INDEX

Explanations

contextual references and conditions related to arrangements or classifications

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

"Whether" and related conditional words

np_acts-logits-general · gemini-2.0-flash

whether, if, either

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

AndEndTag

-0.64

 noDo

-0.52

labus

-0.50

__":

-0.50

AddHtmlAttribute

-0.48

 للمعارف

-0.46

httphttps

-0.46

fxml

-0.45

henswürdigkeiten

-0.43

stasia

-0.43

POSITIVE LOGITS

 ultimately

0.45

 tetap

0.40

 siguran

0.39

tidaknya

0.39

Ultimately

0.38

总之

0.38

 ultimate

0.38

ilarang

0.37

 kuitenkin

0.36

 comunque

0.35

Activations Density 0.431%

contextual references and conditions related to arrangements or classifications

"Whether" and related conditional words

whether, if, either

No Comments

No Known Activations