INDEX

Explanations

words and phrases related to comparison and compatibility

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

action completion or statement

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

DMETHOD

-0.41

modelBuilder

-0.40

 deleteById

-0.39

 plufieurs

-0.39

":[{

-0.38

bootstrapcdn

-0.38

 dataTable

-0.38

 betweenstory

-0.38

 onCreate

-0.37

 zwierzę

-0.37

POSITIVE LOGITS



0.52

 autorytatywna

0.46

 Damn

0.41

يكب

0.41

AddTagHelper

0.41

LookAnd

0.41

************/

0.40

 WRONG

0.40

Damn

0.40

 damn

0.39

Activations Density 0.039%

words and phrases related to comparison and compatibility

action completion or statement

No Comments

No Known Activations