INDEX

Explanations

programming syntax and structure-related components

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

programming data types

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

CrossRef

-0.48

 Obrador

-0.36

getLogger

-0.35

VideoCapture

-0.34

zzleHttp

-0.33

bcryptjs

-0.32

xase

-0.32

듈

-0.32

ConverterFactory

-0.32

袤

-0.30

POSITIVE LOGITS

 propOrder

0.74

Personendaten

0.69

хьтан

0.64

 defaultstate

0.61

 незавершена

0.57

ſſung

0.57

 geweſen

0.56

<unused23>

0.55

<unused8>

0.54

<unused14>

0.54

Activations Density 0.005%

programming syntax and structure-related components

programming data types

No Comments

No Known Activations