INDEX

Explanations

key concepts and roles related to various subjects or topics

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

purpose or subject of

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

Thomson

-0.58

liani

-0.57

يكب

-0.53

 Maximal

-0.52

Hochspringen

-0.52

 Tivoli

-0.50

Παραπομπές

-0.50

 Neuroscience

-0.49

 Muses

-0.49

aroo

-0.48

POSITIVE LOGITS

]='\

0.43

 Gegenteil

0.43

CppMethod

0.42

ConstraintMaker

0.40

SharedCtor

0.39

ValueStyle

0.38

 հղումներ

0.35

astify

0.34

 Vordergrund

0.34

 handleMessage

0.34

Activations Density 0.037%

key concepts and roles related to various subjects or topics

purpose or subject of

No Comments

No Known Activations