INDEX

Explanations

instances of the word "the" and other common words or phrases that indicate important information

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

code and styles

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

 cucchiai

-0.41

舺

-0.41

Ecotoxicity

-0.38

Aihe

-0.38

 protección

-0.38

 IFTT

-0.38

🔙

-0.37

 efectivamente

-0.36

endios

-0.36

]<

-0.36

POSITIVE LOGITS

تقاوى

0.65

RenderAtEndOf

0.61

setVerticalGroup

0.60

 Waray

0.56

новниш

0.52

DrawerToggle

0.50

providedIn

0.49

Cyfeiriadau

0.47

 виправивши

0.46

 كومونز

0.46

Activations Density 0.179%

instances of the word "the" and other common words or phrases that indicate important information

code and styles

No Comments

No Known Activations