INDEX

Explanations

specific health-related categories and their associated attributes or conditions

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

category names

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

 ब्रेकडाउन

-0.48

 linkovi

-0.44

delwed

-0.43

±

-0.42

舺

-0.40

bobox

-0.39

 مرئيه

-0.38

 rink

-0.38

حوالہ

-0.37

 continúas

-0.37

POSITIVE LOGITS

 område

0.66

forChild

0.59

 subcategory

0.52

 category

0.49

SpringBootTest

0.47

 doInBackground

0.47

IsMutable

0.46

 createSlice

0.46

områ

0.46

 såsom

0.46

Activations Density 0.147%

specific health-related categories and their associated attributes or conditions

category names

No Comments

No Known Activations