INDEX

Explanations

programming-related syntax elements

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

code variable names and 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

ARROLL

-0.43

 régulier

-0.36

 Inscrivez

-0.36

好

-0.35

私も

-0.33

 estaremos

-0.32

 podrán

-0.32

 pleine

-0.31

epresidente

-0.31

 trouvez

-0.31

POSITIVE LOGITS

Personensuche

0.76

 مشين

0.69

esModule

0.65

رشف

0.65

<unused8>

0.65

<unused41>

0.65

<unused51>

0.65

<unused23>

0.65

<pad>

0.65

<unused14>

0.65

Activations Density 0.013%

programming-related syntax elements

code variable names and types

No Comments

No Known Activations