INDEX

Explanations

unique identifiers or references within a specific context

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

CR followed by numbers

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

 Palatine

-0.41

lossom

-0.39

lossus

-0.38

 Quintet

-0.38

recent

-0.37

 boson

-0.36

frutar

-0.36

กุ

-0.36

 daarna

-0.36

 todav

-0.36

POSITIVE LOGITS

contentLoaded

0.64

RegressionTest

0.62

 <<<<<<<<<<<<<<

0.59

UserScript

0.57

 bezeichneter

0.57

ValueStyle

0.55

 OMITBAD

0.54

 propOrder

0.53

styleType

0.52

fjspx

0.52

Activations Density 0.092%

unique identifiers or references within a specific context

CR followed by numbers

No Comments

No Known Activations