INDEX

Explanations

issues related to convenience and commitment in decision-making contexts

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

linger

-0.15

TOTYPE

-0.14

auge

-0.14

oice

-0.14

neutral

-0.13

Calibri

-0.13

uala

-0.13

 neutral

-0.13

pitch

-0.12

 inheritance

-0.12

POSITIVE LOGITS

 convenient

0.18

 decide

0.18

 inconvenient

0.17

 decided

0.17

 arbitrarily

0.17

åĨ³å®ļ

0.17

 decides

0.17

 Convenient

0.17

isci

0.17

 inconvenience

0.16

Activations Density 0.078%

issues related to convenience and commitment in decision-making contexts

No Comments

No Known Activations