INDEX

Explanations

expressions of acknowledgment or admission of fault

oai_token-act-pair · gpt-4o-mini

New Auto-Interp

Configuration

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

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Hzfinfdu/SlimPajama-3B

Features

32,768

Data Type

float32

Hook Name

blocks.15.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pers

-0.16

pad

-0.15

ì´

-0.14

 pied

-0.14

åĽ²

-0.14

ban

-0.14

Ãºp

-0.14

åĥı

-0.13

 Pied

-0.13

azy

-0.13

POSITIVE LOGITS

ably

0.18

 admission

0.16

 admitted

0.15

/disc

0.15

 defeat

0.15

Fully

0.15

çĮ«

0.14

airs

0.14

worth

0.14

iras

0.14

Activations Density 0.041%

expressions of acknowledgment or admission of fault

No Comments

No Known Activations