INDEX

Explanations

common phrases related to actions and interactions

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

part protected new improper increase interfere thread

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-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

новниш

-0.31

complexContent

-0.31

 comunión

-0.31

얻

-0.31

 gyhoeddwyd

-0.28

 <<<<<<<<<<<<<<

-0.28

 ninguno

-0.28

ándolos

-0.28

 éstos

-0.27

 duquel

-0.27

POSITIVE LOGITS

 مشين

0.69

\{\\

0.60

tagHelperRunner

0.58

ⓧ

0.56

ctools

0.54

centa

0.53

 OMITBAD

0.53

IndentedString

0.52

 يتيمه

0.52

ೢ

0.52

Activations Density 0.131%

common phrases related to actions and interactions

part protected new improper increase interfere thread

No Comments

No Known Activations