INDEX

Explanations

patterns related to numerical data or statistics

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

coordinates and commas

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

 considérons

-0.52

FormTagHelper

-0.42

الدراسه

-0.39

ніципалі

-0.36

ribune

-0.36

AntiForgeryToken

-0.34

gonic

-0.34

 adquis

-0.34

SYNOPSIS

-0.34

 AssemblyCompany

-0.34

POSITIVE LOGITS

хьтан

0.59

Hentet

0.57

 ſche

0.52

 intStringLen

0.48

 ModelRenderer

0.45

 reaſon

0.45

 faſt

0.44

Hochspringen

0.43

tonode

0.43

 cory

0.42

Activations Density 0.103%

patterns related to numerical data or statistics

coordinates and commas

No Comments

No Known Activations