INDEX

Explanations

phrases containing hedging language (e.g., to express doubt or uncertainty).

oai_token-act-pair · gemini-2.0-flash

New Auto-Interp

Configuration

fnlp/Llama-Scope-R1-Distill/400M-Slimpajama-400M-OpenR1-Math-220k/L11R

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Hzfinfdu/SlimPajama-3B and open-r1/OpenR1-Math-220k

Features

32,768

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 totiÅ¾

-0.07

onor

-0.07

iddles

-0.07

tuy

-0.06

enaire

-0.06

oris

-0.06

alles

-0.06

uraa

-0.06

ys

-0.06

/antlr

-0.06

POSITIVE LOGITS

ains

0.06

argins

0.06

Uph

0.06

 ello

0.06

ecies

0.06

 Bulk

0.06

Ð²Ð¸ÑĤ

0.06

bulk

0.06

AndWait

0.06

 ones

0.06

Activations Density 0.037%

phrases containing hedging language (e.g., to express doubt or uncertainty).

No Comments

No Known Activations