INDEX

Explanations

statements and phrases of uncertainty, specifically focusing on calculations or interpretations that may involve errors

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

Not in Any Lists

No Comments

Negative Logits

_Tis

-0.08

Ã©rÃ©

-0.07

undy

-0.07

.sap

-0.07

etting

-0.06

.scalablytyped

-0.06

redi

-0.06

Â©Â©

-0.06

reeze

-0.06

Ã§as

-0.06

POSITIVE LOGITS

or

0.08

 éŁ

0.07

 either

0.07

 Warn

0.07

0.06

-g

0.06

 other

0.06

 Pall

0.06

Å

0.06

est

0.06

Activations Density 0.024%

statements and phrases of uncertainty, specifically focusing on calculations or interpretations that may involve errors

No Comments

No Known Activations