INDEX

Explanations

references to software packages

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

package followed by identifier

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

iner

-0.37

bindingNavigator

-0.35

 canst

-0.33

 Ceramby

-0.33

to

-0.30

IInter

-0.30

Acu

-0.29

like

-0.29

ion

-0.29

 into

-0.28

POSITIVE LOGITS

package

1.23

 AssemblyTitle

0.85

 propOrder

0.82

TagMode

0.81

WriteTagHelper

0.72

 initComponents

0.71

subpackage

0.71

BeginInit

0.71

ValueStyle

0.71

 PACKAGE

0.71

Activations Density 0.001%

references to software packages

package followed by identifier

No Comments

No Known Activations