game

0.18

 comer

0.18

Activations Density 0.022%

references to "late" in various contexts

No Comments

No Known Activations