INDEX

Explanations

patterns in whitespace or formatting variations

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

special characters and syntax

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

PerformLayout

-0.81

LabelTagHelper

-0.68

 surla

-0.68

contentLoaded

-0.66

rungsseite

-0.65

 AssemblyTitle

-0.63

nodoc

-0.62

 kasarigan

-0.62

IntoConstraints

-0.62

 defaultstate

-0.61

POSITIVE LOGITS

emplace

0.41

 calavera

0.38

 great

0.37

batsman

0.36

 kilometre

0.36

 germain

0.36

 favourable

0.35

entretien

0.34

ṁ

0.34

辐

0.33

Activations Density 0.034%

patterns in whitespace or formatting variations

special characters and syntax

No Comments

No Known Activations