INDEX

Explanations

punctuation marks, specifically commas and periods

oai_token-act-pair · gpt-4o-mini

New Auto-Interp

Configuration

fnlp/Llama-Scope-R1-Distill/0-Slimpajama-800M-OpenR1-Math-220k/L15R

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

open-r1/OpenR1-Math-220k

Features

32,768

Data Type

float32

Hook Name

blocks.15.hook_resid_post

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Embeds

IFrame

Link

Negative Logits

.IsActive

-0.08

Î²Î¿Î»

-0.06

åĿĽ

-0.06

.TabStop

-0.06

Pid

-0.06

 Nicol

-0.06

¹

-0.06

 estr

-0.06

 IDEA

-0.06

\u

-0.06

POSITIVE LOGITS

$MESS

0.07

ascar

0.07

 treff

0.06

zell

0.06

ivy

0.06

agram

0.06

ideographic

0.06

AAF

0.06

.scalablytyped

0.06

stroy

0.06

Activations Density 0.179%

punctuation marks, specifically commas and periods

No Known Activations