INDEX

Explanations

statements or assertions highlighted by the word "This."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

This introducing a statement or result

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Houſe

-0.82

 houſe

-0.77

 pleaſure

-0.71

 ſta

-0.65

 Reſ

-0.65

SequentialGroup

-0.62

ſelf

-0.60

 ſtre

-0.57

 ſever

-0.57

 Majefty

-0.57

POSITIVE LOGITS

 conclusão

0.45

 null

0.39

 comptes

0.38

tagHelperRunner

0.37

 finally

0.37

 configuração

0.37

 idéia

0.35

 autorytatywna

0.35

conteudo

0.34

 basically

0.33

Activations Density 0.110%

statements or assertions highlighted by the word "This."

This introducing a statement or result

No Comments

No Known Activations