INDEX

Explanations

code structure elements such as method definitions and print statements

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

printing and display

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

WriteTagHelper

-1.09

 autorytatywna

-0.94

 ब्रेकडाउन

-0.83

tvguidetime

-0.81

principalColumn

-0.80

 الحره

-0.80

 незавершена

-0.79

الحياه

-0.79

 otomatig

-0.79

 ModelExpression

-0.79

POSITIVE LOGITS

display

0.31

 display

0.30

InView

0.29

 Beobachtung

0.29

 displays

0.28

 printing

0.28

data

0.27

 Appreciation

0.26

Printer

0.26

print

0.26

Activations Density 0.026%

code structure elements such as method definitions and print statements

printing and display

No Comments

No Known Activations