INDEX

Explanations

code-related elements and structures, especially in programming languages

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

code method definition

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

 للاسماء

-0.84

 kasarigan

-0.79

WriteBarrier

-0.72

 queſta

-0.72

writeFieldEnd

-0.71

 Biôgrafia

-0.69

 Taktlose

-0.69

expandindo

-0.69

twimg

-0.67

 Meksiku

-0.67

POSITIVE LOGITS

↵↵

0.46

 metod

0.36

 estima

0.34

 прове

0.32

 metody

0.30

 metodi

0.30

the

0.29

得了

0.29

 Método

0.29

mmung

0.28

Activations Density 0.002%

code-related elements and structures, especially in programming languages

code method definition

No Comments

No Known Activations