INDEX

Explanations

numeric values and their specific formatting or contexts

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

structured data and markup symbols

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

########.

-0.74

 Paglinawan

-0.70

 referrerpolicy

-0.68

sizeCache

-0.68

DockStyle

-0.65

stdc

-0.59

GraphicsUnit

-0.58

quested

-0.56

Tembelea

-0.53

mpagne

-0.53

POSITIVE LOGITS

 Italijani

0.39

 actualidad

0.34

fraz

0.31

 éclat

0.28

 renov

0.28

 пи

0.27

 trị

0.27

 تک

0.27

pol

0.27

NavItem

0.27

Activations Density 0.002%

numeric values and their specific formatting or contexts

structured data and markup symbols

No Comments

No Known Activations